API Design on the Scale of

Decades
Learn How to Architect and Design
Long-lasting APIs

Nordic APIs
© 2016 - 2022 Nordic APIs
Tweet This Book!
Please help Nordic APIs by spreading the word about
this book on Twitter!
The suggested tweet for this book is:
Check out ”API Design on the Scale of Decades” - the
latest eBook by the @NordicAPIs team
The suggested hashtag for this book is #APIdesign.
Find out what other people are saying about the book by
clicking on this link to search for this hashtag on Twitter:
#APIdesign
Also By Nordic APIs
Developing the API Mindset
The API Lifecycle
Securing The API Stronghold
API-Driven DevOps
The API Economy
Programming APIs with the Spark Web Framework
How to Successfully Market an API
GraphQL or Bust
API Strategy for Open Banking
Identity And APIs
API as a Product
Developer Experience
This book is dedicated to the speakers, attendees, and
sponsors that continually make Nordic APIs events
wonderful!
Contents

Supported by Curity . . . . . . . . . . . . . . . . . . i

Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii
APIs on the Scale of Decades . . . . . . . . . . . . . iii

Designing a True REST State Machine . . . . . . . . 1

The History Behind REST and Hypermedia . . . . 2
How do we Define REST? . . . . . . . . . . . . . . . 3
Misconception #3: REST APIs Should be Versioned 6
Misconception #4: Hypermedia is Optional for
REST APIs . . . . . . . . . . . . . . . . . . . . . 7
Example State Machine: IoT Toaster . . . . . . . . 8
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . 13

Is GraphQL The End of REST Style APIs? . . . . . . . 14

Defining REST and its Limitations . . . . . . . . . . 15
Round Trip and Repeat Trip Times . . . . . . . . . 16
Over/Under Fetching . . . . . . . . . . . . . . . . . . 16
Weak Typing and Poor Metadata . . . . . . . . . . 17
Improper Architecture Usage . . . . . . . . . . . . 18
REST Has Many Roundtrips - GraphQL Has Few . 19
REST Has Poor Type Systems - GraphQL Has a
Sophisticated One . . . . . . . . . . . . . . . . 19
REST Has Poor Discoverability - GraphQL Has
Native Support . . . . . . . . . . . . . . . . . . 20
REST Is Thin Client/Fat Server - GraphQL is Fat
Client/Fat Server . . . . . . . . . . . . . . . . . 21
CONTENTS

The End Of The Status Quo . . . . . . . . . . . . . . 21

Conclusion . . . . . . . . . . . . . . . . . . . . . . . . 22

Continuous Versioning Strategy for Internal APIs 23

Typical Public API Versioning . . . . . . . . . . . . . 24
Badoo’s Continuous Versioning Strategies . . . . 25
Changing the Verification Process . . . . . . . . . . 26
Updating Banner CTAs for Specific Clients . . . . 27
Use Flags to Avoid Versioning in Complex Busi-
ness Logic Changes . . . . . . . . . . . . . . . 28
Run Experimental Features . . . . . . . . . . . . . . 28
How Continuous Versioning Could Apply to You . 29

Case Study: Spotify Internal Payment APIs . . . . 30

The Evolution of Spotify Payments . . . . . . . . . 31
What’s Really Going on with Online Payments? . 32
The Checkout API . . . . . . . . . . . . . . . . . . . . 33
The Billing API . . . . . . . . . . . . . . . . . . . . . . 35
Sample Use Case: Automatic Alerts . . . . . . . . . 36
4 Reasons Why API Design is Critical to Subscrip-
tion Services . . . . . . . . . . . . . . . . . . . 37
Analysis: Scalable API-Driven Infrastructure will
Power the Future of Online Payments . . . 39

The Benefits of a Serverless API Backend . . . . . . 41

What Does “Serverless” Mean? . . . . . . . . . . . . 42
From Traditional to Serverless Environments . . 43
How Get Started: Understanding the Serverless
Vendors . . . . . . . . . . . . . . . . . . . . . . 45
Designing Event-Driven Serverless Applications . 46
5 Serverless Pro Tips: . . . . . . . . . . . . . . . . . 47
Example: Kickflip SDK . . . . . . . . . . . . . . . . . 48
Building Serverless API Backends . . . . . . . . . . 49

Putting an End to API Polling . . . . . . . . . . . . . . 51

What is the Polling Madness? . . . . . . . . . . . . 51
CONTENTS

One Solution: REST Hooks . . . . . . . . . . . . . . 52

Counter Arguments . . . . . . . . . . . . . . . . . . 53
Implementing REST Hooks . . . . . . . . . . . . . . 54
Alternatives . . . . . . . . . . . . . . . . . . . . . . . . 56
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . 57

6 Ways to Become a Master Microservice Gardener 58

6 Ways to Become a Master Microservice Gar-
dener . . . . . . . . . . . . . . . . . . . . . . . . 59
1: Use Bimodal IT to Avoid Stagnation . . . . . . . 60
2: Avoid the Kafka-esque Monolith . . . . . . . . . 61
3: Design In a Way That Promotes Further Iteration 62
4: Harvest Concepts and Incorporate . . . . . . . . 63
5: Distribute the Seeds: Adopt True Microser-
vices Arrangement . . . . . . . . . . . . . . . 64
6: Prune the Service Surface . . . . . . . . . . . . . 65
Nurture API Ecosystem Growth . . . . . . . . . . . 65

Is Your API Automotive Grade? . . . . . . . . . . . . 67

Test It Like it Has to be Automotive Grade . . . . 68
Automotive Grade and Futureproofing . . . . . . 70
Automotive Grade and Backwards Compatibility 71
Final Thoughts . . . . . . . . . . . . . . . . . . . . . . 72

Why OAuth 2.0 Is Vital to IoT Security . . . . . . . . 74

What is OAuth 2.0? . . . . . . . . . . . . . . . . . . . 75
What Does OAuth Do? . . . . . . . . . . . . . . . . . 76
Unique IoT Traits that Affect Security . . . . . . . . 77
Proof of Possession . . . . . . . . . . . . . . . . . . 78
Disconnected Flow . . . . . . . . . . . . . . . . . . . 80
Real Worth Authorization Failure . . . . . . . . . . 81
OAuth Embeds Trust into the IoT . . . . . . . . . . 82
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . 83

4 Design Tweaks to Improve API Operations . . . 84

1: HTTP GET instead of POST . . . . . . . . . . . . . 85
CONTENTS

2: Letting clients constantly poll APIs . . . . . . . . 86

3: Rigid hierarchy in microservices causes latency 87
4: Generic actions . . . . . . . . . . . . . . . . . . . . 88
Harmon-ious Mantras to Live By . . . . . . . . . . . 89

The API of Me . . . . . . . . . . . . . . . . . . . . . . . . 91
Confidential . . . . . . . . . . . . . . . . . . . . . . . 92
Financial . . . . . . . . . . . . . . . . . . . . . . . . . . 93
Tactile . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
Aggregate . . . . . . . . . . . . . . . . . . . . . . . . . 96
Final Thoughts . . . . . . . . . . . . . . . . . . . . . . 97

Avoid Walking on Eggshells and Use DevOps . . . 98

Silos Are Bad For Business . . . . . . . . . . . . . . 99
Key DevOps Concepts . . . . . . . . . . . . . . . . . 100
DevOps and APIs . . . . . . . . . . . . . . . . . . . . 103
Guides to API Development Management . . . . . 104
Conclusion: Putting Humpty Back Together Again 106

Securing the IoT for Decades to Come . . . . . . . . 107

Looking into the Crystal Ball: The World in 2030 . 109
Identity is the #1 Impediment to Safe IoT Con-
nections . . . . . . . . . . . . . . . . . . . . . . 110
Building on Open Standards will Secure Future
Identity in the IoT . . . . . . . . . . . . . . . . 111
The Nuances of IoT-Based Communication . . . . 113
5 Actionable Steps Toward Improving IoT Security114

5 Ways the OpenAPI Specification Spurs API Agility116

What is the OpenAPI Specification? . . . . . . . . . 117
Why OpenAPI Specification? . . . . . . . . . . . . . 118
API Fastness . . . . . . . . . . . . . . . . . . . . . . . 119
1: Proper Design and Approach . . . . . . . . . . . 119
2: Complete Documentation and Description . . 120
3: Rapid Testing and Iteration . . . . . . . . . . . . 121
4: Shorter (More Secure) Time to Market . . . . . 122
CONTENTS

5: Machine and Human Readability and Transla-

tion . . . . . . . . . . . . . . . . . . . . . . . . . 123
Conclusion: OpenAPI Enables an Agile API Lifecycle124

Case Study: Bosch Ongoing Enterprise API Man-

agement Saga . . . . . . . . . . . . . . . . . . . . . 125
Is building everything yourself always the answer?126
Have an effective strategy . . . . . . . . . . . . . . . 127
Be ready for that strategy to fall apart . . . . . . . 128
Think big, even when you’re small . . . . . . . . . 130

In Summary . . . . . . . . . . . . . . . . . . . . . . . . . 132
TL/DR - 15 Important Takeaways . . . . . . . . . . 132
What’s Next? . . . . . . . . . . . . . . . . . . . . . . . 134

Nordic APIs Resources . . . . . . . . . . . . . . . . . . 136

Endnotes . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
CONTENTS i

Supported by Curity

Nordic APIs was founded by Curity CEO Travis Spencer

and has continued to be supported by the company.
Curity helps Nordic APIs organize two strategic annual
events, the Austin API Summit in Texas and the Platform
Summit in Stockholm.
Curity is a leading provider of API-driven identity man-
agement that simplifies complexity and secures digital
services for large global enterprises. The Curity Identity
Server is highly scalable, and handles the complexities of
the leading identity standards, making them easier to use,
customize, and deploy.
Through proven experience, IAM and API expertise, Curity
builds innovative solutions that provide secure authenti-
cation across multiple digital services. Curity is trusted by
large organizations in many highly regulated industries,
including financial services, healthcare, telecom, retail,
gaming, energy, and government services across many
countries.
Check out Curity’s library of learning resources on a vari-
ety of topics, like API Security, OAuth, and Financial-grade
APIs.
Follow us on Twitter and LinkedIn, and find out more on
CONTENTS ii

curity.io.
Preface
APIs on the Scale of Decades

Roy Fielding, the creator of the REST standard for API

design, once described REST as:

“software design on the scale of decades: every de-

tail is intended to promote software longevity and
independent evolution. Many of the constraints
are directly opposed to short-term efficiency.”

Essentially, he acknowledged how software developers

often execute short term design without long-term de-
sign in mind, which causes inadaptability as technology
evolves.
In October 2016 Nordic APIs held it’s annual Platform
Summit centered on this theme, bringing together API
industry thought leaders to share their insights on what it
means to architect and design Application Programming
Interfaces on the scale of decades. The event detailed
RESTful design techniques for longevity, operational com-
ponents of sustaining an API, and topics like DevOps,
microservices architecture, developer relations, and new
business methodologies for supporting an API platform.
To create API Design on the Scale of Decades, we gathered
the top 15 most watched sessions from the Summit that
were relevant to API Design, and drafted companion
Preface iv

chapters to dive into each topic. This unique volume

thus contains a holistic assortment of insights that our
followers have found to be the most current, mission-
critical ideas for sustaining an API platform.
For API designers and architects, this eBook release presents
a convenient way to tap into a wide range of knowledge
that we’ve been collating on the blog over the last few
months. Guided by some of the most significant Nordic
APIs presentations, it outlines some of the most helpful
advice we’ve published to date.
As we close the chapter on the last Platform Summit we
begin to plan for the future. Stay tuned for updates from
Nordic APIs on our 2017 event, which will be on the theme
of API Scalability. On that note, if you would like to join
the legion of past API speakers, consider submitting a
session here!
So, please enjoy API Design on the Scale of Decades, and let
us know how we can improve. If you haven’t yet, consider
following us, and signing up to our newsletter for curated
blog updates and future event announcements.
Thank you for reading!

– Bill Doerrfeld, Editor in Chief, Nordic APIs

Connect with Nordic APIs:

Facebook | Twitter | Linkedin | Google+ | YouTube
Blog | Home | Newsletter | Contact
Designing a True REST
State Machine

There are a lot of misconceptions surrounding what ex-

actly Representational State Transfer (REST) is. The prime
of which is the concept of hypermedia, or defined in full
context, Hypermedia as the engine of application state
(HATEOAS).
Jargon aside, hypermedia is actually a valuable idea that
many self-touted “RESTful” web APIs do not adhere to. Hy-
permedia places heightened operational significance on
resources, located at URIs. With a hypermedia API, when
a request is sent to a URI, the response lists information
on resource state and acceptable future operations, in
essence creating a state machine that can be manipu-
lated.
Designing a True REST State Machine 2

Hypermedia is truly valuable as it can create more power-

ful APIs that alleviate the need for API versioning. But this
is only one of many facets that makes true REST design
awesome.
In this chapter, we’ll define what exactly REST is and
isn’t. Led by RESTafarian Asbjørn Ulsberg, we’ll uncover
the history of information design that has led to REST,
and debunk some common misinterpretations of REST
design.
We’ll also construct a mock state machine and example
HTTP behaviors, demonstrating how hypermedia could
be used within a REST API to trigger states on an IoT
kitchen device. Actually following the REST constraints by
building a HATEOAS-compliant API could be very bene-
ficial to advancing the web, so let’s give it the focus it
deserves.

Speaker: Asbjørn Ulsberg

This chapter was inspired by Ulsberg’s session
at the 2016 Platform Summit. Watch the full
talk here

The History Behind REST and

Hypermedia

Can you guess how old the concept of Hypermedia is?

It’s nearly 80 years old. That’s right, in 1941, Agentinian
author Jorge Luis Borges wrote The Garden of Forking
Paths, a manuscript that contained pages that referenced
each other; arguably the first form of hypertext. From
Bioshock to Goosebumps, choose your adventure style
Designing a True REST State Machine 3

entertainment and relational data have become com-

monplace, but in his time, the book was unprecedented.
Some other important strides that have led to REST are:

• 1963: Ted Nelson coins the terms hypertext and

hypermedia.
• 1968: Douglas Englebart debuts the On-Line System
in the Mother of All Demos, which was the first appli-
cation of hypertext, including mouse pointing, text
editing, window environment and more, essentially
giving birth to modern word processing.
• 1987: Apple employee Bill Atkinson creates Hyper-
Card, creating the first successful implementation of
hypermedia before the World Wide Web.
• 1989: Tim Berners-Lee creates the World Wide Web
at CERN, implementing the first successful Hypertext
Transfer Protocol (HTTP) implementation between a
client and server.
• 2000: Roy Fielding, Co-author of the HTTP and URI
specification, writes a doctoral dissertation entitled
Architectural Styles and the Design of Network-based
Software Architectures. In Chapter 5 he describes
Representational State Transfer, or as we commonly
call it, REST.

How do we Define REST?

So what is REST? Well, it’s difficult to tackle all the techni-

calities in a single blog post, but let’s first respond to four
misconceptions to understand what REST isn’t.
Designing a True REST State Machine 4

Common Misconception #1: REST is just

CRUD.

CRUD, or Create, Read, Update, and Delete, has become

a hallmark acronym amongst data management profes-
sionals as it represents the four basic actions for commu-
nicating with a database. Though CRUD maps cleanly with
SQL actions, as we can see clearly below, it doesn’t map
well to HTTP methods:
Operation SQL HTTP
Create INSERT PUT / POST
Read SELECT GET
Update UPDATE PUT / POST / PATCH
Delete DELETE DELETE

Though GET and DELETE coordinate well, POST, PUT, and PATCH
aren’t directly synonymous with CRUD operations. For
example, POST doesn’t necessarily only mean “Create”.
It’s actually a very versatile method — so versatile that
the entire SOAP protocol is tunneled through the POST
method when used with HTTP.
Since HTTP methods don’t map cleanly to CRUD, Uls-
berg argues that API providers should consider how they
might describe their APIs in a different way:

“Don’t limit yourself to CRUD when you design a

REST API. You should read the specification and
understand the semantics of each method, and
use it properly.”

What it comes down to is that REST is an architectural

style, not a protocol. So, calling an HTTP API behaving with
CRUD operations “RESTful” is a fallacy.
Designing a True REST State Machine 5

Common Misconception #2: Some URI

Constructions are more RESTful than
Others

Uniform Resource Identifiers (URIs) are vital for defining

resources and acting on resources, and are a core con-
cept of REST. However, many developers seem to think
they can distinguish a REST API simply based on how the
URI is structured. Can you tell which URI is more RESTful
than the others?

http://api.nordicapis.com/authors/contributor?author=doerrfeld
http://api.nordicapis.com/blogpost/getPostById?id=47
http://api.nordicapis.com/blogpost/47/edit-form
http://api.nordicapis.com/blogpost/47
http://api.nordicapis.com/128ndoels-8asdf-12d5-39d3

Contrary to popular belief, within the REST guidelines

there is no such thing as a RESTful URI construction. To
REST, these URIs are simply opaque identifiers — yes,
they are global, unique identifiers that can be used for
many purposes. However, without knowing more context
and the behavior going on under the hood (what the
methods look like, what the request looks like, what the
response is, etc.) there is no way to tell whether these
are RESTful operations or not. What the URI is actually
shouldn’t matter, so any of the ones above are as good
as the other one.

“Since only machines should read the URIs, and

no human, it shouldn’t matter”

The human-readable API design crowd is probably throw-

ing tomatoes by this point. All this isn’t to say that you
shouldn’t give attention to making URIs human readable.
Designing a True REST State Machine 6

However, Ulsberg recognizes that you shouldn’t depend

on them being in any particular way within the client.
For example, say we have very basic documentation for a
Nordic APIs Blog Post API, as identified below:

URI Method Description

http://api.nordicapis.com
/v1/blogposts POST Creates a new
blog post
/v1/blogposts/{id} GET Retrieves a blog
post
/v1/blogposts/{id} PUT Updates a blog
post
/v1/blogposts/{id} DELETE Deletes blog
post
/v1/blogposts/{id}/author
GET Retrieves blog
post author
/v1/blogposts/{id}/comments
GET Retrieves blog
post comments

Just because we’ve listed a URI, Method, and Description

for each one of our API calls doesn’t mean that we’ve
created a REST API — we’ve just documented our URIs as
we would document RPC operations. Stefan Tilkov calls
these URI APIs. According to Ulsberg, this puts a huge
burden on the clients to understand how to build these
URIs, and removes a lot of flexibility from the server.

