Module 1.

Cloudian Hyperstore Introduction

In today's storage world, customers have the challenge of dealing with
both capacity and complexity.
Part of the problem is the incredible capacity growth.
And, as you can see by the image on the left, data is increasing at an
exponential rate.
This increased capacity brings with it storage complexity.
Think tanks such as the IDC are predicting that the global data sphere by
the year 2025 will be approaching upwards of 175 zettabytes.
What is more incredible is that 80% of that global data sphere is
predicted to be unstructured data.
To explain the difference between structured and unstructured data, if
you think of structured data as being stored in databases,
data that is stored in an organized fashion, and unstructured data as
everything else.
Unstructured data can come from a variety of places.
In this slide, we are showing some of the areas where we see customers
generating unstructured data.
Back up in archive is a key area, and we are seeing a huge growth of data
being kept.
Some vendors talk about information lifecycle management, and even have
services available to identify data, which was redundant, obsolete, or
trivial, which we call ROT.
The classification of this was with a view to being able to allow those
customers to decide which data they need to keep and which data they need
to delete.
Now, though, we are finding that customers are not deleting any data.
One reason for that is due to the introduction of software and
intelligence that can derive value from a seemingly unimportant or old
data set.
Also, customers aren't deleting data for things like compliance and where
they have a requirement to keep data for legal and compliance reasons.
Some customers don't want the ownership and infrastructure costs of
hosting their own environment,
and are instead looking to manage service providers to supply them with
their storage capability.
Median entertainment is another huge use case.
The resolution of raw video footage and images are now generating
hundreds of gigabytes per hour.
And lastly, security surveillance and research.
Cloudy and HyperStore is the solution that can help customers deal with
all these challenges.
Meet Cloudy and HyperStore.
Cloudy and HyperStore is a simple, modular, and cost-effective solution.
You can start small and grow.
HyperStore starts as small as three nodes and can grow to theoretically
an infinite number of nodes.
It is also hybrid and multi-cloud ready.
You may be thinking there are other organizations out there, like
Microsoft Azure, AWS, and Google Cloud Platform,
that all provide very similar solutions.
But we will talk about over the next few modules why it is important and
why we are seeing customers go from external cloud providers
to their own internal cloud storage solution.
At this stage, all the benefits that you see in a public cloud, like
transparent scalability and security built-in,
are all available with Cloudy and HyperStore, with the added benefit that
it resides in your own data center.
We talked about cloud in the previous slide, so why is on-premise storage
still essential?
Some of the reasons are shown here.
We are going to jump to the last in the list because it tends to be a
very important reason why customers,
who perhaps have been in the cloud, are now bringing their data back on-
prem.
And that is the cost of access.
The cost of access is not just the cost of storing the actual data.
It is the entire cost of access.
Some public cloud vendors have a very low price per gigabyte, which is
very enticing to customers.
But later discover this is not the full story and the cost of retrieving
that data can in some cases be significantly more.
You do not have that complexity when you have the storage on-prem.
Another reason is compliance.
Some customers are not allowed to store their data outside of their
country.
For that reason, and especially if there is no public cloud vendor in
their country, then they have no option but to store the data on-prem.
Security.
Some customers just don't trust their data with third parties.
They may not be allowed to entrust their data with a third party.
And again, on-premise becomes essential for those customers.
Lastly is performance.
Performance when you are accessing data across a wide area network is not
as performant as accessing that data over a local area network.
So, where does Cloudy and Fit, as discussed previously, there is
structured and unstructured data in the global data sphere.
Around 80% of stored data is unstructured and that is the work-lown where
Cloudy and Fit's.
There is still very much a need for environments with fast data access
and there are many vendors out there that will provide that for customers
which constitutes the other 20%.
As the previous slide suggests, object storage is not a one-size-fits-all
solution and there is still very much in need for traditional storage
offerings.
All flash-based arrays, traditional sand and naz, as well as object
storage, all have their benefits and challenges.
