AWS APPSYNC

SERVERLESS GRAPHQL APIS AT SCALE

INTRODUCTION GRAPHQL COGNITO INTEGRATION DATA TYPES AUTHORIZERS
Since Facebook released its previously internally used query REST APIs can become a burden for clients with an ever AppSync natively integrates with Cognito, which allows you Every GraphQL object in AppSync has a name and fields. You can protect your API with an Authorizer.

language GraphQL in 2016, it has seen an outstanding growing number of endpoints. Also, there’s maybe never a to do fine-grained access control at a per-field level. Fields in turn can have sub-fields.


increase in adoptions for all kinds of applications.

perfectly fitting one to only fetch the data actually needed, Your authorizer can be one of the type
so clients often end up receiving more data than necessary.

Each field’s type must resolve to a scalar type, either from API_KEY - authorized by API key
So it didn’t come as a surprise when AWS built a managed REAL-TIME DATA VIA SOCKETS ️ the default set of GraphQL or from AppSync’s extensions AWS_LAMBDA - authorized by a Lambda function
Subscriptions give you the ability to send real-time data to
solution called AppSync and released it in 2018. AppSync is GraphQL was built to solve those problems by enabling the prefixed with AWS, or to one of your own custom object AWS_IAM - authorized via AWS IAM
your clients without the need of any additional services.

becoming more and more popular and is already being used client to specify how to structure the
types.
OPENID_CONNECT- authorized via OpenID Connect
It is natively integrated into AppSync.

by large companies, including BMW and Sky. data when it is returned by the server.

AMAZON_COGNITO_USER_POOLS - authorized by a
For every mutation your client gets updated automatically.
Cognito identity pool.

For example in React simply use useSubscription.

GraphQL Scalars AWS Scalars

GraphQL’s top-level operations are
CACHING ️ Query - read-only fetc
ID: Unique identifier, serialized as a String
String: A UTF-8 character sequence
AWSDate: Extended ISO-8601 date (YYYY-MM-DD)
AWSTime: Extended ISO-8601 time (hh:mm:ss.sss) This secures your whole API. 

AppSync allows you to cache data on the server-side and by You can also use enhanced filtering via JSON in the response Int: An integer value AWSDateTime: Combination of Date & Time (T separator)
Mutation - write, followed by a fetc You can also set authorizers on 

that enjoying several benefits mapping template of the subscriptions resolvers.
Float: An floating point value AWSTimestamp: seconds before/after 1970
Subscription - long-lived connection for receiving data Boolean: A Boolean value, either true or false AWSEmail: email address as defined by RFC 822 field level. For example if you want

In-memory cachin Even multiple rules can
AWSJSON: valid JSON construct - automatically parsed
to secure your Customer with

Improved performanc be used together either
AWSPhone: phone which can contain dashes/hyphens
IAM you can add the annotation @aws_iam. Similar to all
DATA SOURCES & RESOLVER in the same filter (AND)
AWSURL: URL as defined by RFC 1738
Decreased latency

AWSIPAddress: a valid IPv4 or IPv6 address. other methods.

A resolver defines how AppSync processes your queries and or while having multiple

In the background, ElastiCache for Redis is used.

mutations. It’s the glue between your schema and the actual filters inside a group (OR).
data sources.

SCHEMA CODEGEN
Caching can be used at different levels Schema files contain your GraphQL definitions incuding Codegen is a CLI tool that allows you to create types,
Full-request: $context.arguments and $context.identity One of the main advantages of AppSync is that you can MONITORING query, mutation & subscription.

queries, and mutations directly from your GraphQL schema.

CloudWatch already collects a lot of metrics for you,

is used as key for caches natively integrate with different AWS services without
including the number of HTTP 4xx and 5xx errors, request
Per-resolver: Each resolver will be cached. Individual writing any code.

You can create your own object

You simply need to define where your

latency, the number of API requests and caching statistics
keys based on the context. types which include fields with
schema is located (schema.graphql)

like hits, misses or evictions.

Execution Summary Logs

A resolver is attached to a field, for example to the orders of Request Summary Logs
scalar types or even other object
and define where to save your 

a customer.

Enable Logs types. Meaning, you can deeply

generated files & it’s done.
Amplify For detailed insights
Include Verbose Content
GraphQL Request
Request & Response nest objects.

Amplify allows you to generate VTL resolvers from your about resolvers, you
ield Resolver Log Level
F Headers

There are three different kinds of resolvers ERROR For all custom types like Customer

data sources automatically.

need to enable logging, which

with $context
Tracing Logs

enhanced

By adding mutations, you define

ALL
VTL Resolver - connect your data directly to AWS Request Mapping Logs in our example you can simply define custom scalars and
With their CLI you can simply define an annotated GraphQL comes with many options.

resources like DynamoDB, RDS, or OpenSearch with the Response Mapping Logs
how items will be inserted or
map them to specific types.

schema and all of your resolvers are created automatically.

Apache Velocity Template Language. This does not updated and what are the return

Verbose logging for all requests will generate very deep

require you to write any code, but only mappings types.

Codegen has tons of plugins out there for creating all sorts
Simply start with amplify add api and run through the insights, but also increase the number of logs by an order of
between inputs and outputs of types and code from your GraphQL schema.
wizard with selecting GraphQL as your target service. You’ll magnitude. Be careful, as CloudWatch ingestion is pricy.

Direct Lambda Resolvers - if you want to use Lambda as Subscriptions allow you to listen

be asked if you want to edit your schema in your favorite

your resolver you don’t need to use VTL and attach your to certain events in your API, e.g.

text editor.

For easy observability out-of-the-box, make use of external

Lambda function directly. This is the most flexible a creation of a new customer.
services like Dashbird.io which doesn’t require you to
approach that has basically no limits
Via amplify push the API will be deployed to AWS so you create custom alarms or notifications.
Pipeline Resolvers - execute multiple resolvers before
can immediately start using it.
and after the execution.