Misconception #3: REST APIs Should

be Versioned

Imagine if we had a database table called ‘Referer’. After

much use, we realized that the database name is mis-
spelled, and that we have to change it to ‘Referrer’. Since
the clients have already hardcoded the table name in
Designing a True REST State Machine 7

their SQL statements, you can’t update the table, since

this means you would have to update all clients— it would
be very messy.
Similarly, if we wanted to change one of our /blogposts/
URIs from above, we would be in the same tricky situation
— we would have to update all clients. This leads to
creating a v2, updating documentation, and asking clients
to kindly update everything. In short, having hardcoded
versioning in the URI is a painful process.
Upon whether or not to version web APIs, Roy Fielding’s
keynote presentation for the 2013 Evolve Conference
simiply stated:

“DON’T”

Misconception #4: Hypermedia is

Optional for REST APIs

Nope. As Roy Fielding himself stated in a 2014 interview

with Mike Amudsen:

” ‘Hypermedia as the engine of application state’

is a REST constraint. Not an option. Not an ideal.
Hypermedia is a constraint. As in, you either do
it or you aren’t doing REST.”

So what is hypermedia exactly? Well, REST consists of 6

major constraints. Out of these, HATEOAS is arguable the
most important and unique to the REST stipulation, but
also the least understood.
Designing a True REST State Machine 8

1. Client-Server
2. Stateless
3. Cacheable
4. Layered
5. Code on demand (optional)
6. Uniform Interface
• Identification of resources
• Manipulation of resources
• Self-descriptive messages
• Hypermedia as the engine of application state
(HATEOAS)

To understand HATEOAS, Ulsberg recommends we apply

the same thinking proposed by Don Norman in The Design
of Everyday Things. As a cup wants to be held and lifted,
or a button wants to be pushed, hypermedia wants to
tell you what to do with the resource. Hypermedia is
links and metadata for operations that help developers
or machines perform additional actions. As Ulsberg says:

“If you look at hypermedia as a recipe of how

the next request is supposed to look like, you
will grasp what hypermedia is all about”

Example State Machine: IoT Toaster

So let’s delve into what hypermedia actually looks like. To

describe hypermedia as the engine of application state,
let’s take a simple example of a possible state machine —
a connected toaster that can be manipulated through the
internet.
Designing a True REST State Machine 9

A state machine is a concept within computer

programming that is used to describe a ma-
chine that has a set of states, usually with
input and output events.

Our toaster begins in an off state, and when it is turned

on eventually reaches a heating state. Then it reaches
its upper temperature limit, enters an idle state, which
reduces the temperature. By being idle it cools, and goes
back into a heat state. It continues this loop, maintaining
a constant temperature until the bread is toasted, after
which it shuts down.
How could you manipulate such a toaster through REST
and hypermedia? Well, as Roy Fielding described, hyper-
media is the engine of application state. Therefore, each
page on the web represents a single state of a single
resource, which can be obtained using a GET call. It is
implicit that you are able to retrieve anything with an ID
at a URI. So, let’s first call our toaster to see what we get
back:

GET /toaster HTTP/1.1

The response looks something like:

Designing a True REST State Machine 10

HTTP/1.1 200 OK
{
"Id": "/toaster",
"state": "off",
"operations": [{
"rel": "on",
"method": "PUT",
"href": "/toaster",
"Expects": { "state": "on"}
}]
}

As you can see, we have an id at the top to denote what

resource we are communicating with. The current state
is displayed, as well as a list possible operations that we
may enact on the machine.
So let’s try to actually alter the state of the toaster. We’ll
send a PUT HTTP request for this:

PUT /toaster HTTP/1.1

{
"state": "on"
}

Our response will look something like:

HTTP/1.1 200 OK

{
"Id": "/toaster",
"state": "on",
"strength": 0,
"operations": [ {
"rel": "PUT",
"method": "PUT",
"href": "/toaster",
"expects": { "state": "off" }
}, {
"rel": "strength",
Designing a True REST State Machine 11

"method": "PUT",
"href": /fcesj48fl29304d827434j
"expects": {
"strength": [1,2,3,4,5,6]
}
}]
}

We have now turned on the toaster! However, as we can

see above, the strength is still set to zero, so it isn’t heating
yet. Let’s see what happens when we make a call to affect
the strength operation.

PUT /fcesj48fl29304d827434j HTTP/1.1

{
"strength": 3
}

So when we execute this strength operation, we see that

the state has changed, and the toaster is now heating:

HTTP/1.1 200 OK

{
"Id": "/toaster",
"state": "heating",
"strength": 3,
"operations": [ {
"rel": "PUT",
"method": "PUT",
"href": "/toaster",
"expects": { "state": "off" }
}, {
"rel": "strength",
"method": "PUT",
"href": /fcesj48fl29304d827434j
"expects": {
"strength": [1,2,3,4,5,6]
}
}]
}
Designing a True REST State Machine 12

We still have other options, as laid out in the hypermedia

above. We can chose to turn it off, or adjust the strength
again. But instead, let’s send another GET to the toaster ID.

HTTP/1.1 200 OK

{
"Id": "/toaster",
"state": "idle",
"strength": 3,
"operations": [ {
"rel": "PUT",
"method": "PUT",
"href": "/toaster",
"expects": { "state": "off" }
}, {
"rel": "strength",
"method": "PUT",
"href": /fcesj48fl29304d827434j
"expects": {
"strength": [1,2,3,4,5,6]
}
}]
}

Since our last request, the toaster has entered an idle

state. You can see that we can still opt to turn it off, adjust
the heating temperature with the strength
Wait a few minutes, and another request to the state
machine will likely result in a state of “shutting down”
which may include no operations, or “off” — the initial
resting state with the list of possible operations still in the
response.
Designing a True REST State Machine 13

Conclusion

The web functions as a web of interconnecting ideas,

linked with hypertext. Ulsberg believes that web APIs
should mimic this, and an important facet in doing so
is implementing hypermedia within our APIs. For those
newcomers to REST, Building a HATEOAS-compliant API
is a huge improvement over RPC-style APIs:

“If you use hypermedia, you can add relations

and links, and operations to the resources with-
out breaking existing clients, and at the same
time, giving new functionality to new clients.”

This also means rethinking the traditional stance on ver-

sioning. On versioning, Ulsberg thoughts echo Fielding’s:
don’t. When did you last see a versioning number on a
website? As HTML doesn’t need a version number, JSON
shouldn’t either.
By putting more emphasis on the resources and URIs, we
can retrieve operations directly from them. This is the
power of REST. This is the power of hypermedia.
Is GraphQL The End of
REST Style APIs?

The world of APIs is one of innovation and constant

iteration. The technologies that once drove the largest
and best solutions across the web have been supplanted
and replaced by new, more innovative solutions.
That is why it’s surprising, then, that many developers
have clung to what they consider the bastions of web
API development. Such a bastion is the REST architecture.
To some developers, REST is an aging and incompleted
proposition that does not meet many of the new devel-
opment qualifications required by the unique challenges
faced by modern development groups.
In this chapter, we’re going to look at a technology that is
poised to replace, or at the very least, drastically change
Is GraphQL The End of REST Style APIs? 15

the way APIs are designed and presented — GraphQL.

We’ll discuss a little bit of history, what issues REST suffers
from, and what GraphQL does differently.

Speaker: Joakim Lundborg

This chapter was inspired by Lundborg’s ses-
sion at the 2016 Platform Summit. Watch the
full talk here

“The way we design our APIs structures the way

we think about the tools and applications we
build.”

Defining REST and its Limitations

REST or Representational State Transfer, is an API design

architecture developed to extend and, in many cases,
replace older architectural standards. Objects in REST are
defined as addressable URIs, and are typically interacted
with using the built-in verbs of HTTP — specifically, GET,
PUT, DELETE, POST, etc. In REST, HATEOAS (Hypermedia
As The Engine Of Application State) is an architecture
constraint in which the client interacts with hypermedia
links, rather than through a specific interface.
With REST, the core concept is that everything is a re-
source. While REST was a great solution when it was first
proposed, there are some pretty significant issues that
the architecture suffers from. According to Lundberg, the
circumstances have changed, giving rise to the need for
new technical implementations:
Is GraphQL The End of REST Style APIs? 16

“Many things have happened. We have a lot

of mobile devices with lots of social and very
data rich applications being produced…We now
have very powerful clients, and we have data
that is changing all the time. This brings some
new problems.”

Here are some issues Lundberg sees with REST:

Round Trip and Repeat Trip Times

REST’s defining feature is the ability to reference resources

— the problem is when those resources are complicated
and relational in a more complex organization known
as a graph. Fetching these complicated graphs requires
round trips between the client and server, and in some
cases, repeated trips for network conditions and appli-
cation types.
What this ultimately results in is a system where the
more useful it is, the slower it is. In other words, as
more relational data is presented, the system chokes on
itself.

Over/Under Fetching

Due to the nature of REST and the systems which often

use this architecture, REST APIs often result in over/under
fetching. Over fetching is when more data is fetched
than required, whereas under fetching is the opposite,
when not enough data is delivered upon fetching.
Is GraphQL The End of REST Style APIs? 17

When first crafting a resource URI, everything is fine —

the data that is necessary for functionality is delivered,
and all is well. As the API grows in complexity, and the
resources thus grow in complexity as well, this becomes
problematic.
Applications that don’t need every field or tag still receive
it all as part of the URI. Solutions to fix this, such as ver-
sioning, result in duplicate code and “spaghettification” of
the code base. Going further, specifically limiting data to
a low-content URI that is then extensible results in more
complexity and resultant under fetching in poorly formed
queries.

Weak Typing and Poor Metadata

REST APIs often unfortunately suffer from poor typing.

While this issue is argued by many API providers and
commentators (often with the caveat that HTTP itself
contains a typing system), the fielding system solutions
offered simply do not match the vast range and scope of
data available to the API.
Specifically, this is an argument in favor of strong typing
rather than weak typing. While there are solutions that
offer typing, the delineation between weak and strong is
the issue here, not an argument defused by simply stating
“well there is typing”. The strength and quality of typing
does matter.
This is more a matter of age and mobility rather than an
intrinsic problem, of course, and can be rectified using
several solutions (of which GraphQL is one).
Is GraphQL The End of REST Style APIs? 18

Improper Architecture Usage

REST suffers from the fact that it’s often used for some-
thing it wasn’t really designed for, and as a result, it often
must be heavily modified. That’s not to say that REST
doesn’t have its place — it’s only to say that it may not
be the best solution for serving client applications. As
Facebook says in its own documentation:

“These attributes are linked to the fact that

“REST is intended for long-lived network-based
applications that span multiple organizations”
according to its inventor. This is not a require-
ment for APIs that serve a client app built within
the same organization.”

All of this is to say that GraphQL is functionally the end of

REST — but not in the way that terminology implies. Until
now, REST has been seen as the foundational architecture
of modern APIs, and in a way, the last bastion of classic
API design.
The argument here is not made to fully sever REST from
our architectural lexicon, but instead to acknowledge that
there are several significant issues that are not properly
and fully rectified by the solutions often proffered by its
proponents.
Therefore, the answer to the question of this piece — is
GraphQL The End of REST Style APIs? — is quite simple.
Yes, using GraphQL is the end of REST style APIs as we
know it — specifically through the extension of base
functionality and a reconsideration of data relations and
functions. ## 4 Things GraphQL Does Better than REST
Is GraphQL The End of REST Style APIs? 19

GraphQL declares everything as a graph… You

say what you want, and then you will get that.

Now that we’ve seen the issues with REST, how, exactly,
does GraphQL solve them?

REST Has Many Roundtrips -

GraphQL Has Few

The biggest benefit of GraphQL over REST is the simple

fact that GraphQL has fewer roundtrips than REST does,
and more efficient ones at that. GraphQL unifies data
that would otherwise exist in multiple endpoints (or even
worse, ad hoc endpoints), and creates packages.
By packaging data, the data delivery is made more effi-
cient, and decreases the amount of resources required
for roundtrip calls. This also fundamentally restructures
the relationship between client and server, placing more
efficiency and control in the hands of GraphQL clients.

REST Has Poor Type Systems -

GraphQL Has a Sophisticated One

While REST can have a type system through implemen-

tations of HTTP, REST itself does not have a very so-
phisticated typing system. Even in good implementations,
you often end up with variants of type settings — for
example, clientdatamobile and clientdatadesktop — to fit
REST standard calls.
Is GraphQL The End of REST Style APIs? 20

GraphQL solves this with a very sophisticated typing sys-

tem, allowing for more specific and powerful queries.

REST Has Poor Discoverability -

GraphQL Has Native Support

Discoverability is not native to REST, and requires specific

and methodical implementations of HATEOAS, Swagger,
and other such solutions in order to be fully discover-
able. The key there is “fully discoverable” — yes, REST
has HATEOAS as a “native” discovery system, but it lacks
some important elements of effective discoverability —
namely known document structure, server response con-
straint structures, and an independence from standard,
restrictive error mechanisms in HTTP.
While this and many other points of negative considera-
tion towards REST is often answered with “but you can
add that functionality!”, the fact that it lacks it by default
only adds to the complexity we’re trying to move away
from.
Because GraphQL is based on relational data and, when
operating on a properly formed schema, is self describing,
GraphQL is by design natively discoverable. Discoverabil-
ity is incredibly important, both in terms of allowing for
extensible third-party functionality and interactions and
for on-boarding developers and users with an easy to
understand, easy to explore system of functions.
Is GraphQL The End of REST Style APIs? 21

REST Is Thin Client/Fat Server -

GraphQL is Fat Client/Fat Server

In REST design, the relationship between client and server

is well-defined, but unbalanced. REST uses a very thin
client, depending on processing from the server and
the endpoints that have been defined for it. Since the
bulk of the processing and control is placed firmly on
the server, this strips power from the client, and also
stresses server side resources. Until now that has been
fine, but as devices grow in processing power and ability,
this client/server relationship may need rethinking.
GraphQL, however, is different. By offloading specifica-
tion of expected data format to the client and structuring
data around that call on the server side, we have a Fat
Client/Fat Server (or even a Thin Client/Thin Server de-
pending on approach) in which both power and control
are level across the relationship.
This is very powerful when one considers that the data
type being requested will be used for specific purposes
as regulated and requested by the Client itself — it makes
sense, then, that moving from a Thin/Fat relationship to
a Fat/Fat or Thin/Thin relationship would improve this
functionality on the Client side while freeing up Server re-
sources. Of course, this assumes that the client is capable
of handling this burden.

The End Of The Status Quo

There’s a tendency in the tech space for providers and

developers of new technologies to proclaim the end of an
Is GraphQL The End of REST Style APIs? 22

era with each solution. While it’s common to discussion

in the field, the fact is that there are very few complete
paradigm shifts that signal an irrevocable end to existing
technologies.
Innovation depends on prior technologies to create new
functionality. Therefore, when a new solution is designed,
it’s not replacing the solution, but rather iterating. The
same is true here. While GraphQL may not be the com-
plete demise of REST, it is the end of the status quo.
While there are a great many solutions to the issues
raised here, they all depend on further integrations and
modifications. GraphQL is essentially an overhaul, and
one which improves the base level functionality of the API
itself.

Conclusion

What we have here is a basic value proposition. GraphQL

does what it does well, but the question of integration lies
directly on what kind of data you’re processing, and what
issues your API is creating. For simple APIs, REST works
just fine, but as data gets more complex and the needs of
the data providers climbs, so too will the need for more
complex and powerful systems.
Adopting GraphQL as an adjunct or extension of the REST
ideology, while removing REST from the intellectual space
of “too big to not use”, will directly result in more powerful
APIs with easier discoverability and greater manageability
of the data they handle.
Continuous Versioning
Strategy for Internal APIs

Recently, there has been debate over what the best prac-
tices are for versioning an API. Many public web APIs are
retired as new versions replace them, but if you were to
ask Roy Fielding, creator of REST, he may tell you not to
version your API at all.
Some companies are taking matters into their own hands,
and seeking out innovative ways to handle the cumber-
some process of keeping their Application Programming
Interfaces up to date in a way that makes sense to their
business model. These new strategies place more empha-
sis on evolution rather than deprecation.
The typical v1, v2, v3 etc. versioning approach focuses
on releasing large sweeping updates to improve the API
experience. But the downside of this method is that it
causes a major breaking change on the client side. For
internal API-first companies that have granular control
over their various web, desktop, and mobile clients, con-
tinuous versioning could be a more attractive strategy.
In this chapter, we’ll review how public web APIs are
typically versioned within our domain, and discuss why
companies may want to consider a continuous versioning
strategy for handling complex APIs that are subject to
Continuous Versioning Strategy for Internal APIs 24

continual, iterative evolution. Led by Platform Summit

speaker Konstantin Yakushev, we’ll use Badoo as a case
study to peek into an alternative approach to versioning.
Benefits like feature negotiation, and allowing for exper-
imental development tracks could make continuous ver-
sioning strategy a win, especially for private API systems.

Speaker: Konstantin Yakushev

This chapter was inspired by Yakushev’s ses-
sion at the 2016 Platform Summit. Watch the
full talk here

Typical Public API Versioning

Within most public scenarios, an API service is updated

by creating an entirely new v2 and slowly deprecating
the original v1. Problems with v1 are tracked — perhaps
a product order is misspelled, or the business logic has
changed. All these edits are accumulated and introduced
in a v2 that solves these issues, but introduces a complete
breaking change with the previous version.
An API with an endpoint such as http//api.example.com/orders,
is typically reworked with a URI extension to something
like ‘http//api.example.com/v2/orders’. The v1 is then sched-
uled for retirement, usually in accordance with a depre-
cation policy. Though this is the norm, there are some
significant negatives of this approach:

• Long Timeline: Instead of incremental edits, with

versioning you must wait for all changes to be bun-
dled. This means you can’t be that agile in respond-
ing to specific user requests.
Continuous Versioning Strategy for Internal APIs 25

• Breaking: Whether you like it or not, releasing an

v2 is inherently breaking the connection, and will
require all clients to eventually update their connec-
tions.
• Communication: Time and resources must be spent
to communicate API changes. With a v2, documen-
tation must be updated, and deprecation timeline
notices must be sent to consumers.
• Fielding as a Friend Factor (3F ): Roy Fielding de-
fines evolvability as the ability to change over time in
response to changing user needs or a changing en-
vironment without starting over. It’s actually against
Roy Fielding’s own recommendation to version your
API, saying it’s “only a polite way to kill deployed
applications.”

Many typical versioning strategies focuses too heavily on

the URL construction, which to Yakushev, is “the least
important step, in my opinion.” Instead, it may be better
to consider the entire process from a more holistic van-
tage point. When we look at the API update process, we
see that perhaps there is no v2 — after all, much is often
salvaged, and introducing an entire new version may not
be worth the effort in updating all clients.

Badoo’s Continuous Versioning

Strategies

When API-first companies consistently iterate with con-

tinuous versioning, the problems listed above dissolve.
To see how this actually works in practice, let’s consider
some specific use cases from Badoo, the Russian dating
network and app.
Continuous Versioning Strategy for Internal APIs 26

Badoo is the largest dating network in Russia, and they

have been evolving an internal API since 2010. They’ve
never had a breaking change as they’ve been incremen-
tally updating all this time. Konstantin frankly admits
that the API is not strictly RESTful, rather RPC-style and
Protobuf based for mobile clients, and JSON based for
their web clients
With nearly 600 commands and over 1,200 classes, the
API receives around 9 updates each week, and supports
5 clients (iOS, Android, Windows, Chrome, Safari) with a
healthy backwards compatibility for older client versions.
Let’s take a look into Badoo’s internal API strategy to see
how they’ve used a continuous versioning mindset for
specific updates to avoid major, breaking changes.