Some of the challenges of traditional storage is in their ability to
scale and what you will find in lots of non-object storage solution,
islands of storage start to become prevalent because they are very good
at scaling up but not good at scaling out.
Scaling is something the object storage was designed to be able to do and
achieve and does that exceptionally well.
One of the challenges with an increase in the exponential growth that
customers are experiencing is the price of that storage.
Typically, with flash or sand-based arrays, is the cost per gigabyte is
extremely high.
Object storage however, is an extremely low cost per gigabyte in
comparison. There is no such thing as a free lunch though and with these
benefits do come some consideration,
especially in the terms of latency. Object storage is not designed for
highly transactional workloads in the hundreds of thousands of IOPS.
This is still very much a use case for all flash arrays. Their ability to
deal with highly transactional workloads,
requiring microsecond latency is their sweet spot and are perfect for
things like virtual machines or databases,
whereas running these workloads on object storage would not be a good use
case. Why Cloudion?
Due to the modular architecture of Cloudion, it makes it inherently
capable of scaling out.
We can present this data through a single global namespace and we have
the ability to provide an on-prem storage solution with hybrid capability
and multi-cloud ready for our customers.
This training is specifically geared towards HyperStore object storage.
However, we also have another solution which complements HyperStore
called HyperFile.
HyperStore is Cloudion's object storage solution. Each object storage
solution vendor has a slightly, and in some cases, significantly
different programming interface.
One of those programming interfaces developed by Amazon Web Services is
the S3 programming interface.
S3 stands for Simple Storage Service. During the design of HyperStore, it
was decided to follow the AWS S3, such that any applications created to
use the AWS standard would natively work with Cloudion HyperStore.
However, not all applications are able to write to an object store
because they have not been architected with an object storage programming
interface in mind.
That could be for many reasons, one of which being there are legacy
applications that were created before the object standard,
and before object storage was prevalent in the marketplace.
HyperFile helps customers who have such applications where those
applications would still be a good use case for object storage, but don't
have the ability to utilize the S3 API.
HyperFile as a NAS controller allows legacy applications using protocols
such as SMB and NFS to store data on Cloudion HyperStore.
HyperFile, which we will discuss later, has additional features to help
customers around compliance and work, while is also allowing the ability
to integrate with the customer's active directory.
Importantly, all of this technology is available and can reside in the
customer's own data center.
As already discussed, Cloudion HyperStore's ability to scale out is one
of its defining features. A single Cloudion HyperStore cluster provides
X-Sabyte scalability.
In this example, 200 nodes in a single region with 10 data centers in a
region, supporting 20 regions would give the potential of 3 X-Sabytes in
a single system.
This is a significant amount of storage, but still can be managed as a
single device and provides immense scale, not just in capacity, but also
in front-end compute.
Cloudion HyperStore is proven in multi-pedabyte deployments worldwide.
So, we have mentioned the Amazon S3 API, simple storage service, but why
does this matter?
It matters because our customers were keen to adopt the S3 API and
because customers adopted that application programming interface.
Many vendors started to write applications in modules specifically to be
compatible with Amazon's S3 API.
Cloudion benefits from that decision because all of those applications,
which have been written to interface with the Amazon S3 API natively, are
usable with Cloudion HyperStore.
Cloudion delivers the highest compatibility of any S3 object vendor.
Our commitment to the S3 API is such that we have an S3 guarantee, which
effectively means that any application that works with the AWS S3 API in
Amazon will work with Cloudion HyperStore.
And if it doesn't, then we will work to resolve that difference.
Some other competitors who also wrote their using an S3 API are doing so
using an S3 access layer.
Cloudion's core product is written around the S3 API, which allows us a
greater level of compatibility with no translation.
Cloudion HyperStore has integrated data protection.
We will talk at greater length on the options for this in a later module.
However, at a high level, we have two methods in which we store the data
in a protected manner.
The first is replication.
Although this slide states replication for DR, replication is also a
valid method of protection in single site deployments.
The second is a ratio coding, where data is striped across a number of
nodes.