Changing the Verification Process

Yakushev describes how Badoo needed to rework their

verification process. In the past, if a user signed up with
their social account, they received a social checkmark as-
sociated with their account. As time grew on, the design-
ers wanted to have more rigorous checks. For example, if
a user were to verify with photo verification, they should
receive a different badge.
The problem was that the original verification had a bi-
nary logic that affected other aspects of the app — the
users were either verified (true) or not verified (false).
Since that was the case, adding a new verification com-
plexity meant instituting a dramatic change to their API
behavior.
The Badoo team was able to solve this issue by using
Continuous Versioning Strategy for Internal APIs 27

GraphQL to list the acceptable fields for clients. Now,

when clients request the verification status, they receive
more customizable options. Allowing clients to negotiate
new fields is a way Badoo can update their API while
keeping endpoint consistency. The old clients can use old
fields, whereas the new clients use new fields.

Updating Banner CTAs for Specific

Clients

However, Yakushev recognizes harder challenges in keep-

ing their API updated and consistent across various clients.
For large changes, he advises releasing new features
on the server, and making clients end supported types
explicitly.
For example, Badoo needs to serve various call-to-action
banners for different screen sizes and device-specific
interactions. If a new banner type is introduced, however,
when the client asks for banners, the server could send an
unknown or old banner. Typical versioning is not flexible
enough here.
To solve this issue, Badoo introduced a list of supported
banner types to easily decide which banners will be
shown to the client. Now, client specific banners, such as
swipeable mobile-only logic can be paired with the right
receiving device using the same, albeit stateful, API.
Continuous Versioning Strategy for Internal APIs 28

Use Flags to Avoid Versioning in

Complex Business Logic Changes

What about more complex high-level changes to busi-

ness logic? Yakushev explains how all Badoo profiles
have a photo feed attached to them. Over time, the
design team wanted to mix in videos with the photos, and
add a play button to watch the videos from within the grid
view.
To resolve the issue without versioning the entire API,
Badoo introduced a supported changes array. This way,
the client knows that the server may send videos along
with photos. A similar approach can work in many other
cases — essentially you release changes behind a version
flag, and make the client control these flags.

Run Experimental Features

A benefit of Badoo’s hands on approach to the entire API

lifecycle is the ability to run quick experimental features
on select platforms. To do this they create a superset
experimental API that is only used on a select platform,
such as the Windows phone, as it las low usage. Having
multiple development tracks enables new features to be
tested and engagement monitored.
Continuous Versioning Strategy for Internal APIs 29

How Continuous Versioning Could

Apply to You

Depending on the situation, continuous versioning could

be a powerful ally in developing and scaling agile web
APIs. Instead of instigating breaking change, fields for
new features are added, and the client has a list of sup-
ported items to send to the server. Yakushev recom-
mends covering new changes with change flags, and
letting the server control enabling and disabling features.
At the end of the day, client developers and product
owners are happy. An iterative versioning strategy my put
additional pressure on backend developers, but they may
like the ability to split their work into parallel tracks for API
supersets and experimental features.
In practice, Badoo has nearly 260 feature flags, and 160
negotiable features. Implementing feature negotiation
at this level of complexity is more easily accomplished
within an internal scenario, where communication be-
tween teams is united and both client developers and API
developers are working toward similar end goal.
In Q&A discussion it was found that continuous version-
ing still may may not be an ideal method for public API
providers. Since within continuous versioning new fields
essentially equate to new features, some believe this
level of feature negotiation is only acceptable when you
control both the API and clients. Performing continuous
versioning within a public API scenario may still need
thought, nonetheless is an alluring proposition.
Case Study: Spotify
Internal Payment APIs

Adding new layers of complexity within a digital service

without sacrificing user experience is a difficult endeav-
our. Especially for platforms that accept online payments
and subscription formats, maintaining support for an
increasing number of payment methods is deceptively
complex.
Users don’t want to think too much about the payment
process. They simply want payments to work with their
preferred, custom method. However, the number of on-
line purchase methods has soared in recent years —
digital platforms must now consider support for not only
credit cards or bank accounts — but Paypal, cryptocur-
rencies like Bitcoin, Facebook payments, Google Wallet,
Apple pay, Klarna, Boku, Sofort… the list goes on.
All this is expected to behave with zero issues, but un-
der the hood, handling diverse payment methods in an
efficient way is becoming increasingly complex, requiring
some serious forethought.
At the 2016 Nordic APIs Platform Summit, we were of-
fered an exclusive sneak peak into the architecture that
Spotify software engineers deploy in order to accept
millions of user subscription payments each month. As
Case Study: Spotify Internal Payment APIs 31

they build internal APIs to handle the complexities of their

payment subscriptions, their platform is an excellent case
study into the power Private APIs have to streamline
internal operations.
Despite architecting a custom system for their own busi-
ness domain, the Spotify payment model could surely
inspire other digital enterprises to introduce similar, API-
driven scalable payment subscription formats into their
own service framework.

Speaker: Horia Jurcut

This chapter was inspired by Jurcut’s session
at the 2016 Platform Summit. Watch the full
talk here

The Evolution of Spotify Payments

To begin with, let’s consider how Spotify evolved. In 2006,

Spotify was developed by a small team in Stockholm,
and was launched to the Scandinavian market in October
2008. At that time, they only accepted credit/debit cards
as a payment method.
By the end of 2011 their market increased again, spread-
ing throughout Europe and into the United States. It was
then that they added support for Paypal. In 2013 their
market size increased again, and they added Boku, Sofort,
Klarna, Google in-app purchases (iAP), and Facebook pay-
ments as optional payment methods.
Within this two year period their growth was impressive.
As Spotify was emerging into many local markets, they
Case Study: Spotify Internal Payment APIs 32

decided to shift their focus from only accepting global

payment methods to adding support for more local ones.
To do this, in 2015 they partnered with the payment
service Adyen, which allows companies to support over
250 different payment methods and 150+ currencies —
truly a globalized, yet locally aware solution. They added
9 other payment methods to their stack, including bank
transfer and preloaded cash cards, which increased the
brand engagement into areas that have seen low credit
card penetration.
In 2016, they now have roughly 40 million worldwide
subscribers in 60 markets. Beside delivering great musi-
cal playlists and audio experiences, their remittance goal
remains consistent across the globe — to help users
select most convenient payment method.

What’s Really Going on with Online

Payments?

Things like subscriptions seem like an easy process, but

the workflow required for even a simple credit card pay-
ment has hidden complexity, with many working actors
as well as unique edge cases to consider. First, let’s break
down the workflow of a typical online payment made
using a Visa credit card.

1. The user opens a line of credit with a Bank and is

given a card.
2. They type their card number into a Checkout screen
that ties into a Payment Backend.
3. Next, the Payment Backend will attempt to autho-
rize the payment details with the Payment Provider.
Case Study: Spotify Internal Payment APIs 33

4. If it succeeds, the Payment Provider will contact

the Bank through the Credit Network to set aside
money for a subscription.
5. Finally, the Bank confirms that the purchase is pos-
sible, and the response comes back to the Payment
Backend through the Credit Network.

After this common flow is complete, an online subscrip-

tion will be engaged. How is this communication archi-
tected? For Spotify, the process is supported by three
main actors: the Client with the checkout page, the Pay-
ment Backend, and the Payment Provider. In reality,
however, Spotify supports 16 different payment providers,
each with unique APIs that have different implementa-
tions.
So how should a platform manage this complexity? To
do it Spotify developed two smart tools: The Checkout
API to help build flows that make it easy for users to
enter payment details, and the Billing API to interface
with the various details of Payment Providers and Credit
Networks, enabling the Payment Backend to determine
if they can charge a user for a subscription with a single
call.

The Checkout API

The first main component is the Checkout API. To begin

a subscription, the Payment Backend must try to validate
payment details, and initiate the workflow mentioned
above. However, in order to do this, it first needs to
gather information from Spotify internally, which is in
effect sourced from user input on clients like mobile or
Case Study: Spotify Internal Payment APIs 34

desktops. Since the type of clients are manifold, Spotify

uses a state machine.
Therefore, their Payment Backend dictates the opera-
tions, controlling the workflow and requesting data from
the clients. The Checkout API thus facilitates the commu-
nication between the Payment Backend and the clients.
This is necessary for the fact that extra steps may exist
in a check out workflow. For example, in some countries,
filling in tax information may be required. The Spotify Pay-
ment Backend needs to inform clients of the existence of
this step, and the clients need to respond with a nuanced
form, and collect other data as well. Other special cases
may involve an offer code from an email invite to join.
For all these scenarios, the backend decides what data
is required, and the client must be populated with the
correct form.
The key API that is driving these experiences is the Check-
out API, sitting in between the Spotify clients and Payment
Backend. This API enables them to create customized
forms in an infinite number of iterations within Spotify
clients.

Simple Checkout API Flow

1. The Client initiates a purchase, which triggers the

Checkout API.
2. The Checkout API asks the Payment Backend what
the next step is.
3. The Payment Backend responds saying that credit
card information needs to be collected.
4. The Checkout API sends this to the Client, who uses
this information to display a custom form for the
user to input data into.
Case Study: Spotify Internal Payment APIs 35

5. The Checkout API then sends the card data to the

Payment Backend.
6. Then the Payment Backend processes the payment,
confirms with the Checkout API, which initiates the
Client to display the confirmation page.

These steps could easily be extended for custom form

scenarios. For this to work, the business logic is always
resting on the backend to dictate processes and for the
clients to respond with appropriate information.
There are a few benefits to architecting a payment pro-
cess this way. Having the Checkout API between the
backend and client means that you can alter the checkout
experience quite easily. Clients themselves can also now
create native experiences, meaning that data can be mod-
erated throughout a device to alter other apps or device-
specific functions.

The Billing API

After the Payment Backend receives payment informa-

tion, it needs to communicate with the Payment Providers
to charge the user. Spotify’s Billing API is the bridge
between the Payment Providers and their Payment Back-
end; the second interior API necessary for this payment
flow to occur.
Since Payment Providers have different business logic
(a credit charge is instant, whereas a bank transfer may
take 3-5 days, for example), the Billing API is very use-
ful. It translates these different implementations into
usable data, hiding the complexity of 16 different Pay-
Case Study: Spotify Internal Payment APIs 36

ment Providers from the Payment Backend, consolidat-

ing things into a single call.
The Billing API is also responsible for handling financial
data, settlement, monitoring, and other duties. This al-
lows the team to compare different payment methods,
such as seeing the usage differences between Paypal and
a method where SMS confirmation is required.

Sample Use Case: Automatic Alerts

A main facet of the DevOps approach is having auto-

mated alerts to allow for faster response times to sys-
tem issues. Spotify leverages the internal Billing API to
provide rich platform monitoring features, that are then
repurposed throughout the enterprise in what they call
their “anomaly detection system”. The Billing API is able
to turn actions from the Payment Providers into consis-
tent strings of events, which can then be monitored and
filtered.
The varying actions for all remittance methods are boiled
down into JSON exchanges with fields such as message,
amount, provider, failure_code, and more. Producing stan-
dardized events for all types of payment methods en-
ables alerting which can inform development and oper-
ations.
Keeping track of received payment transactions, and mod-
eling this with past transaction histories, means that the
system can detect trends in payment flaws. So, when a
developer deploys a change that breaks the production
environment, the team receives an alert. Using this same
data to populate visualizations, the team can also detect
Case Study: Spotify Internal Payment APIs 37

if a certain payment provider is having an issue, as the

data allows them to isolate individual provider activities.
Having a monitoring system for a billing API allows a
company to follow trends, and become aware of local
events, like bank holidays. It also can help automate
internal accounting; Spotify’s approach is so fine-grained
that if a small payment provider misses a due date/time
for sending batch requests, the Spotify alert system will
immediately notify the team.

Repurposed from original diagram within Hurcut/Spotify presentation

4 Reasons Why API Design is Critical

to Subscription Services

Of all the approaches to handling payments, why should

subscription services design their own internal APIs? There
Case Study: Spotify Internal Payment APIs 38

are a few benefits:

• APIs are platform independent: First, web APIs are

designed to be used by any client. This helps hide
complexity when dealing with a large number of
clients and payment providers. Since the purchase
method landscape is still evolving, adaptable inter-
faces are very important.
• Internal integrations: Other benefits include inte-
grating with accounting platforms, payment security,
or user activity monitoring. Having well-defined APIs
for subscription services enables powerful data inte-
gration capabilities with other internal systems.
• Scale for growth: As a business grows, the need to
build more complex products will become a concern.
Therefore, having good API design helps prepare for
compatibility with new business logic that will affect
the company.
• Customization potential: There are many payment
providers, aggregation services, or payment gate-
ways within the market, but building your own APIs
is a way to ensure customization and efficiency.

Horia Jurcut, Software Engineer at Spotify, also describes

his personal affection for APIs and the positive reverber-
ations they create within a modern business:

• APIs make it easier for multiple teams to collaborate

• APIs help you take a hard problem and divide it into
more manageable domains
• APIs require documentation
• APIs enable you to rapidly test, experiment and learn.
Case Study: Spotify Internal Payment APIs 39

Analysis: Scalable API-Driven

Infrastructure will Power the Future
of Online Payments

Spotify has impressively balanced the small and large

with their payment architecture. They have successfully
arbitrated complexity that has arisen out of a need to
serve customization options to their users, and in doing
so, have created a scalable architecture for accepting
multiple payments that could be modeled at other com-
panies.
By building consistent constructs that interface with mal-
leable components (payment providers and clients) through
APIs, they are able to both scale for global use while
integrating with small payment providers, allowing their
brand to cater to very local, niche audiences in specific
markets.
On the user side, adding support for local payment meth-
ods is one way Spotify has honored a pledge for great con-
sumer experiences. As the Spotify music service excels at
offering custom, curated musical playlists based on user
behaviour, it makes sense that their payment models
would allow for custom methods as well. Platform-wide
consistency with an end goal to improve UX should,
similarly, be the end goal at other companies. It seems
that a significant portion of Spotify’s rapid, global growth
is due, in part, to a well-architected payment backend.
The Spotify payment strategy is conformable within other
businesses, and can act as a great architectural model
for new end-user facing digital services that intend to
expand to specialized markets with local payment op-
tions. Since the number of payment providers continues
Case Study: Spotify Internal Payment APIs 40

to climb, digital companies will very soon need to plan

for how they can most efficiently accept and monitor new
payment types within their subscriptions.
As the decisions made now regarding online payments
will be crucial for digital platform longevity, consider how
your business could parcel settlement functionality more
efficiently. Whether or not you need a custom built solu-
tion is up to you. Exciting emerging financial technology
APIs like Klarna, Plaid, Stripe, Transpay, or other FinTechs
will likely aid SMBs, unlocking differing payment methods
so they may become globally recognized and supported
services.
Though fine-grained payment options may not currently
be necessary for all digital services, their ubiquity is on the
rise. At the very least, from this case study we can see the
benefits of creating distributed internal microservices:
abstracting functionalities is key to scalability.
The Benefits of a
Serverless API Backend

Imagine if your backend had no infrastructure. No perma-

nent server, nowhere for your API to call home. Sounds a
bit bleak, doesn’t it? As it turns out, serverless backends
could be the next big thing for implementing truly scal-
able cloud architecture.
How do you make an Application Programming Interface
lightweight on the client side, yet scalable to heightened
traffic demands? SaaS vendors have been migrating to
serverless architecture to solve this dilemma, as well as
many other operational issues found in hosting their web
applications.
In this chapter, we’ll identify what the serverless craze is,
and why some providers may want to consider having
The Benefits of a Serverless API Backend 42

a serverless API backend. Led by Rich Jones of Gun.io,

we’ll define what we mean by serverless API backends,
provide an example of one practice today, and aim to
outline some potential benefits and pitfalls of adopting
this approach.

Speaker: Rich Jones

This chapter was inspired by Jones’ session at
the 2016 Platform Summit. Watch the full talk
here

What Does “Serverless” Mean?

Traditional cloud hosting is permanent.

As in, you pick a server provider, and
they run you software on multiple servers
worldwide. There are precise, recurring
physical locations for where your data is
A serverless
stored and functionality processed. backend scales
to the needs of
Serverless computing is a strategic devi- API requests.
ation from this model — it is an event-
driven setup without permanent infrastructure. This
doesn’t mean servers are no longer involved, rather, it
means that servers are auto-created on a per-need basis
to scale to the demands of your app.
But for developers, what serverless really means is less
time spent on operations, since they no longer have to
worry about traditional server maintenance. The benefits
of a serverless infrastructure really add up:
The Benefits of a Serverless API Backend 43

• No more over capacity issues

• Servers are autoscaling
• You don’t pay for idle time
• Consistent reliability and availability
• No load balancing, no security patches

In general, serverless simply equates to peace of mind —

but perhaps not for some, Operations may need to find
another job altogether.

From Traditional to Serverless

Environments

To understand the subtleties between traditional and

serverless approaches, let’s walk through a basic step-by-
step sample implementation of each.

Traditional Web Request

An interaction with a traditional web server will often

occur in a format similar to this:

1. An Apache or NGINX web server listens for events as

they come in.
2. The server then converts this to a Web Server Gate-
way Interface (WSGI) environment.
3. This is sent to an application to process the request.
4. Then the web server sends the response back to the
client.
5. The web server resumes listening.
The Benefits of a Serverless API Backend 44

There are a few drawbacks of this approach. For one, if

you encounter a huge spike in traffic, this system will
deal with the requests as they came in. If the end user
isn’t ahead in the queue, they will likely experience a
timeout, and the page will look like it’s down. Medium to
late visitors to the queue face very slow speeds. Secondly,
when it’s not processing a request, the web server is left
in an idle state to poll, wasting valuable resources that
could be used elsewhere.

Serverless Web Request

Within a serverless infrastructure, each request corre-

sponds to it’s own server. After the server processes the
function, it is immediately destroyed. For example, here’s
how Jones’s Zappa handles a web request:

1. The request comes in through an API Gateway.

2. The API request is mapped to a dictionary using
Velocity Template Language (VTL).
3. A server is created.
4. The server then converts the dictionary to a stan-
dard Python WSGI and feeds it into the application.
5. The application returns it, and it passess it through
the API Gateway.
6. The server is destroyed.

Astoundingly, all this occurs under 30 milliseconds, so

that “by the time the user actually sees the [content appear
on the] page, the server has disappeared… which is actually
a pretty zen thing if you think about it,” says Jones.
So what are the advantages to spawning servers on a
moment’s notice? To Jones, the top reason is scalability.
The Benefits of a Serverless API Backend 45

Since a single request matches to a single server creation,

this relationship can be scaled indefinitely, on a scale of
literally trillions of events per year.
Second is cost savings. Paying by the millisecond means
that you are only spending money on actual server pro-
cessing. AWS Lambda charges around $0.0000002 per
request. But since Lambda tier offers 1M free requests
per month, this means it could stay free for small projects
or young startups.
This infinite scalability make serverless infrastructure a
boon for both small breadth projects like microservices,
APIs, IoT projects, or chatbots, but also for larger tradi-
tional enterprise content management systems like Django
as well.

How Get Started: Understanding the

Serverless Vendors

Sound interesting? An easy way to get started is with a

serverless framework like Zappa, Serverless Framework,
or Apex (more here). With some frameworks, like Zappa,
you can adopt serverless computing for existing APIs. All
three are built around AWS Lambda, Amazon’s cloud com-
puting service, but other significant serverless computing
providers are within offerings by Microsoft Azure Func-
tions, Google Cloud Functions, and IBM Bluemix Open-
Whisk. However, according to Jones:

“AWS Lambda is by far the leader in the space…

it’s just far more capable in pretty much every
regard. The others are still playing catchup.”
The Benefits of a Serverless API Backend 46

Designing Event-Driven Serverless

Applications

Within a serverless environment, a main design element

that will be novel to newcomers is that code is going
to execute only in response to events. Since building
a robust, event-driven application means designing in-
response to events, what can we define as our event
sources?
An event may be related to file operations — for ex-
ample, say a user uploads an image and the application
needs to resize a large picture into a small avatar. Using
a serverless architecture, you could have a thumbnail
service execute a response in an asynchronous and non-
blocking way. Instead of setting up an entire queuing
system, having a native cloud hosted queue can handle
this.
Support notifications like receiving an email, text, or
Facebook message could also be interpreted as events.
Rather than polling for new emails to come in, an ac-
tion could be executed specifically in response to these.
Where it gets really interesting is how you can treat HTTP
requests as an event. This paired with other event trigger
types is usually called a hybrid architecture.
Database activity could also be used as an event trigger.
A change to a table’s row could trigger an action to hap-
pen, for example. However, Jones reminds us to treat the
“API as the primary source of truth in your application” —
don’t make SQL calls inside your event functions, rather,
funnel this through your API.
Jones reminds us that time is also an important factor
that can be used as an event source, and will be needed
The Benefits of a Serverless API Backend 47

to initiate regular occurring tasks or updates. Throughout

these varying event sources, instead of creating machines
that constantly poll your resources for changes, you’re
essentially setting up triggers within your applications to
execute a response.

5 Serverless Pro Tips:

All this sounds awesome, but what are the downsides

of building applications with serverless backends? In his
presentation, Jones covers some ground on potential
downsides, how to avoid them, and some general tips for
getting the most out of a serverless arrangement:

• Avoid vendor lock-in: This can be a big issue when

adopting any new technology. To avoid vendor lock-
in Jones recommends integrating software that pro-
vides open source compatible offerings, and to de-
couple vendor interactions from your application.
Rather than hardcoding interactions, Jones recom-
mends decoupling this logic — creating a dispatcher
inside of a function to add an item to the queue is
one way of doing so.
• Mock your vendor calls for testing: When writing
a mock or sample app that behaves as if synced to
the cloud, you may want to test your cloud functions.
Placebo is an interesting package that will record
your actions with AWS and replay them as if you
were interacting with the server.
• Think “server-lessly” and avoid infrastructure: It
can take a while to develop the serverless mindest.
When developing, consider if you actually need a
database, or if a queue can be adopted instead.
The Benefits of a Serverless API Backend 48

• Stage different environments: When testing and

staging, Jones recommends using CI for multiple pro-
duction environments (Blue/Green Deployment).
• Deploy globally: Using a geographically distributed
server arrangement can increase speed and security.
AWS Lambda services can host on 11 regions, so that
anywhere on the planet can reach a 20 millisecond
ping.

Example: Kickflip SDK

So how do we build an authenticated API, with low-la-

tency, low cost, that is infinitely scalable, without having
to worry at all about server operations? Let’s turn to an
example serverless implementation in action.
Kickflip is an SDK that brings live streaming video to
mobile applications. A “live stream” is essentially just a
combination of separate MP4 files, along with a manifest
that determines the order of the videos. Since a real-
time video stream service wouldn’t need to keep large
amounts of video data around for later use, it is an ideal
application for a serverless environment.
Kickflip uses a hybrid architecture of HTTP and non-HTTP
event sources to trigger server creation from a mobile
phone upload, which updates the manifest file so that
end users view the latest video chunks. To do all this,
Kickflip uses a combination of services: API Gateway for
authentication, an API constructed with Lambda, Zappa,
and Flask, file storage using S3, and CloudFront for global
content delivery. The simplified flow is as follows:

1. The client authenticates with the API. Kickflip uses

The Benefits of a Serverless API Backend 49

Amazon’s authentication API key generation service,

but a custom identity access management handler
could work here as well.
2. The API returns a short-lived federation access to-
ken which can only be used to upload a file into a
specific S3 bucket.
3. The client receives the token, and uses it to upload
the video.
4. An AWS Lambda server is executed in response to
the new video upload, and the stream manifest is
updated. This upload acts as the event-source.
5. Content is served on the CloudFront delivery net-
work for low-latency.
6. Users see the latest video stream on their device.
7. The server is destroyed and temporary access token
revoked.

Jones demonstrates that with the strategic pairing of

technologies, a serverless video streaming service can be
developed in only 42 lines of Python.

Building Serverless API Backends

The serverless movement represents a profound paradigm

shift in our ability to create impressively scalable web ser-
vices. Rethinking how events can spark temporary server
iterations can be an extremely cost effective solution for
microservices and large projects alike.
With all the small connected services being deployed in
this manner, the serverless arrangement also reiterates
the rise of composable enterprises that depend on many
The Benefits of a Serverless API Backend 50

different services to thrive; cementing the web API’s po-

sition as an important cog in modern and future web
communication.

Additional Resources

• Awesome Serverless: Helpful curated list of server-

less tooling and helpful information
• What is Serverless Computing?
• Ten Attributes of Serverless Computing Platforms
• The Serverless Stack: Step-by-step tutorials for creat-
ing serverless React.js apps
Putting an End to API
Polling

APIs have been around for a long time. While that means
there’s a lot of great tools from a lot of amazing develop-
ers, it also means that, as a community, the API space has
held on to some practices for a long time – some would
argue too long in many cases.
One such holdover, according to some developers, is the
concept of polling. While polling itself is not a bad thing
– after all, it’s a simple implementation of an endpoint
call – many argue the effects of constant polling require
a solution, and an immediate one at that.
Enter REST Hooks. Today, we’re going to take a look at
the concept and application of REST Hooks, and exactly
why some argue for their necessity. We’ll discuss some
objections to the idea of polling, and the responses from
their supporters.

What is the Polling Madness?

Polling Madness is a concept championed by API provider

Zapier that simply states the pattern of calling an end-
point for new data, or “polling”, is wasteful. The idea is
Putting an End to API Polling 52

that the constant polling of an endpoint is wasteful in

terms of resources committed to the action from the
developer, in terms of the traffic seen by the vendors,
and in terms of actual result to effort. While there are
some objections to this concept, it’s got some merit to it,
especially in the modern day of lean, efficient processing.

Polling is the same as the refresh button. It’s not a

viable solution.

Part of why polling has stuck around for

so long is because it is ubiquitous – un-
til recently, there were very few effective
ways to limit the negatives of polling, and because ev-
eryone was doing it, it was a very hard proposition to
get other developers to move away. At best, you’d get a
new solution that was a distant third option, and at worst,
you’d get a powerful but ignored product.
The problem is that polling is essentially just hitting a
refresh button – and, just as if the user was hitting the
refresh button themselves, depending on the refresh
button for vital functionality is simply unacceptable.

One Solution: REST Hooks

A solution from Zapier is actually quite simple – POST a

subscription /api/hooks that collates hooks at a target
URL, which then pings the resource requester when a
change is noted. It’s a subtle change, but it moves re-
sources from constant rechecking in active fashion to
passive waiting for updates.
Putting an End to API Polling 53

Counter Arguments

Many argue that this issue is not really an issue at all -

many seem perfectly fine with the current status, or at
the very least, the current solutions on hand.
One of the chief arguments against REST Hooks is that it
flies in the face of REST itself. REST is specifically supposed
to be session-free, and so the idea of creating a constantly
polling, static URL on a session-less system seems coun-
terintuitive.
Further issues are raised with the idea that there’s no
current solution which does what REST Hooks is intending
on doing. Some would argue that this, too, is false –
TCP/IP websockets should be capable of doing what is
being asked of the REST Hook.
The issue really, then, is one of effectiveness. Yes, it is true
that REST is session-less in most implementations, but
that does not mean that one can’t benefit from the pos-
itive attributes of sessionful communication while main-
taining the overall benefits of REST. So too is the argu-
ment for TCP/IP somewhat flawed - yes, TCP/IP websock-
ets can somewhat do what is being asked, but there are
issues (loss of control for the user, less customizability,
etc.) in those approaches.

Watch: Audrey Neveu

This chapter focuses heavily on REST hooks.
For more advice on real-time streaming APIs,
watch Audrey Neveu present at the 2016 Plat-
form Summit. Watch her talk here

So REST Hooks should be seen for what it is - one of many

Putting an End to API Polling 54

solutions, unique in its application, that can be used to

solve a complex, consistent issue in a new, unique way.

Implementing REST Hooks

Conceptually, the implementation is rather simple, though

it does come with specific concepts that must be adhered
to if the implementation is to be considered properly as a
REST Hook. REST Hooks are essentially URLs which collate
changes that would usually be polled, providing a URL
that can be actively monitored.

Create a Subscription

In order to escape the issues of polling a webhook, the

webhook itself needs to be tied to a subscriber URL.This
can be done with a simple POST element:

POST <subscribe_endpoint> \
-H Authenticated: authenticationSolution \
-H Content-Type: application/json \
-d ‘{“target_url”: “https://hooks.zapier.com/hooktest>”,
“Event”: “user_created”}’

This chunk of code sets up the URL by which you can

direct polling webhooks to the REST Hook. It uses the
standard authentication/authorization solution you’ve al-
ready implemented. Zapier responds with three elements:

• Authenticated User
• target_url
• event
Putting an End to API Polling 55

All three of these items are stored in local persistent data,

which can be called upon when accessing the REST Hook.
Of note is that Zapier recommends a few status codes
for certain behaviors. For successful subscriptions, a 201
code, signifying “Created”, should be returned. Likewise,
when a non-unique subscription URL is set, a 409 code
should be returned, signifying “Non-Unique”.

Sending Hooks

Now that the hook is setup, we need a way for the API to
actually send the data. We can do this by implementing
the following POST code:

POST https://hooks.zapier.com/testhook \
-H Content-Type: application/json \
-d <json payload>

Zapier notes that they typically expect an array of objects

– if the API sends a single object only, it needs to be
wrapped in the following element array:

[ {“element1”: “Content1”, “element2”: “Content2”} ]

Unsubscribing and Setting Up a Global URL

Finally, a DELETE call is made to unsubscribe using the

following code:
Putting an End to API Polling 56

DELETE <unsubscribe_endpoint> \
-H Authenticated: authenticationSolution \
-H Content-Type: application/json \

In order to define a polling URL permanently, there is an

option to set up a trigger. This trigger will allow a user
to set up a permanent data point, rather than having
to create a new one to poll the endpoint. The polling
URL alleviates this issue, creating a permanent location
to draw information and data from.

Alternatives

WebSockets are bi-directional - you tell both the

client and the data to send that data to each
other. Server-Sent Events is unidirectional - its
an open channel where data can be streamed
from the server to the client.

There are some alternatives, of course. WebSockets can

provide a constant back and forth connection, negating
polling in a way by moving the relationship from single-
directional “pull” to “push”. This is questioned by some,
though, as being essentially “polling 2.0”, in which the
polling is from two sources rather than one.
Server-Sent is yet another solution. In this approach, the
server does not respond to a poll request, but instead
constantly “pulses” changes to the client itself. The prob-
lem here is that, while it removes the polling technically,
it also removes much of the control from the user, and
forces them into a passive state.
Because of this, many feel that both solutions are es-
sentially inverse versions of polling, and are thus not
Putting an End to API Polling 57

acceptable solutions. Given the use case, they might be

appropriate, and in others, they could be opposite of
what is needed.

Conclusion

Implementing REST Hooks can solve a huge problem, but

it rests on the developer to figure out how much an
issue it truly is - on the one hand, polling is a consistent
resource dedication that can result in slower response
times when multiplied over many hundreds of connec-
tions.
On the other hand, there are many solutions that already
exist, though they each have strengths and weaknesses
opposite that of the Zapier REST Hook solution. If a devel-
oper does not have a polling solution, this is a fine one
indeed - if, that is, the developer believes they have a
problem to begin with.
6 Ways to Become a
Master Microservice
Gardener

The API space is an extremely interesting one, largely due

to it straddling a dichotomy of opposing requirements.
On the one hand, an API must be innovative, quick to
change, and ever evolving. On the other, consumers de-
mand stability, and with it, the security that comes with
proven solutions.
The constant battle between old and new, centralized and
distributed, has led to a unique range of solutions and
approaches to solving the dilemma. One such solution
6 Ways to Become a Master Microservice Gardener 59

is the concept of microservice gardening, the idea of

being a “landscaper” for an API or suite of microservices.
Microservice gardeners must use innovative build tech-
niques that embrace a distributed architecture, avoiding
the Kafka-esque monolith at all costs.
In this piece, we’re going to discuss what it means to
be a “microservice gardener”, and present six ways API
providers can excel as masters.

Speaker: Eric Wilde

This chapter was inspired by Wilde’s session
at the 2016 Platform Summit. Watch the full
talk here

6 Ways to Become a Master

Microservice Gardener

“Companies are increasingly seeing themselves

as restructuring themselves into something that
has organizational boundaries that have techni-
cal alignments. […] The idea is that you set up
your company in a way that what you’re doing
is also aligned with how you’re doing it. Then
it’s easier for you to restructure and rebuild
yourself when you have to.”

Imagine a garden — a range of flowers, each with their

own needs and purposes. This is the API microservice
landscape. Just as each flower or vegetable needs spe-
cific fertilization and specialized care, it is the role of
6 Ways to Become a Master Microservice Gardener 60

the provider to give attentive care to API microservice

consumers as an API gardner.
As a landscape manager, it is your duty to control and aug-
ment the environment to allow for the greatest growth
and prosperity. Failing to do so does not just hurt the en-
vironment, it could possibly destroy it. Within an API land-
scape, not supporting users where they are can choke
your system and any potential user base you have yet to
develop.
Both of these issues arise from a fundamental shift in
how APIs are produced and managed. Whereas APIs used
to be monolithic and shackled to larger systems, we are
now in an era of microservices, all designed to do specific
things in concert with other solutions, which forces a
containerization and decentralization of the API model.
Now that we know what a Microservice Gardener is, let’s
look at some techniques we can apply to master this art.

1: Use Bimodal IT to Avoid

Stagnation

“Imagine that you have the castle, which is your

legacy system. Now you have these new things
coming around where people say ‘oh we should
do microservices, and it will be fast, and we can
do interesting things.’ They’re highly enthusias-
tic at what they’re doing, but they also need
guidance.”

Bimodal IT is the concept of dealing with disparate styles

of work that nonetheless function perfectly well separate
6 Ways to Become a Master Microservice Gardener 61

from one another. “Bimodal” means two modes, and in

this case, the term matches the function — one focused
directly on predictability, and the other on exploration.
The exploratory type focuses on unique and innovative
solutions that result in a requirement to explore rather
than to depend on known, constant functionality. The
second development track is the inverse, a dependable
codebase that is known, proven, and well documented,
thus resulting in predictable results.
While an API provider could simply direct their API to
only support one of these types of content and the re-
sultant users through the effect of eschewing innovative,
unproven solutions, developers would be excising a large
portion of their potential power and user base by do-
ing so. Bimodal IT approaches are designed to support
both concepts and approaches. By combining predictable
product evolution and documentation with innovation
and exploratory tools, developers can harness the “best
of both worlds.”

2: Avoid the Kafka-esque Monolith

Erik provides a very good example of what a centralized

monolith results in by example of The Castle, a novel by
Franz Kafka. Throughout the events of The Castle, the
protagonist, known only as K., suffers the tribulations
of bureaucracy from a government which erroneously
called him into their village for a job that was non-exis-
tent.
Redirected from contact to contact, ignored by higher-
ups with the simple instruction to talk to their subor-
dinates, and never given explanation for any action of
6 Ways to Become a Master Microservice Gardener 62

the officials, K. exposes the bureaucratic nonsense of

the village surrounding the castle, much to the villagers’
chagrin.

Poorly designed and implemented APIs

(and even the well designed and imple-
mented ones) sometimes result in this
To Wilde,
sort of behavior. The user, attempting to monolithic
use the API for whatever function they architecture is
similar to the
deem, are lost in the structure of subor- nightmarish
dination and redirection. The API design bureaucracy
of Franz
methodology used can somewhat aug- Kafka’s The
ment this, but giving APIs too much struc- Castle
ture could end in a bureaucratic result.
It doesn’t have to be this way, of course. APIs need struc-
ture, but just as important as structure is the need for
guidance. Designing an API to not only get the job done,
but to get it done in an understandable way while al-
lowing for innovation and explorability is a key tenant of
bimodal IT. Essentially, you’re joining the old — that is, the
bureaucratic — with the new.

3: Design In a Way That Promotes

Further Iteration

First and foremost, consider that all of these issues, even

those in The Castle, derive from the principle foundations
upon which the system is built. No single API specification,
documentation, and design approach is perfect, and as
such, avoiding dependence on a single monolithic solu-
tion will do a lot to resolve this.
6 Ways to Become a Master Microservice Gardener 63

When developing an API, keep in mind not only the in-

tended functionality, but the ability to extend that func-
tionality. If an API is designed to perfectly fit only the
current requirements, it will be a perfect API — but only
for a moment in time.
This is much more of a “top level” concept, but it’s one that
needs to be adopted as a mantra for any API provider.
When considering an API, don’t think about what the
user today wants, or even what the user a year from
now will want — future proofing is difficult as technol-
ogy fluctuates so rapidly. Rather, Wilde recommends the
best solution is not to code the functions you expect to
need in the future, but to enable the easy additions of
those features to an extensible code base. Allowing fur-
ther functionality through extensible classes and hooks
to combine functions creates powerful, longevity-driven
solutions.

4: Harvest Concepts and Incorporate

Harvesting and introducing new concepts should be easy.

In The Castle, K. not only runs into confusion, he also en-
counters people actively resisting his attempts to resolve
the confusion. Your API should not do this.
Part of a solution is simply allowing your API to be ex-
plorable. Don’t lock everything down via obfuscation and
layers of haze — explain how an API does what it does,
how this can be extended, and how to call this functional-
ity.
Essentially, it should be easy to identify what an API does
and to use it for new purposes — failing to provide a
system to do this kills any innovation. This has been seen
6 Ways to Become a Master Microservice Gardener 64

time and time again, and is a prime reason open source

has been as successful as it has been. A locked down
ecosystem works perfectly — but only if everything is kept
internal.

5: Distribute the Seeds: Adopt True

Microservices Arrangement

The greatest way to become a master microservice gar-

dener is to move away from centralization. Give teams
of developers more autonomy in how solutions are de-
veloped and implemented, and allow for greater service
discovery on the user side. Essentially, move away from
a centralized API approach and more towards a true
microservice arrangement.
Developers have the tendency to thinks of themselves
as the “central authority”. The developer has made the
code, has restricted it, has documented it, and by this
process, has complete control. While this is fine for single
use APIs or small teams, this is not how an interactive
API platform should function. When a provider functions
this way, they are being despotic — and with despotism
comes inefficiency and bureaucracy.
Instead, approach API development as a benevolent
democratic leader. Allow for more personal use, and do
everything you can to support growth and understanding.
If a gardener cuts too close to the root, they can kill a
flower — if a provider does not provide any amount of
support for anything other than their specific user case,
they’ll do the same to the API.
6 Ways to Become a Master Microservice Gardener 65

6: Prune the Service Surface

When it comes down to it, what truly matters to the

microservice consumer is the surface of the service. The
surface is unique to each offering, but may consist of
varying degrees of public developer portals, API meta-
data, documentations, or other functional summaries
and integration methodologies.
While it’s tempting to take the advice here and reveal
the entirety of your API, that’s a bit too far in the wrong
direction. Consider the simplicity of offering a service
surface and making well-defined concepts that are fun-
damentally shallow in their functionality. By doing this,
you’re creating an extensible solution while mitigating the
potential damage of innovation for innovation’s sake.

Nurture API Ecosystem Growth

“Instead of having a top-down process to de-

sign and document and describe APIs, use more
of an ecosystem-like approach… Try to build
things in such a way that you make the condi-
tions under which good things will happen.”

As providers and developers create API ecosystems, they

need to have a sea change in concept. To do so, API
providers must eschew centralization in favor of innova-
tion and new developments.
The best a developer can hope for is this — create a
system in which the API is understood and extensible,
and from there, create the conditions for success. Even
6 Ways to Become a Master Microservice Gardener 66

if that success does not immediately happen, preparing

should mitigate those possible issues while magnifying
the potential for success.
Is Your API Automotive
Grade?

Imagine that a user has plugged into your API and triggers
a request, only to be met with an error message. Now
imagine all of this is happening while they’re traveling
at 70mph on the highway. For Henrik Segesten, cloud
domain architect at Volvo, making sure that this sort of
thing doesn’t happen isn’t the stuff of bad dreams — it’s
part of his day to day routine.
Segesten and co. can’t afford to make the same mistakes
that API developers working on a plugin for, say the
Nest thermostat can. Rather than cause a minor incon-
venience for the user, the potential impact of flaws in an
automobile scenario could be life-threatening. From the
transmission to the software powering onboard electron-
Is Your API Automotive Grade? 68

ics, automotive grade equates to rigorous testing and

extremely high reliability standards.
But this is about more than just ensuring that automobile
APIs are robust and bug-free – after all, that should be
the aim of any API developer. Rather, it’s about improv-
ing longevity (something we’re writing extensively about
currently) and compatibility.
Right away, we can see that there’s something of a dis-
connect here: it’s extremely difficult, if it’s even possible
at all, to ensure that every step in the API process can
be monitored with the sort of obsessive quality control
demanded by true automotive grade standards. Imagine,
for example, telling API consumers that they’re only al-
lowed to use your APIs if they do all of their coding on
a MacBook Pro, produced no later than 2015, running OS
X 10.10.5 or above.
But, as we’ll see below, there are a number of useful
concepts relating to automotive grade that can make a
lot of sense in the context of how we design and update
APIs as well.

Speaker: Henrik Segesten

This chapter was inspired by Segesten’s ses-
sion at the 2016 Platform Summit. Watch the
full talk here

Test It Like it Has to be Automotive

Grade

We’ve previously written about the importance of testing,

both your APIs themselves and your documentation, yet
Is Your API Automotive Grade? 69

it’s something that some API developers continue to over-

look.
Think about some of the processes that a car, and its
component parts, need to go through before the latest
model rolls off the production line:

• Testing for extreme temperatures

• Analysis of how the vehicle performs on different
terrains
• Adequate warning when something isn’t working
properly
• Lots of redundancy built in

There’s an important lesson from automotive grade in

there for API developers: it’s highly likely that the tests
outlined above are far more thorough than anything you
put your APIs through…unless you work at Volvo with
Segesten, where they’re trying to establish API architec-
ture that can remain solid for decades.
While “bank grade” and “military grade” are often used
in relation to security and construction materials respec-
tively, automotive grade is a term that’s a little more
tightly defined.
Segesten outlines automotive grade as being a hardware
concept with “a measure on the quality level of all parts
in a car…with an expected lifespan for each and every
component, down to every nut and bolt, that is equal to
or greater than the expected lifespan of the car.” That
usually equates to around 30 or 40 years.
In the context of software development, we need to
loosen up that definition a little bit. We aren’t necessarily
trying to create an API that will last for that period of
Is Your API Automotive Grade? 70

time so much as, as Segesten jokes, giving some thought

to “the poor girls and guys who need to re-implement
your API in twenty years’ time.” He highlights the fact that
“automotive grade translates to longevity, so you need
to design with longevity in mind.” In other words, just
because something seems to work for now doesn’t mean
that it’s the right choice.

Automotive Grade and

Futureproofing

Segesten advises looking backwards in order to plan for

the future to consider what the main drivers have been in
web-oriented software and API development throughout
the past 30 years. It’s worth reproducing that list here:

00s: 90s 80s

SOAP HTTP FTP
JSON CORBA SMTP
XML DCE TCP/IP
PGP Telnet
X509 PPP
ONC/RPC

But this doesn’t mean that all API developers should be

shunning RESTful development in favor of standards that
have been around for 30+ years. Rather, it’s an argument
to ensure that your API uses technology that can stand
the test of time. In their lifespan, typical APIs will need
to be “re-implemented several times along the way …
because the old system has died for some reason. Pick
your technology wisely.”
Now, let’s think about the protocols/languages etc. com-
Is Your API Automotive Grade? 71

monly used in the API space in 2017:

• JSON
• REST
• SOAP
• XML
• JavaScript
• Open API/Swagger
• GraphQL

It’s interesting to note that, of the above standards listed

by Segesten, only JSON, SOAP and XML appear across
multiple decades. This doesn’t mean that you should only
consider using these data formats when building an API.
Rather, it’s evidence that ten years isn’t a long time in the
world of software development; developers can ill afford
to use the latest, hippest languages and standards unless
they’re pretty confident that they have what it takes to
stand the test of time. Based on what we’ve seen so far,
REST appears to fit that bill.
But it’s worth remembering, as Segesten says, that there’s
more to an API than just the technology you use; the best
APIs can thrive regardless of the technology that power
them.

Automotive Grade and Backwards

Compatibility

As important as it is to futureproof an API, it’s just as

important to ensure that it’s backwards compatible too.
For example, what would happen if a car buyer disables
all the connectivity features in a car, uses it for ten years,
Is Your API Automotive Grade? 72

then sells it? When the next buyer turns on a connectivity

feature, the car will be consuming the associated APIs for
the first time.
The lesson here is that old versions of APIs never die, and
you’ll inevitably need to maintain them because:

• You can’t be sure that all users will migrate to later

offerings
• Certain products/hardware may not be capable of
communicating with your newest API
• You need to remain compatible with hardware that
will go offline, i.e. asynchronous communications.

Designing APIs with backwards and forwards compatibil-

ity in mind from the start – using extensibility, optionality,
and so on – is much easier than trying to simultaneously
manage tens of different versions … as the folks trying to
support users with versions of Windows that are decades
old will probably tell you. In fact, there is a strong case that
versioning shouldn’t even happen at all with web APIs.
Lastly, Segesten reminds us that it’s a good idea to keep
things as simple as possible. “If you have to have complex-
ity…try to put all of that complexity on the server side, be-
cause you can update that all the time but the API should
never have to be updated.” When that hypothetical ten
year old car we mentioned above finally hooks up to a
Volvo API, it should be more or less good to go if you make
most of your changes server-side.

Final Thoughts

As we’ve seen above, particularly given the lack of control

API developers have over how and where their products
Is Your API Automotive Grade? 73

are used, actually producing an automotive grade API is

extremely difficult. If we take the term at face value, i.e.
remaining valid for 30 or 40 years, it may not be possible
at all.
Any experienced API developer knows that there are too
many variables to control to guarantee that a service will
still be working in 2047. With that in mind, it might be bet-
ter to consider employing the principles of “automotive
grade”, such as:

• Using futureproof, or at least as futureproof as

possible, languages and methodologies
• Have compatibility strategies in place – how do you
version handle your APIs?
• Keep things simple, both in terms of functions and
documentation, and testing as thoroughly as possi-
ble to make sure it all works as expected

The good news is that, for the majority of us, our API
consumers will never be using our products or services
at 70mph. But that doesn’t mean that it’s not a good idea
to prepare like they might be…
Why OAuth 2.0 Is Vital to
IoT Security

The internet is fundamentally an unsafe place. For every

service, every API, there are users who would love nothing
more than to break through the various layers of security
you’ve erected.
This is no small concern, either — in the US alone, security
breaches cost companies in excess of $445 Billion USD
annually. As the Internet of Things (IoT) grows, this
number will only climb.
The problem is our considerations concerning security
are for modern web services and APIs — we rarely, if
ever, talk about the coming wave of small connected and
unconnected IoT devices that will soon make this an even
greater concern.
Why OAuth 2.0 Is Vital to IoT Security 75

From the connected fridge to the smartwatch, the IoT is

encompassing many new web-enabled devices coming to
the market. As we’re designing new API infrastructures,
Jacob Ideskog believes that the “IoT is going to hit us hard
if we’re not doing anything about it.”
Thankfully, there’s a great solution by the name of OAuth.
OAuth 2.0 is one of the most powerful open authoriza-
tion solutions available to API developers today. We’re
going to discuss OAuth 2.0, how it functions, and what
makes it so powerful for protecting the vast Internet of
Things.

Speaker: Jacob Ideskog

This chapter was inspired by Ideskog’s session
at the 2016 Platform Summit. Watch the full
talk here

What is OAuth 2.0?

OAuth 2.0 is a token-based authentication and autho-

rization open standard for internet communications. The
solution, first proposed in 2007 in draft form by various
developers from Twitter and Ma.gnolia, was codified in
the OAuth Core 1.0 final draft in December of that year.
OAuth was officially published as RFC 5849 in 2010, and
since then, all Twitter applications — as well as many
applications throughout the web — have required usage
of OAuth.
OAuth 2.0 is the new framework evolution that was first
published as RFC 6749 alongside a Bearer Token Usage
definition in RFC 6750.
Why OAuth 2.0 Is Vital to IoT Security 76

What Does OAuth Do?

While by definition OAuth is an open authentication and

authorization standard, OAuth by itself does not pro-
vide any protocol for authentication. Instead, it simply
provides a framework for authentication decisions and
mechanisms.

“OAuth does nothing for authentication. So in

order to solve this for the web, we need to
add some sort of authentication server into the
picture.”

That being said, it does natively function as an autho-

rization protocol, or to be more precise, as a delegation
protocol. Consider OAuth’s four actors to understand
how it accomplishes this:

• Resource Owner (RO): The Resource Owner is the

entity that controls the data being exposed by the
API, and is, as the name suggests, the designated
owner.
• Authorization Server (AS): The Security Token Ser-
vice (STS) that issues, controls, and revokes tokens
in the OAuth system. Also called the OAuth Server.
• Client: The application, web site, or other system
that requests data on behalf of the resource owner.
• Resource Server (RS): The service that exposes and
stores/sends the data; the RS is typically the API.

OAuth provides delegated access to resources in the

following way. Below is a fundamental flow in OAuth 2.0
known as implicit flow:
Why OAuth 2.0 Is Vital to IoT Security 77

• The Client requests access to a resource. It does this

by contacting the Authorization Server.
• The Authorization Server responds to this request
with a return request for data, namely the username
and password.
• The Authorization Server passes this data through to
an Authentication solution, which then responds to
the Authorization Server with either an approval or
denial.
• With an approval, the Authorization Server allows
the Client to access the Resource Server.

Of note is that OAuth 2.0 supports a variety of token types.

WS-Security tokens, JWT tokens, legacy tokens, custom
tokens, and more can be configured and implemented
across an OAuth 2.0 implementation.

Unique IoT Traits that Affect

Security

Now that we understand OAuth 2.0 and the basic work-

flow, what does it mean for securing the Internet of
Things (IoT)? Well, the IoT has a few unique caveats that
need to be considered. Ideskog notes that IoT devices are
typically:

• Battery powered: IoT devices are often small and

serve a particular function, unlike server resources
which have massive calculation-driven platforms and
consistent, sanitized power flow.
• Asynchronous: They are partially or completely of-
fline, connecting only asynchronously via hub de-
vices or when required for functionality.
Why OAuth 2.0 Is Vital to IoT Security 78

• Lean: Lastly, IoT devices usually have limited calcula-

tion capabilities, and depend on central devices and
servers for this processing functionality.

Despite all of these caveats, IoT devices, are extremely

attractive targets to attackers due to their known sin-
gle use functions and relatively lax security.

Proof of Possession

Due to all of these caveats, the OAuth workflow is strik-

ingly different — we are, in fact, using a methodology
called Proof of Possession. Consider a healthcare sce-
nario, in which a doctor must access an EKG IoT device.
Since the IoT device cannot perform the same authenti-
cation process as a full client device can, we need to do a
bit of redirection.
The start is normal. The Client sends an access request
to the Authorization Server. From here, the Authoriza-
tion Server contacts the Authentication Server, which
prompts the Client with Authentication Data. When this is
provided, the Authentication Server authenticates to the
Authorization Server, which issues an authorization code
to the Client:

authorization_code = XYZ

From here, we deviate from the standard OAuth work-

flow. The Authorization code is a one-time use proof that
the user is Authenticated, and this code can be used to
further contact that IoT device as an authorized device.
The code is not something that can be used to directly
Why OAuth 2.0 Is Vital to IoT Security 79

access data as other OAuth tokens are, it is simply proof

that we are who we say we are and that we’ve been
authenticated and authorized.
The Client then generates a key (though this key can
also be generated server side) to begin the connection
process with the IoT device, sending a packet of data that
looks somewhat like this:

Client_id = device123
Client_secret = supersecret
Scope = read_ekg
Audience = ekg_device_ABC
authorization _code = XYZ
…
Key = a_shortlived_key

With the data in hand, the Authorization Server now

responds to this packet by providing an access_token; a
reference to data held in the Authorization Server mem-
ory to serve as proof of possession of both authentication
and authorization:

Access_token = oddfbmd-dnndjv…

This is the final step — the client is now fully and truly
authenticated. With this acess_token, the client can start
a session on the IoT device. The IoT device will look at
this access_token, and pass it to the Authorization Server (if
it’s a connected device) asking for verification to trust the
device. When the Authorization Server accepts the verifi-
cation, it passes a new key to the IoT device, which then
returns it to the Client, establishing a trusted connection.
Why OAuth 2.0 Is Vital to IoT Security 80

Disconnected Flow

What happens if a device is unable to ask the Authoriza-

tion Server for verification due to power or calculation lim-
itations? In this case we can use something called Discon-
nected Flow. A key point for Disconnected Flow is unlike
other OAuth 2.0 solutions, this eschews TLS (Transport
Layer Security) by nature of the Resource Server being
a disconnected device with intermittent connectivity and
limited communication and processing power.
In this case, we’re actually shifting the parties around
somewhat. The EKG machine is now the client, and an-
other IoT device, a test tube, is the Resource Server. First,
the EKG machine authenticates and authorizes in the
same way as before:

Client_id = ekg_device_ABC
Client_secret = supersecret
Scope = read_result
Audience = connected_tbie_123
Token = original_token
...…
Key = a_shortlived_key

Once this is received by the Authorization Server, the

server replies not with the access token in the former
structure, but instead an access_token in JWT (or JSON
Web Token). This token is a by-value token, meaning it
contains the data fed to it and the response. Whereas
our first string referenced a memory location in the Au-
thorization Server, the JWT has all of the data in a single
key string.
From here, the JWT can be converted into other formats
for easier reading by the test tube. By design, the test
Why OAuth 2.0 Is Vital to IoT Security 81

tube is crafted to trust the Authorization Server in a

process called Pre-provisioning. Because of this, when
we send the Client token in JWT (or whatever format that’s
been chosen), the tube implicitly trusts the key as long as
it originated from the Authorization Server, and begins a
connected session with the Client.

Ideskog notes there would technically be 2 to-

ken types involved in the flow above: a signed
JWT would contain an encrypted token (JWE),
which has a key in it that is later used for the
communication channel. The JWS (commonly
called JWT) isn’t necessarily encrypted, and is
usually in plain text and signed.

Real Worth Authorization Failure

To see exactly why this is all so important, consider some

real world authorization failures. One of the most
visible failures is known as The Snappening, a leak of over
90,000 private photos and 9,000 private videos from the
Snapchat application.
Most of the blame for the Snappening came from users
using unauthorized third party applications to save
Snaps. These third party applications did not utilize OAuth
solutions, meaning when remote access users attempted
to use the undocumented Snapchat URL that the third
party application relied on, they were able to spoof as
authorized users and retrieve content without proper
token assignment.
This a great example of OAuth 2.0 vs. no implementation,
as we have essentially a “control” application (Snapchat
Why OAuth 2.0 Is Vital to IoT Security 82

secured by OAuth) and a “test” application (the unautho-

rized applications tying into the undocumented API). With
improper authorization integration, content was allowed
to leak through an insecure system with relative ease.
Had the third party application properly implemented an
authorization scheme, this would never have happened.
This issue is only relevant to non-IoT things, though — it’s
just a photo sharing application, right? Wrong. Consider
now this same fault of security for something like an IoT
button that triggers replacement business items. Attack-
ing this device can result in man-in-the-middle attacks to
capture addresses of order processing servers and even
spoof orders to the tune of thousands or hundreds of
thousands of dollars.

OAuth Embeds Trust into the IoT

Applying OAuth to the IoT makes it truly extensible and

customizable. Using OAuth, we can build systems based
on trust that use fundamentally secure resources and
communications protocols. OAuth is, by design, all about
trust.
This trust is key to securing the IoT. For connected devices,
Proof of Possession can solve most security issues. For
constrained environments by either connectivity or pro-
cessing calculative power, devices can be secured using
pre-provisioning that is independent of TLS, and does not
require the device to be online at all times.
It should be noted that JWT, JWS, and JWE are all helpful
tools, but all work with JSON. For lower processing envi-
ronments, sibling binary tokens such as CWT, CWS, and
Why OAuth 2.0 Is Vital to IoT Security 83

CWE can be used as they cater well to building on low

power and limited scope devices.

Conclusion

This isn’t a game — though having lax security can be

convenient for innovation and experimentation, when
it comes to the IoT, this is a dangerous approach. IoT
devices might be underpowered and single use, but they,
as a network, are powerful.
Remember that a network is only ever as secure as the
sum of its parts and the weakest point of entry to its
ecosystem. Failing to secure one IoT device and adopting
a security system based on inherited security can result
in a single IoT device comprising every device connected
to it.
OAuth 2.0 can go a long way towards solving these issues.
4 Design Tweaks to
Improve API Operations

We’ve previously discussed best practices when it comes

to designing an API with quality developer experience.
But what will the long term operational repercussions
be for the design moves we make now?
For example, if URLs are designed without metadata to
describe actions, later on, product owners will have a
difficult time staring at unintelligible logs. Or, if microser-
vices aren’t orchestrated correctly, you run the risk of
long load times queuing multiple API calls in a mobile
environment. These are only two of the many operational
consequences that many API owners overlook while de-
signing their APIs.
Today, we’ll consider some methods to make web Applica-
4 Design Tweaks to Improve API Operations 85

tion Programming Interfaces more operationally efficient

and responsive to the way that clients will consume them.
Led by Nordic APIs veteran Jason Harmon, we’ll cover
the most common operational anti-patterns that could
break an API. We’ll offer ways to remedy these poor
design situations, that if not addressed, could cause some
serious migraines later on in the API lifecycle.

Speaker: Jason Harmon

This chapter was inspired by Harmon’s ses-
sion at the 2016 Platform Summit. Watch the
full talk here

From simple, yet often overlooked issues, such as using

the appropriate HTTP method, to more complex compo-
sition advice, let’s consider four anti-patterns found in
the wild all too often.

“To some extent, the way you design your API

can set you up for failure.”

1: HTTP GET instead of POST

API providers commonly implement an HTTP GET call where

a POST should have been used instead. A reminder to use
the correct HTTP method is by far the most basic piece of
advice here, yet surprisingly, it’s a common error that can
cause some major damage.
Harmon related a story from his experience at Typeform,
where a small subset of users in web browsers were
hitting the “back” button, causing their sessions to lose all
4 Design Tweaks to Improve API Operations 86

data. By changing the call type from GET to POST, Typeform

was quickly able to solve their caching issue, as the HTTP
RFC states, POST is not allowed to be cached by anything.
The lesson here is to watch out for cached calls from
browsers or proxies. If you do encounter unexpected be-
havior, Harmon recommends to first “look for the GETs…
using POST instead is an easy fix.” If you’re stuck in a
situation with erratic cache issues, he adds that adding
an extra query string with a randomly generated cache-
buster to the GET call (i.e. ?cache_buster=[random]) could
solve a recurring issue.

2: Letting clients constantly poll APIs

Many API providers should reconsider how they allow

clients to update data. Too often, the client sets up con-
stant polling on API endpoints. When a large dataset is
involved, and queries are occurring continually, such as
every 30 seconds, the number of calls can really add up.
Large volumes of calls to an AWS server can be expensive
as well.
Typeform is not immune to the API polling issue. Type-
form end users want to know if there is updated data on
their form, and therefore set up a constant polling service
to see if the data has been updated. Since Typeform is
among the Zapier compatible apps, meaning that users
can create customizable “zaps” that tie in Typeform func-
tionalities, the number of services continually requesting
new data skyrockets.
To avoid constant polling, Harmon recommends you build
and launch webhooks for your service, and convince
your consumers to use them. But first, find out what
4 Design Tweaks to Improve API Operations 87

the rate of change for your data is. For example, The
average rate of change for Typeforms are typically 1-2
weeks. Identifying a pattern for the rate of change in your
data can help you design your webhooks to be as lean as
possible.
Harmon recognizes one catch to this approach — he rec-
ommends still using a polling API alongside the webhook
payload to carry out sporadic system checks and perform
large downloads.

3: Rigid hierarchy in microservices

causes latency

Lately, many SaaS providers have been transitioning to

the microservices architectural style. Microservices are
lean, well-bounded components dedicated to processing
a specialized functionality. This is great for structural
segmentation, but if microservice communication is not
orchestrated well, requesting data can cause serious la-
tency.

“Build your microservices to be externalized

from the ground up”

The problem with microservices is that a client requesting

a large number of functions could end up sending dozens
of separate calls, leading to a massive query load. This
is especially problematic in mobile environments, where
calls must be queued in series instead of being executed
in parallel to one another. For some environments, this
process is simply too slow.
4 Design Tweaks to Improve API Operations 88

According to Harmon, the solution lies in a BFF. No, not

a best friend forever, but a Backend-for-Frontend that
acts as a shim to help compose microservices.
A BFF is a lightweight layer that acts as an orchestration
API. It could be built as a Node.js service, or however
internal developers see fit. The goal of such a shim is
to, with one call, have the client receive all the resources
packaged together in the way the client wants. That way,
to put it in Harmon’s words:

“They’re not gluing together a model in the

browser, the model’s already given to them the
way the wanted it in the first place.”

insert picture of BFF[A BFF has been implemented to

solve microservice design issues at both Paypal and Type-
form. By decreasing JSON packages, a BFF can cut down
JavaScript processing.]
Allowing clients to call a microservice directly without any
composition layer is poor design as it doesn’t consider the
use cases and client limitations. Constructing a BFF layer
is a possible solution, but it should be noted that GraphQL
is another potential solution as well.

4: Generic actions

When we design URLs, Harmon reminds us that detailing

actions matters. He recommends to note state transi-
tions within the URL, and to pass along short descriptions
as metadata. If you don’t describe state transitions out-
side of the typical CRUD verbs, you run the risk of having
very generic, unreadable logs.
4 Design Tweaks to Improve API Operations 89

This problem is often called protocol tunneling; with

generic URL names you often lose perspective, and then
when an action breaks, it’s hard to locate the affected
calls and analyze trends. If we design URLs as generic
phrases that tell us nothing about the entire story, then
error diagnosis will be difficult.

“URLs are a key component in how you opera-

tionalize the API”

Harmon notes this is especially important for product

owners using tools such as the ELK stack to visualize data
sources to determine insights. When designing URLs, vis-
ibility is a plus, so consider a good method of identifying
these URLs with naming schematic, along with metadata
for the actions that are being taken.

/resource/:id/generic-name + {action:process}

Harmon-ious Mantras to Live By

We’ve covered much ground with some specific issues to

avoid. For more generic design advice, the wise Harmon
provides some mantras to design by:

• Use cases first, then design: Ensure your HTTP

methods are correctly ascribed, and that user behav-
ior won’t adversely affect the outcome.
• Design can influence performance: Put an end to
polling. Rather, design with subscription webhooks
and insist that clients use them instead.
4 Design Tweaks to Improve API Operations 90

• Structure is good, but be prepared to blur those

lines: Though microservices architecture is about
separate division, having too rigid of a structure can
result in “unhappy clients and crappy performance.”
• Design can put out fires: Having a DevOps ap-
proach to design early on can put out fires in the long
term. Try to make performance more visible.
• It’s all about the logs: Ultimately, your logs are the
transcript of what clients are actually doing. Con-
sider how you can make this performance more
intelligible by crafting URLs and metadata that will
aid metric analysis and make product owners happy.

Developer experience is the mainstay of most API design

discussion, but we mustn’t ignore how design moves will
affect API operations as well. As Harmon points out:

“API design is not just fanciful usability discus-

sions and lot of fluffy emotions about develop-
ers… Developer Experience really is just the first
layer.”

Following these guidelines, we can improve specific areas

to help design on the scale of decades, as operational
efficiency is intimately correlated with platform longevity.
By planning for operational improvements now, we can
“design in a way that writes sentences, to tell a bigger
story later on.”

Speaker: Jason Harmon

Also watch Nordic APIs veteran Jason Harmon
present his tips on scaling API design. Watch
the full talk here
The API of Me

Here’s an interesting fact: If you live in the EU your per-

sonal data is yours. You are the owner of your data, a fact
enshrined in law under the General Data Protection Reg-
ulations (GDPR). As owners of data and citizens of many
economies — internet, application, information, API — we
have a myriad of tools and technologies available to mine,
mash up, and generally manage our data as we see fit.
Moreover, we can exploit our data for our own benefit,
selling it to the highest bidder for our own profit.
However, as consumers we don’t often consider how we
can exploit our data. Typically we are only concerned with
data security and are not as conscious of using it for
our ends. Perhaps this is because the banks, insurance
companies, and other corporations that hold our most
valuable data are not naturally disposed to making it
easily accessed or shared with whomever we see fit.
Enabling the sharing of our personal data in a manner
controlled by us is at the heart of some unique concepts.
This would allow us to create a Personal Data Store
to better visualize personal data, and would also enable
the API of Me, the endpoint by which we can share our
valuable personal data in a digital world.
The data revealed by the API of Me is not a single plane;
there are different types of personal data from a variety
The API of Me 92

of sources that make up our dataset and describe who

we are, what we do, and what we touch. They come with
varying connotations and likelihoods of being exposed
to an external audience. In this chapter, we’ll explore
these types of personal data that if unlocked could finally
enable the API of Me — the programmable endpoint to
our digital lives.

Speaker: Chris Wood

This chapter was inspired by Wood’s session
at the 2016 Platform Summit. Watch the full
talk here

Confidential

Confidential data is the data most important to us. These

attributes uniquely describe us and allow important things
to happen in our life, from our birth to our death. It also
includes subsets of information that we would prefer to
be extremely closely guarded; data about our health and
ongoing health records, for example.
It goes without saying that confidential data is extremely
sensitive, and when obtained by miscreants can be used
to impersonate us and falsely make applications for bank
accounts or mortgages without our consent. Most coun-
tries and regions have established extensive laws and
regulations, like GDPR, to safeguard confidential data.
Though safeguarding this data is crucial, selectively mak-
ing it available can aid our daily lives by simplifying the
process of data exchange.
The API of Me 93

In general, due to data privacy constraints, organizations

that hold our confidential data limit it to a specific, autho-
rized realm: The owner of data themselves, or a third-
party who is expressly authorized to access this data. The
majority of APIs that access this information are therefore
either private or partner APIs, with little or no public
access. Enrollment, authorization and access are tightly
controlled. Such private APIs include the closed APIs that
power our online banking platforms, or partner APIs that
are the heart of price comparison websites for insurance
or other products.
This closed approach to confidential data introduces con-
siderable friction to us as consumers, and requires us
to endlessly confirm our identity using physical artifacts
like a driver’s license, passport, or proof of address. Only
where a formal scheme exists for sharing our identity
and confidential data digitally (such as the schemes in
the Nordics like NemID and BankID) does that friction
disappear. The technology also exists to extend the scope
of the platforms that hold this information and to share
this data on our behalf; both in terms of protocols or
products like Trunomi. It remains to be seen whether we
as consumers have the appetite and trust to allow this
information to be shared via an API to unlock it for our
own benefit.

Financial

Our financial data consists of the transactions we make,

our banking history, and creditworthiness. Like confiden-
tial data, it is of considerable value to us as consumers.
Exposing this data to applications like personal financial
The API of Me 94

management (PFM) tools can provide helpful visualiza-

tions that the atomic data doesn’t necessarily deliver.
Our financial data also has intrinsic value to businesses
who attempt to sell us products and services based on
what we spend our money on. Such data is essential
to market analysis and demographic trends. Obtaining
such data is easier for major retailers, especially those
with loyalty schemes who can easily correlate spending
over time with a real individual.
Financial data and APIs presents an interesting juxtaposi-
tion for consumers: Having this data available to PFMs or
accounting packages like Xero that help us make sense
of our spending is extremely valuable, but often only
private APIs (such as online banking) expose it. Gaining
access to the data therefore becomes a question of using
workarounds such as screen-scraping, which can contra-
vene the terms of use for the services we access. More-
over, this data is also not available to businesses who
might sell us a product or service we are truly interested
in. This can negatively affect our perception as consumers
of businesses who resort to spamming us based on either
spurious correlation deduced from limited data or no
analysis at all. Controlling how we expose financial data
via APIs is therefore highly valuable to us, but again we
are frequently thwarted by those who hold our data on
our behalf.

Tactile

It’s more difficult for consumers to comprehend the value

of tactile data: It’s what we ‘touch’, both in the physical
and virtual world. Tactile data includes:
The API of Me 95

• The sensors we activate when travelling on public

transport, using either a travel or contactless card or
a mobile or smart device (which may indeed present
a cross-section with our financial data)
• Our GPS movements as tracked by our smartphones;
• The items we browse at a clothes store that are
labeled using smart tags;
• Our browsing history that shows what we view on-
line;
• Qualified data we catalogue about ourselves, such
as calorie information collected through apps like
MyFitnessPal.

These events tie us to particular locations and activities

and provide a clearer view of us as an individual. If we
chose to share tactile data with the organizations we
want to do business with, we could exploit unique offers
specifically tailored to our daily activities, which could
bring about fantastic consumer experiences.
At the time of writing tactile data is only sporadically avail-
able to us as consumers and indeed subsets of that data
are tightly controlled in most countries, such as the geo-
graphic tracking of an individual. There are exceptions in
certain subject areas of course, like the Garmin Connect
API (although being able to access this data carries a con-
siderable price tag!). Browsing data is also becoming the
subject of intense scrutiny with more tools being created
to protect it. Few, however, are allowing consumers to
exploit it for their own purposes. As consumer awareness
of tactile data increases we will value it more and thus
our desire to exploit it will become greater. However,
exploiting it and making it available via APIs, given the
regulations and laws involved, may be extremely difficult
for the foreseeable future.
The API of Me 96

Aggregate

Aggregating our confidential, financial, and tactile data

helps to build a picture of our lives that is extremely
valuable to us: It could be a way to exploit the Intention
Economy, a concept described by Doc Searls that he later
expanded upon on his book on the topic. The Intention
Economy focuses on the fact that consumers come ready
made, and that advertising is unnecessary: As buyers
of goods know what they want, they can make rational
decisions in approaching the market to make a purchase,
allowing sellers to “bid” for them. This is incongruent with
the vast majority of markets today, as they are seller-
orientated and based largely on advertising.
The Intention Economy has yet to come to fruition in
a universal sense, but pockets of behaviors that loosely
follow the idea have emerged. Examples include price-
comparison websites for insurance, energy supplies or
credit cards, but these only bid in silos i.e. the companies
offering deals don’t overtly attempt to outperform each
other. The concept and implementation of Personal Data
Stores is becoming a reality through software like Meeco
and Mydex, enabling us to corral, organize, and allow
specific access to our data.
We are on the cusp of being conscious of our real value
as consumers. However, the architecture, protocols, and
networks to support the ready aggregation and inter-
change of our personal data is in a nascent state, with
some of the key tools like vendor relationship manage-
ment (VRM) software (a means to manage our relation-
ship with the vendors we wish to do business with) under
relatively early development. The API-of-Me concept will
need to mature for us to truly elicit the value of our
The API of Me 97

personal data.

Final Thoughts

The ownership and stewardship of consumer data is a

subject much wider than the API economy: It covers all
areas of technology, and for the majority is a subject that
only surfaces in conscious thought when a business or
organization uses their data in a way that they should not.
However, as the population becomes more tech-savvy,
consumers will begin to understand that their data (whether
exposed by an API or another means) has intrinsic value
to them and controlling it can deliver them significant
benefits. Generally speaking, we are at a standoff be-
tween the desire to access this data for our own means,
the willingness of the organizations that harbor our data
to make it available, and the manner in which we can
correlate it, understand it, and disseminate it for our own
benefit.
The cross-section of this data forms a picture of us as
individuals that is hugely interesting to potential con-
sumers of this data. That value must be controlled and
safeguarded for our own benefit.
We are on the verge of utilizing the data we create for
completely new ventures — both the API of Me and the
APIs that allow us to access our data will certainly provide
a vehicle to help this opportunity become a reality.
Avoid Walking on
Eggshells and Use
DevOps

Humpty Dumpty sat on a wall, Humpty Dumpty

had a great fall. All the king’s horses and all the
king’s men Couldn’t put Humpty together again.

That sure sounds like the last time your site went down
and every team in the office started panicking, right?
When things go wrong, everybody starts pointing fingers.
“The developers are saying it’s a server issue, IT is saying
Avoid Walking on Eggshells and Use DevOps 99

it’s a code issue. It’s absolute chaos,” says Emily Dowdle,

of Wazee Digital.
The real truth? No-one is quite sure what’s wrong. Oth-
erwise, they’d already be doing what’s needed to fix the
problem and get the site back online. In many cases, the
problem is miscommunication: person A knows some-
thing that person B doesn’t, and that’s preventing person
B from coming up with a fix.
The idea behind DevOps is that facilitating interactions be-
tween teams, particularly development, operations and
QA, can alleviate some of that chaos.

Speaker: Emily Dowdle

This chapter was inspired by Dowdle’s session
at the 2016 Platform Summit. Watch the full
talk here

Silos Are Bad For Business

There’s a problem with the modern workplace that not

many people talk about. Namely, that everybody works
in silos. And, where that’s the case, no amount of team-
building exercises or open office space will change things.
Let’s talk about developers and operations: they don’t al-
ways get along. Emily Dowdle, of Wazee Digital, describes
the situation like this:

“There’s tension, could be described as a fric-

tion or an attitude … a general inability to tol-
erate each other without eyerolls and audible
sighs.”
Avoid Walking on Eggshells and Use DevOps 100

She talks about how the Operations side of things is

evaluated not by the features they release, as developers
are, but by quantifiable measurables like site uptime. Ops
teams strive for five nines (or better) uptime, i.e. a site
being up for 99.999% of the time, which works out to 5
minutes a year of downtime.

“Fundamentally, we have different priorities. It

means we have conflict, we butt heads. It’s no
wonder we’re natural enemies.”

But it doesn’t have to be this way.

Key DevOps Concepts

With the rise of agile development – which, of course,

results in more frequent releases – embracing DevOps as
a concept is not just advantageous, but a necessity.
But – like “big data”, “agile development” and other mots
du jour – DevOps is such a large, often poorly defined,
concept that it’s easy to get lost in.
Dowdle has a few ideas for practical implementation of
DevOps in the workplace:

1. Bridge the skills gap

Acknowledging that development and operations teams

don’t work in the same way, and don’t have the same
priorities, is key to making changes.
Avoid Walking on Eggshells and Use DevOps 101

No developer is ever going to fully understand the ins and

outs of operations, and vice versa, but some comprehen-
sion of why the other team would want to take a certain
approach is extremely valuable.

2. Share information

Everyone hates meetings, but they do have their uses.

If a new feature is in the works, it’s important for develop-
ment and operations to discuss the safest way to iterate
without breaking… well, everything.
From the other side, it’s helpful for developers to have
access to read-only logs that let them see the impact of
their deployments and any other changes they’ve made
to a site. Adding ChatOps, colourfully named so by GitHub,
into the mix isn’t a bad idea either.

3. Consistency

Whether we’re talking about the process of deployments

themselves – which should be such an intuitive process
that anyone on the team can do it – or how closely a
staging area replicates its live equivalent, consistency is
key.
Making small, incremental changes – no more bloated
feature releases – is the best way to make sure everything
stays recognizable while still moving forward with devel-
opment timetables.
Continuous integration/delivery is, as we’ve written be-
fore, here to stay. A cornerstone of agile development,
it’s also incredibly useful for trying to keep things as
consistent as possible.
Avoid Walking on Eggshells and Use DevOps 102

4. Increased accountability

In most workplaces, teams are guilty of passing the buck

sometimes.
Sales and marketing teams might close deals on projects
that don’t technically exist in the product yet (but the devs
can take care of all that!), while developers might make a
deployment that really needed more QA in order to meet
a deadline.
Including devs in on-call rotation, i.e. involving them in
that 4am “the site is down” conversation, is one example
of a way to change the way developers think about de-
ployments.
Again, this is as much a culture thing as it is a process
thing: in her presentation, Dowdle talks about the fear of
heads rolling when something goes wrong, and that fear
makes people want to divert responsibility.
Making sure everyone is up to speed with best practices
(e.g. security) in the workplace is key, but it’s just as vital
to make sure that employees feel able to do their job
without worrying that one mistake will cost them it.

5. Embrace failure

Let’s think about those mistakes a little more. So the site

went down? That’s bad. But it doesn’t need to be the end
of the world.
It’s more important to figure out what went wrong, which
automated alerts can seriously help with, and figuring
out how to make sure it doesn’t happen again. Assigning
personal blame isn’t helpful, but enabling a team to see
where they went wrong definitely is.
Avoid Walking on Eggshells and Use DevOps 103

Expecting and preparing for mistakes or accidents, and

not setting unrealistic uptime targets of 99.99999%, can
help to ease some of the tension between development
and operations.
If (or maybe that should be when) you perform a post-
mortem on what went wrong, figuring out why and whose
fault it was — without a lot of pointing fingers – is the best
way to make sure that it doesn’t happen again.

DevOps and APIs

Some of the problems outlined above become even more

significant when APIs are thrown into the mix: According
to Dowdle, in most cases of API problems, there will be at
least two development teams and two production teams
in the mix.
In that respect, APIs actually offer a great jumping off
point for an approach that takes DevOps more seriously;
when it comes to creating a great API, creators are keen
to avoid issues because they know that multiple teams
will inevitably be involved.
That’s especially true if we’re talking about a public API, in
which operations and testing are even more critical than
one designed for internal use.
As a result, the following are key in successful API devel-
opment:

• A very specific approach to requirements e.g. param-

eters, response codes etc.
• Thorough documentation to walk through a number
of potential problems
Avoid Walking on Eggshells and Use DevOps 104

• Need for rigid testing frameworks that functions

exactly like what’s live

What API developers don’t necessarily realize is that ev-

erything they’ve applied to the development of their API is
the perfect recipe for a balanced DevOps-style approach
to building a core software offering too.
So, next time you’re working with your team on a new
feature, ask yourself the following question: “is this how
we’d be doing it if we were making changes to our API?” If
the answer is no, then you might find the concepts above
useful to consider.
Dowdle doesn’t come right out and say it, but her strong
focus on documentation, interdisciplinary communica-
tion and parity between environments means that it’s dif-
ficult not to reach this conclusion: the way we build APIs
is an effect a model for the entire software ecosystem.

Guides to API Development

Management

API developers aren’t always given the same freedoms as

traditional software developers. Or perhaps we should
phrase that another way: API development is typically
directed by feedback and requirements, or a product
backlog.
In some – but not all! – workplaces, developers make
additions based on what marketing/sales asks them to
add, what features THEY think a product should have
and sometimes (let’s be honest here) what’s trending that
week on HackerNews or Reddit.
Avoid Walking on Eggshells and Use DevOps 105

Let’s consider the following three diagrams for slightly

different development models:

Accenture DevOps diagram

IBM continuous integration process

Microsoft development management cycle

The first of the above diagrams is a DevOps approach

to building and testing API features outlined by Brajesh
De, of Accenture, and the latter two show continuous
integration/delivery/feedback DevOps approaches used
elsewhere.
You’ll notice that, while the order and the specifics may be
a little different, the overarching ideas are basically iden-
tical – requirements are designed, deployed, tested and
feedback is gathered, combining both the development
and operation sides in synchronicity.
The fact that the Accenture API management framework
is so similar to other DevOps templates only seems to
further the idea that the approach adopted during API
development – which typically involves rigorous testing
and continuous feedback from Operations and/or those
who use the API – effectively offers a valuable DevOps
model.
Avoid Walking on Eggshells and Use DevOps 106

Conclusion: Putting Humpty Back

Together Again

The bottom line is that much of this comes down to

culture: a DevOps approach, or one that mimics the style
of API development, won’t be successful if your Develop-
ment and Operations teams are completely unwilling to
co-operate.
The DevOps mindset isn’t a new one but it’s one that,
while hugely valuable, can still be difficult to adopt be-
cause of problems with coordination or company struc-
ture. Still, it’s one that’s increasingly important as agile de-
velopment and continuous delivery/feedback becomes
the norm.
Adopting an approach that works is even more critical
when it comes to APIs because, when API development is
done badly, it’s not just your own business that’s at risk;
you’re also compromising the businesses of everyone
who uses your API.
And if there’s one thing harder than putting Humpty back
together again, it’s putting a thousand Humpties back
together again with an angry mob shouting at you the
whole time.
Securing the IoT for
Decades to Come

In 2007 Kevin Kelly gave a TED talk in which he forecasted

how the World Wide Web would look 5000 days into
the future, prophesizing the emergence of the IoT and
AI. He envisioned a more connected planet where all
manufactured goods tap into a single, global, intelligent
network.
At the time, the Internet of Things (IoT) was only a nascent
idea thrown around by futurists. Now, however, it’s a hot
potential market, accelerating large corporate moves and
entrepreneurial initiatives throughout the world. Gartner
predicts there will be over 20 billion IoT devices by 2020.
So, let’s model Kelly’s thought process and take the long
view yet again. What will the next 5,000 days bring? What
Securing the IoT for Decades to Come 108

will the world look like in 2030? Drone deliveries, self-

driving cars, and more innovation is surely coming, but
so are unprecedented vulnerabilities to the user. In 2030,
the way IoT sensors and devices share data must be
secured, or else we are creating our own doomsday.
According to Travis Spencer, founder of Twobo Technolo-
gies, the IoT can only exist if identity checks are properly
implemented at the API level — as any IoT device must
know who is requesting data, and what they are allowed
to do with that data.
The creation of a global device-oriented network is only
possible by using APIs for standard information design,
as well as adopting internally recognized open standards
(like OAuth, OpenID Connect, SCIM, etc.) as a foundation
for the identity management that is so critical for pro-
tecting user safety and data.
We’ve discussed at length the need for standardized APIs
within the IoT, but scaling IoT platforms from a security
perspective is another beast entirely. In this chapter, we’ll
look decades into the future to see how IoT and API
security must unite, so that developers can begin to scale
their platform and security measures accordingly.

Speaker: Travis Spencer

This chapter was inspired by Spencer’s ses-
sion at the 2016 Platform Summit. Watch the
full talk here
Securing the IoT for Decades to Come 109

Looking into the Crystal Ball: The

World in 2030

The world is changing rapidly, and IoT is gaining true

momentum. By 2030, we’ll see the ubiquity of driverless
cars and drones. Highways are already being designed for
driverless cars, drones have expanded beyond hobbyist
usage into numerous applications, and sensors are driv-
ing the fog computing movement.
As the IoT leaves the realm of hobbyism and integrates
into our daily lives, Spencer sees connected apparel, space
tourism, 3D printing, smarter healthcare, and secure elec-
tronic voting as realistic implementations.

“The IoT is not hype. It’s everywhere — it’s in the

room.”

There will also be major restructuring within our industry

and workforce. Entire companies will vanish and new
ones will emerge. Futurist Thomas Frey predicts that by
2030, 2 billion jobs will disappear from the planet.
But this transformation will also reinvigorate the work-
force; It’s been said that by 2030, 1 in 5 jobs will require
programming as a basic skill, and the need for program-
mers is growing 12% faster than other areas.
Today’s tech will appear absolutely archaic in comparison
to future devices, yet the software architecture could
remain constant if open standards are adopted. Let’s see
what IoT evolution will mean for scaling security architec-
ture for decades to come…
Securing the IoT for Decades to Come 110

Identity is the #1 Impediment to

Safe IoT Connections

As we embed devices into our bodies and immediate

surroundings, Spencer sees political, religious, and social
systems being reinvented and questioned. But some-
thing that won’t ever change is the human desire to form
relationships. People will always want to know who or
what it is that they are interacting with.

“Within social, big data, and all different aspects

of computing impacting our lives, we need to
know who we are talking to — who’s on the
other end of that API?”

Without identity control in place, trusting an IoT gadget

poses a serious threat. Consider a skier using smart
goggles that describe upcoming conditions on the slope.
If someone hacked this platform and edited the terrain,
this could lead to a fatal crash. To prevent this, we need
to intimately understand who is accessing IoT devices.
Whereas humans intimately understand relationships, all
a computer understands are 0s and 1s. So how do we
train IoT devices to recognize identity? Spencer sees the
solution lies in incorporating an open standards-based
identity management system within your API manage-
ment makeup, using specialized systems to authorize and
federate access across API connections.
Securing the IoT for Decades to Come 111

Building on Open Standards will

Secure Future Identity in the IoT

The open standards we’re talking about are communi-

cation specifications defined and generally agreed upon
by the international web development community. Ac-
cording to Spencer, deploying these open, internationally
recognized standards will enable long-lasting IoT/API plat-
forms to be constructed.
The Neo-security Stack achieves this, and is built on the
following open standards:

• Authentication: U2F & Web Crypto

• Provisioning: SCIM
• Identities: JSON Identity Suite
• Federation: OpenID Connect
• Delegated Access: OAuth 2.0
• Authorization: ALFA

Within the Neo-security stack, these standards are com-

bined within three subsystems: An Identity Manage-
ment System, API Management System, and an En-
titlement Management System — three fundamental
systems that should be recognizable and supported within
all IoT organizations. All are cleverly powered by open
standards to ensure platform longevity.
For example, the goal of an Identity Management Sys-
tem is to answer the fundamental question of who is
calling. Once it figures out the identity, it then forwards
this information to other subsystems. An identity man-
agement system should be comprised of:
Securing the IoT for Decades to Come 112

• A user management service to perform Create

Read Update Delete (CRUD) operations on things
like user information, groups, or connections be-
tween users. It uses SCIM, the standardized identity
API by IETF.
• A federation service to allow you to reuse cre-
dentials in another trusted domain. It brokers cryp-
tographically secure credentials using OpenID Con-
nect.
• A security token service that issues a security cre-
dential token, implementing the OAuth standard.
• An authentication service is another important
facet of the overall identity management system. It
covers many bases — it can log you in, collect creden-
tials, act as a self-service registration, and provide
Single-Sign On (SSO) capabilities. It also integrates
with different directories such as SQL databases,
other integration providers, social networks, or ex-
terior identity providers. An authentication service
is often built to the specific requirements of the
organization, and in doing so may need to deal with
older, non-standard protocols.

Acknowledgement: Open Standards

Fluctuate

Spencer recommends that developers build on new se-

curity standards that will stand the test of time, yet they
should loosely couple them in a way that allows for revis-
ing or adding new standards to the stack. This is because
standards, and the products they support, will inevitably
evolve:

“The world is changing and shifting… if we don’t

Securing the IoT for Decades to Come 113

build in a loosely coupled way where the inter-

faces between these systems are governed by
open standards then it’s going to be hard to
replace those products.”

By 2030, your API platform may not be replaced in it’s

entirety, but it will be likely be reiterated and extended.
Ensuring the platform is open from the start, based on
standards you can implement yourselves or find replace-
ments for, goes a long way to extending lifespan. It also
ensures that security will evolve alongside this evolution.

The Nuances of IoT-Based

Communication

Much of what we’ve discussed so far has been related to

internet-based communication. In which, the API security
management system is talking to APIs that are accessing
the internet over HTTP or HTTP 2. However, IoT devices
may be working with other protocols, such as CoAP, a
specialized web access protocol designed for machine-
machine dialogue.
We typically use JWT (JSON Web Token) as the standard
token, as it creates lower technical barrier, is easier parse
JSON over XML, and integrates well with lower level lan-
guages. With a little maneuvering, we can send an OAuth
message inside of a CoAP stream.
Since HTTP 2 redesigned HTTP as a binary protocol for
compactness, it would make sense to use the same mes-
sage format but as a binary representation. To do so, you
can use CBOR Web Tokens to create a compact profile of
Securing the IoT for Decades to Come 114

the JWT. This could be used within an IoT setup as a way

to optimize throughput.

5 Actionable Steps Toward

Improving IoT Security

We glimpsed into the near-distant future, and discovered

a rapidly changing world greatly impacted by the advent
of the IoT. In the Internet of Things, identity control will
be paramount for protecting users, but to get there, we
need a long-lasting API security architecture founded on
open standards.
With all this information, where do we go from here?
Spencer recommends 5 solid next steps to begin imple-
menting better API security:

1. Gap analysis assessment: Could your systems ben-

efit from a more secure identity architecture? How
important are the components for your business?
2. Gauge the impact on the platform: Assess how a
new security service will impact your existing databases.
How would this alter how clients must call the APIs?
3. Pilot and deploy in baby steps: Assemble a proof
of concept and test it with your users, assess the
strengths and weaknesses, and iterate.
4. Go live with HTTP 2!: We should all embrace HTTP
2 as it’ll make the internet a speedy, happier place!
5. Research IoT-specific protocols: To make data trans-
fers truly compat, consider CoAP as an IoT specific
protocol, and CBOR for distributing your JWT tokens.

Our future digital world will be truly incomprehensible

compared to the present, and perhaps, somewhere in
Securing the IoT for Decades to Come 115

the cloud exists the world’s biggest internet company, yet

to emerge. Now that the IoT has become a reality, we
must future proof the API security platform at it’s core.
As identity will be pivotal for understanding relationships
between devices, the best we can do now is to build
identity management systems that embrace open stan-
dards that are poised to stick around, decoupling them
so that as products evolve we can synergize them with
new components.
5 Ways the OpenAPI
Specification Spurs API
Agility

The API lifecycle is a topic of much discussion — and

rightfully so. The API space is agile, ever changing, and
its participants must continually meet the shifting kalei-
doscope of user needs and demands.
Speeding up and securing this lifecycle is seen by many,
then, to be the Holy Grail. Being able to leverage proven
API specifications is a powerful method, as proper specifi-
cation usage opens up a whole new realm for developers.
Whether this be in efficient, automated documentation,
improved usability and developer experience with dy-
namic sandboxing, or decreasing speed to market through
5 Ways the OpenAPI Specification Spurs API Agility 117

faster iteration, the benefits can’t be overstated.

Today we’ll look at a wonderful solution to help reach this
lofty goal: the OpenAPI Specification. We’ll talk a little
bit about what it actually is, and how it can enable faster,
more secure API development.

Speaker: Arnaud Lauret

This chapter was inspired by Lauret’s session
at the 2016 Platform Summit. Watch the full
talk here

What is the OpenAPI Specification?

The OpenAPI Specification is a specification and frame-

work implementation designed to create machine-read-
able interface files for describing, producing, consuming,
and visualizing web services in a RESTful architecture. As
Lauret describes, the use cases are numerous:

“The OpenAPI specification universe extends

way beyond generated documentation. It’s bound-
less. The OpenAPI specification can be used to
accelerate and secure API creation and evolu-
tion.”

Before there was OpenAPI, there was Swagger. Swagger

was originally designed for use on the Wordnik word
discovery service, a web service that was touted to “find
meaning in words.” Swagger was then designed under
this same guideline, though with a slight deviation — to
“find meaning in APIs.”
5 Ways the OpenAPI Specification Spurs API Agility 118

In 2015, API provider SmartBear acquired the open-source

Swagger framework, and began to expand it out of the
confines of Wordnik’s use case. In 2015, SmartBear helped
to create a new organization under the banner of the
Linux Foundation called the Open API Initiative, working
with Google, IBM, and Microsoft to craft new tools and
solutions. As part of their efforts, they released the Swag-
ger Specification to the group, dubbing it the OpenAPI
Specification. It should be noted that that the tooling
around the specification still retains the Swagger brand
(Swagger Editor, UI, Codegen, etc). At the time of this
writing, OpenAPI Specification is in a stable v2 with a
version 3.0 in draft status.

Why OpenAPI Specification?

The OpenAPI specification has some really special fea-

tures that make it so well-received in the API world. First
and foremost, OpenAPI is language-agnostic. Unlike other
solutions, this specification has no language preference
and can work across any language with minor instruction.
Another huge benefit of the OpenAPI Specification is
that, as a result of its specific dependence on declarative
resource specification, clients have a much larger, more
robust and holistic view of functionality. Instead of
depending on restricted server code or documentation,
the OpenAPI Specification, in many ways, describes itself.
Wrapped up into the OpenAPI Specification is the Swagger-
Codegen, a piece of Swagger technology that utilizes a
template-driven engine to generate documentation, server
foundational stubs, and clients.
For many reasons, OpenAPI is often seen as the best
5 Ways the OpenAPI Specification Spurs API Agility 119

solution to date for API development. With this in mind,

how specifically does it help us toward our goal?

API Fastness

“API fastness is probably what every API provider

aims to do — being able to deliver an API quickly
but securely.”

Lauret defines API fastness as accelerating the devel-

opment of an API and its resources while maintaining
security and quality control. Essentially, you’re aiming
for the best of both worlds, and as part of that, you’re
accepting some limitations.
In the classic development lifecycle, you’re playing a game
of balance — you sacrifice quality for speed to market,
and you give up some speed when you focus on securing
resources. Though not every project will have the same
pitfalls, slow and methodical generally counters quick
and innovative. As Mark Zuckerberg once said, “move fast
and break things.”
Luckily, the OpenAPI Specification takes away a lot of the
“break things” part of that motto and focuses more on the
“move fast”. Here are 5 ways it does that.

1: Proper Design and Approach

Any good engineer can tell you that structures don’t fail
because of the last brick, they fail because of the first. If
you start with a shaky foundation, or bad bedrock, you’re
going to collapse, and the same is true of an API.
5 Ways the OpenAPI Specification Spurs API Agility 120

The OpenAPI Specification provides perhaps one of the

strongest foundations upon which one can build an API.
It does this not by solely providing tools or approaches,
but by providing a proper design approach towards how
your API will be interacted with.

“A beginning is a very delicate time, especially

for an API. Whether an API is internal or exter-
nal, its design is crucial. A bad design can lead
to a total disaster. It can kill projects and even
companies.”

Perhaps the most powerful example of how the OpenAPI

Specification does this is in the synchronization of source
code and client libraries. Failing to synchronize these
elements can result in divergent code sets and awful
interactions for the consumer.
On the other hand, adopting a proper methodology for
ensuring that content is properly synchronized can make
for better functionality. The greatest part of all of this
is that the three main ways this specification genera-
tion is done — Codegen, automatic generation (swagger-
node-express and swagger-play), and manually — are
completely adaptable to specific API functionalities and
design aesthetics.

2: Complete Documentation and

Description

SwaggerUI, an element of OpenAPI, is supremely pow-

erful and capable when it comes to generating descrip-
tive documentation. Due to where the OpenAPI Speci-
fication came from — specifically, Swagger — the main
5 Ways the OpenAPI Specification Spurs API Agility 121

selling point of “accurate, easy documentation” that came

with Swagger is also shared within OpenAPI. As Lauret
describes:

“The OpenAPI specification is able to easily and

efficiently describe an API. Every single aspect
can be described easily in a simple but struc-
tured document.”

That’s not the end, either — the OpenAPI Specification

is open by design, which has given rise to some serious
contenders for functionality mirroring that of the Swag-
gerUI. For instance, APIs.guru have released their ReDoc
solution as an alternative system.
That’s the real power here. Any documentation service
worth its salt will get the job done, but the OpenAPI Speci-
fication is specifically built around the idea of opening up
an API to multiple solutions, languages, and approaches
in an effort to reach the best solution possible for any
given permutation.

3: Rapid Testing and Iteration

The OpenAPI Specification utilizes a declarative resource

specification, and as part of this, allows for easy explo-
ration of the API without access to the server code or an
understanding of the server implementation.
This means that, unlike other solutions where the dev has
a great sandbox, but the average user is restricted, the
API can be sandboxed in a way that allows for both devel-
opers and non-developers to understand and explore in
an interactive sandbox.
5 Ways the OpenAPI Specification Spurs API Agility 122

This is all done via that magic implementation called the

SwaggerUI. SwaggerUI reads the API description that is
stated as part of the OpenAPI Specification, and renders
it as a web-page.
While this is great from a consumer viewpoint, the real
benefit here is in testing and iteration. By being able to
actively test and modify API content during development
in a sandboxed environment, endpoints can be tested
against any permutation, resource access can be tested
in a multitude of environments, and the API can be tested
in interactions with other APIs.
This drastically aids in iteration, and enables truly rapid,
dynamic testing with minimal overtime.

4: Shorter (More Secure) Time to

Market

Due to the aforementioned drastic reduction in resource

and time dedication for testing and iteration, this also
naturally results in a decrease regarding time to market.
By taking less time to iterate and using less resources be-
tween initial revisions, this results in a much quicker cycle
as features are rolled into the main codeset dynamically.
A key sidenote here is that this also drastically increases
the security level of the API. Because the developer is
able to test and iterate with relative ease and quickness,
issues in the code base and its interactions are discovered
earlier in testing, and bugs are attached directly to stated
resources.
Imagine building an engine. When putting the pieces
together, you have to test each element, and upon the
5 Ways the OpenAPI Specification Spurs API Agility 123

addition of a new part, you must test the engine as a

whole. Now imagine the same situation, but with a super
computer that allows each individual part to be tested
no matter how deep the part is, how surrounded by
other parts it is, and how core or ancillary to the overall
functionality it is.
This is functionally what the OpenAPI Specification allows
you to do — test anything dynamically and with low
overhead and time between testing.

5: Machine and Human Readability

and Translation

“The OpenAPI specification can also be used to

create human-readable documentation.”

The OpenAPI Specification is a Rosetta Stone for the

API space. Whereas other solutions generate machine-
centric data for the purpose of functionality and then
derive from this human readable documentation and
specification (or vice versa), the OpenAPI specification
starts from a different position — a progenitor language
from which all other content is derived.
By functioning as a specification which describes the API,
the OpenAPI Specification can then derive both machine
and human readable content in a better, more complete
way. This approach led Swagger to become a household
name in the API space, and it continues to influence
documentation trends to this day.
The OpenAPI Specification leverages its powers to create
the most complete and useful content in both machine
5 Ways the OpenAPI Specification Spurs API Agility 124

and human readable format; an important ability that

simply cannot be overstated.

Conclusion: OpenAPI Enables an

Agile API Lifecycle

As developers, API providers, and evangelists, the biggest

issue we face in this industry is moving at a speed quick
enough to iterate and innovate, while securing the re-
sources our consumers depend on. It’s no easy task —
many have tried and failed (and quite spectacularly so).
The OpenAPI Specification is possibly the best implemen-
tation at this moment for this exact purpose. Due to how
it generates documentation, APIs from a description, and
does so in such a dynamic, diverse way, the OpenAPI
Specification bridges both worlds, enabling, as The API
Handyman calls it, API fastness.
Case Study: Bosch
Ongoing Enterprise API
Management Saga

Have you ever tried to implement a new process in a big

company? Believe us when we tell you that it’s not an easy
(or fun) task. From obtaining initial approval all the way
through to disseminating information, it takes time and
there are often bumps along the road.
This is something that Josh Wang, project manager at
Bosch Automotive Aftermarket, knows all about. Recently
tasked with implementing an API management solu-
tion at Bosch, he talks about the incredible amount of
data associated with Bosch’s products, which range from
mobility solution and industrial technology through to
consumer goods and building technology.
The aim of the project – to implement an API manage-
ment platform that would allow them to use their APIs
more effectively and make more useful products for cus-
tomers – was a slam dunk for Bosch. With billions of
data points being generated via end user usage of their
products, it made perfect sense for them to want to
streamline and solidify their API management practices.
This chapter will cover some of the lessons learned from
Case Study: Bosch Ongoing Enterprise API Management Saga 126

enterprise API management implementation — though

that’s not to say that there’s nothing to garner if you’re
working at a much smaller company — many of these
best practices apply whether you’re managing a very
large number of different APIs or creating and maintain-
ing just a couple.

Speaker: Josh Wang

This chapter was inspired by Wang’s session
at the 2016 Platform Summit. Watch the full
talk here

Is building everything yourself

always the answer?

For developer consumers, there’s a certain stigma associ-

ated with building your property on someone else’s land –
What happens if they shutter the API? How will downtime
affect our app? The question posed above is an important
one to ask.
When beginning a developer program, you must consider
whether or not there’s already an API out there that can
do the type of thing that you’re looking for. For example,
there’s no need for a website looking to give personal
recommendations for movies to build their own API when
they can just use IMDB’s data instead.
That’s a basic example, but the idea holds true even
in large enterprises – it’s much easier to convince your
team to use an existing product than it is to get C-level
execs, developers, marketing and operations on board
with building one of your own.
Case Study: Bosch Ongoing Enterprise API Management Saga 127

There are certain parallels here between using a third

party API management platform and trying to build
your own API in that making use of what’s already out
there can save a lot of time. In the case of Bosch, however,
one thing was for sure – whatever the solution was, it
needed to be extremely robust. As Wang says, “we’re
building stuff that can potentially burn your house down,
crush you or do harm to you if it’s not done 100% cor-
rectly.”
In this context, it’s clear that relying on many different
developers to maintain APIs that need to work seamlessly
together wasn’t the best idea for Bosch. They settled on
using a third party API management solution and, while
that might sound like a luxury for smaller organisations,
it makes good sense when planning for scalability. We’ll
talk more below about cases in which that might be
particularly important.

Have an effective strategy

Regardless of what you’re doing with APIs, it’s best to start

with a list of requirements – what do you need to be able
to do with your API(s)? How will existing infrastructure
impact what you can do? Can your current setup handle
the extra load?
In Wang’s case, this list comprised more than 200 re-
quirements. Large enterprises have to deal with multiple
security zones, each with clearances, manual registra-
tion processes AND different users required to complete
those processes. “This is far, far away from continuous
delivery,” Wang jokes.
Smaller companies are lucky in that they’re generally
Case Study: Bosch Ongoing Enterprise API Management Saga 128

better equipped to develop in a more agile fashion, but

security, data protection, and uptime of existing services
are still things that need to be taken into account when
you’re planning your next API move.
And, as much as we hate to bring it back to this, organi-
zational complexity will always be a huge factor in API
deployment and management. “Within Bosch, for exam-
ple,” says Wang, “there are so many divisions, business
units and people with authority, that you need a platform
that is flexible enough to facilitate the entire API lifecycle
workflow.”
Without that buy-in from others in the company, you’re
going to be spinning your wheels every step of the way
during any API-related activity.

Be ready for that strategy to fall

apart

“Life is what happens to you while you’re busy making

other plans”, said Allen Saunders, and he might as well
have been talking about software development when he
did so.
Wang admits that he was naive in coming up with a
timeframe for the execution of his project to implement
a third party API management system from Axway that
could handle the many APIs used by Bosch. “The team
we were working with said it would take two days, maybe
five. Ok, let’s say ten days.” Just to be on the safe side!
The reality was that implementing a management system,
designed to simplify the maintenance of and interaction
Case Study: Bosch Ongoing Enterprise API Management Saga 129

between existing APIs, took far longer than this: man-

agement nodes couldn’t communicate properly with the
current security setup, necessitating a design of excep-
tion, firewall issues, etc. The infrastructure for the new
domain needed to be changed four or five times. Even
tiny problems like Bosch’s existing system not allowing
the execution of scripts in the temp folder, which was
required by Axway, seemed to snowball.

“And those are just the small problems!” says

Wang.

Whether we’re talking about API management or imple-

menting a new API, flexibility is key. On the former, Wang
prompts enterprises to ask big questions like “how can
you define your API management architecture in a way
that supports on-premise hosting and cloud solutions?”
and to “think about the different use cases with security.”
But he’s eager to point out that platform providers have
plenty of work to do as well, saying that there are “chal-
lenges that providers of API management solutions and
gateways need to address if they want to offer their
solutions to large enterprises.” It’s important to note here
that this doesn’t mean that management providers like
Axway are necessarily doing anything wrong, only that
the relationship between them and large organizations
are so (relatively speaking) new that it’s common for
unpredictable issues to arise.
For example, Wang talks about some of the legal require-
ments associated with API management, joking that this
is something most developers don’t like to think about:

“In Germany, you need to record the version

and date of when somebody has agreed with
Case Study: Bosch Ongoing Enterprise API Management Saga 130

the terms and conditions in the privacy state-

ment. If the system does not support a dedi-
cated database that can store this kind of infor-
mation, it’s very difficult to convince the legal
department that you can go live with the plat-
form.”

That’s just one example, but it’s a compelling one that

highlights the difficulty creators of API management plat-
forms face: they don’t only need to measure up to the
technical requirements proposed by developers, but must
address a number of other issues as well.

Think big, even when you’re small

Some of this may feel like overkill for startups and smaller
businesses, but APIs are like Pringles – once you pop, you
just can’t stop. Salesforce, for example, has 10 APIs and
Microsoft Cognitive Services have 23. Google has such
a large suite of APIs that they have an API explorer to
navigate them. And, with companies increasingly using
freemium or paid API models, the trend of multiple APIs
and microservices will only become more prevalent.
Wang highlights the way in which the size of a company
doesn’t always correlate with the complexity of an organi-
zation anyway: “Even small companies have very complex
authorization structures in roles sometimes, depending
on what kind of domain they’re working in, and it’s really
important to have a platform that supports different
roles, additive roles etc.” Even smaller companies need to
think about all of this, sooner rather than later, if they’re
thinking about building a suite of APIs.
Case Study: Bosch Ongoing Enterprise API Management Saga 131

There’s a certain element of future proofing inherent in

the idea of building or implementing an API management
platform: the insinuation is that APIs will become, or at
least have the potential to become, a key part of the busi-
ness. That can be a smart move since lots of companies
develop a single API then, before they know it, they’re
hosting multiple APIs that probably aren’t synchronised
as well as they could be.
To go one step further, API Evangelist suggests that “per-
sonal APIs” are not only on their way, but already here. As
employees bring their own APIs to the table, potentially
using them in their own individual workflow, an approach
that’s very open to API management platforms makes
sense.
It might seem far-fetched for some companies to start
thinking about API management, and the problems as-
sociated with it in large enterprises before they’ve even
built their first API, but it could just end up being one of
the smartest business moves they ever make.
In Summary
From REST to securing the Internet of Things, in this vol-
ume we’ve covered a lot of ground. To summarize, here
are the key takeaways that API designers and enterprise
architects can glean from each chapter:

TL/DR - 15 Important Takeaways

• Understand true REST API design: We responded

to misconceptions of REST API design, reviewed hy-
permedia, and explored what it takes to create a
HATEOAS-compliant state machine.
• But consider GraphQL: GraphQL performs select
functions better than REST, but it means a significant
reversal of modern REST API design standards.
• Private APIs benefit from continuous versioning:
Eradicating the typical URI versioning schematic (v1,
v2, etc) could withhold the server to client bond,
equating to consistency and better API agility, how-
ever is largely unproven in public scenarios.
• API-fy internal processes: Spotify brilliantly uses In-
ternal APIs to streamline their varying payment type
subscription options. Consider how internal APIs can
bring platform-wide consistency to improve your UX.
• Have a serverless API backend: Serverless archi-
tecture offers an infinitely scalable cloud backend
for APIs and web applications, equating to a lean
platform and cost reduction.
In Summary 133

• Put an end to polling: Allowing clients to continu-

ally poll APIs can be a huge, wasted drain on your
resources. Instead, use REST Hooks, or alternative
means such as websockets or Server-sent.
• Master microservice gardening: API providers must
eschew monolithic centralization in favor of inno-
vation and new developments. This includes using
Bimodal IT for parallel tracks, and a microservices
architecture.
• Model automotive IT for API longevity: API providers
could take a lesson or two from automotive grade
manufacturing - automakers must build long-lasting,
reliable IoT-centric APIs that stand the test of time.
• Use OAuth 2.0 to secure the IoT: The IoT is coming,
and OAuth 2.0 is the way to secure it.
• Avoid common API design anti-patterns: Always
consider the operational repercussions for the de-
sign moves we make now. Avoid improper HTTP
method usage, protocol tunneling, polling, and rigid
microservices structure.
• Personal data is valuable: With the rise of open
banking, programmatic accessibility to the “API of
Me” is becoming more realistic. Keep in mind the
value of user data, and the government regulations
mandating its liquidity.
• Use DevOps: When Humpty Dumpty falls and cracks,
instead of pointing fingers, development, operations
and QA should work together. This means having a
DevOps mindset.
• Secure the platform for decades: IoT and API secu-
rity must unite, so that developers can begin to scale
their platform and security measures accordingly.
This means building on open standards.
• Use the OpenAPI Specification: For growing and
securing the API lifecycle, use a powerful API speci-
In Summary 134

fication format. The OpenAPI Specification is a great

solution to boost platform agility.
• Enterprise API management techniques: For lessons
in enterprise grade API management, we studied
Bosch’s experience implementing Axway’s API man-
agement solution across billions of data points.

What’s Next?

The 2016 Platform Summit focused on API longevity,

emphasizing the strategies architects must deploy to con-
struct everlasting, holistic API programs. As we gear up
for the 2017 Platform Summit, our focus will shift to
scalability. If you are interested in joining the lineage of
great Nordic API speakers, consider submitting a speaker
session!
We hope you enjoyed API Design on the Scale of Decades,
and let us know how we can improve. Our eBook releases
usually come paired with an announcement for a new
title, and this release is no different. Guess what… it’s a
GraphQL book!
As we’ve covered before, GraphQL is the query language
making ripples throughout the economy. GraphQL or Bust
will aim to once and for all determine the position of
GraphQL within the API ecosystem. We’ll explore things
like the benefits of GraphQL, the differences between it
and REST, nuanced security concerns, extending GraphQL
with additional tooling and GraphQL-specific consoles,
making a transition to GraphQL from an existing web API,
and much more. Follow our progress on Leanpub!
In Summary 135

Stay Connected

Thank you again to our readers, event attendees, and

event sponsors and partners. If you appreciate what we’re
doing, consider following @NordicAPIs and signing up to
our newsletter for curated blog updates and future event
announcements.
Nordic APIs Resources

2016 Nordic APIs Summit

Talks

You can watch full videos of the talks featured in this

eBook below. In order of appearance:

1. The REST and Then Some (Asbjørn Ulsberg, PayEx)

2. Is GraphQL the end of RESTstyle APIs? (Joakim Lund-
borg, Wrapp)
3. Versioning Strategy for a Complex Internal API (Kon-
stantin Yakushev, Badoo)
4. How Spotify Payments Creates APIs to Manage Com-
plexity (Horia Jurcut, Spotify)
5. Building Serverless API Backends (Rich Jones, Gun.io)
6. Stop the Polling Madness and Adopt REST Hooks
(Audrey Neveu, Streamdata.io)
7. Bimodal IT for Microservice Gardening (Erik Wilde,
CA Technologies)
8. Look to Automotive IoT for a Lesson in API Longevity
(Henrik Segesten, Volvo)
9. OAuth 2.0 and the Internet of Things (IoT) (Jacob
Ideskog, Twobo Technologies)
10. API Design Anti-Patterns for Microservices (Jason
Harmon, Typeform)
11. Will Open Banking Trigger the API of Me? (Chris
Wood)
Nordic APIs Resources 137

12. Avoid Walking on Eggshells and Use DevOps (Emily

Dowdle, Wazee Digital)
13. Platform Security that will Last for Decades (Travis
Spencer, Twobo Technologies)
14. Secure and Accelerate the API Lifecycle with Open
API Specification (Arnaud Lauret, AXA Banque)
15. Assembling The API Management Jigsaw In Big (And
Small) Companies (Josh Wang, Bosch)
Nordic APIs Resources 138

More eBooks by Nordic

APIs:

How to Successfully Market an API: The bible for project

managers, technical evangelists, or marketing aficiona-
dos in the process of promoting an API program.
The API Economy: APIs have given birth to a variety of
unprecedented services. Learn how to get the most out
of this new economy.
API Driven-DevOps: One important development in re-
cent years has been the emergence of DevOps — a disci-
pline at the crossroads between application development
and system administration.
Securing the API Stronghold: The most comprehensive
freely available deep dive into the core tenants of modern
web API security, identity control, and access manage-
ment.
Developing The API Mindset: Distinguishes Public, Pri-
vate, and Partner API business strategies with use cases
from Nordic APIs events.
Endnotes
Nordic APIs is an independent blog and this publica-
tion has not been authorized, sponsored, or otherwise
approved by any company mentioned in it. All trade-
marks, servicemarks, registered trademarks, and regis-
tered servicemarks are the property of their respective
owners.

• Select icons made by Freepik and are licensed by CC

BY 3.0
• Select images are copyright Twobo Technologies and
used by permission.

Nordic APIs AB Box 133 447 24 Vargarda, Sweden

Facebook | Twitter | Linkedin | Google+ | YouTube
Blog | Home | Newsletter | Contact