The primary difference between these two mechanisms is in the efficiency
of the storage of the data and its tolerance to failure.
But more about that later.
Our data protection features are also able to protect this data, not only
locally, but also into the cloud or remote sites.
Cloudion HyperStore is Cloud-ready.
We support distributed, hybrid, and multi-cloud.
All of the data management is managed with HyperStore, and there is no
extra software to buy in manage.
This is incredibly powerful, and as can be seen from this slide, it is
possible to store your data using HyperStore in the majority of cloud
platforms out there today.
Why would you want to do that?
Considering the previous statements around performance, cost of access,
compliance, and security, it may be that some customers don't have the
luxury to use the data.
They don't have the luxury of a multi-data center infrastructure.
For those customers, and for their most important data, they may require
or need an off-site copy of that data for disaster recovery scenarios.
Cloudion HyperStore fully supports that requirement.
There are a choice of deployment models for both HyperStore and
HyperFile.
HyperStore truly is a software-defined storage solution.
The software can be deployed on any X86 hardware or virtual machines, not
so with some of our competitors, who are only able to provide their
solution as an appliance.
The benefit to our customers is that there is no hardware lock-in, and
customers are free to change their hardware choice at any stage.
It can be deployed on-prem or in the cloud.
Customers can also opt to choose an appliance if they wish.
With the benefits of choosing an appliance being end-to-end support for
the solution.
HyperFile also supports deployment using an appliance, or it can be
deployed on virtual machines on-prem or in the cloud.
Cloudion provides that flexibility to deploy across multiple
environments.
HyperStore appliances come in several configurations.
The smallest appliances are the 1500 range.
They are ultra-gents, one-use servers with 12 hot-swappable, 3.5-inch
disk drives, and up to 4 hot-swappable, 2.5-inch SSDs.
Disk drives sizes vary and are available in 8, 10, and 12 terabyte sizes,
giving a storage capacity of 96, 120, and 144 terabytes respectively.
For larger deployments, the HyperStore 4000 range are ultra-dense 4U
platforms.
These have two nodes in a single chassis and support 70 hot-swappable
drives, with 35 in each node.
These appliances also support up to 4, 2.5-inch SSDs.
High-capacity disk drive sizes of 8, 10, 12, and 14 terabytes are common,
giving a raw capacity storage of
560, 700, 840, and 960 terabytes.
Our largest appliances are our HyperStore Extreme Appliances.
This is the industry's first 96-drive 4U server, with each supporting 1.5
petabytes in a single 4U chassis.
And up to 18 petabytes in a single rack, HyperStore Extreme was the first
to market with 16 terabyte high-density disks.
Again, these are two nodes in one chassis with 96 hot-swappable 3.5-inch
drives, and up to 4 hot-swappable SSDs.
With high-density drive size of 16 terabytes, this gives a raw storage
capacity of 1,536 terabytes.
Cloudion has also partnered with the following strategic partners and are
certified on multiple leading platforms,
such as HP, where typically we will see the deployments using the Apollo
4000 range of servers.
Lenovo with DX8200 servers and Cisco's UCS-S3260 servers.
Why do customers choose Cloudion?
Simply put, we have the industry's highest S3 API compatibility, which is
key for most of our customers.
We can provide buy modal access to both file and object storage.
We can start small with just three nodes and grow to a theoretically
unlimited number of nodes
and provide granular management at the bucket and object level.
We also, as previously discussed, support both hybrid and multi-cloud
requirements.
We have the ability to protect data through the use of multiple storage
policies,
which allows a single storage cluster to protect data using efficiency
that is appropriate for the customers
requirements, all within a single cluster.
Due to the self-healing and distributed nature of Cloudion Hyperstore, up
to 14 nines of data durability
is achievable with a cost per gigabyte, around half a cent per gigabyte a
month.
What are some of our common use cases?
Backup and Archive is a key use case with most of the major backup and
archive vendors
supporting integration with an S3 object store.
Some of them are shown on the slide.
rubric, veritas, Veeam, and convo being the most common.
With storage as a service being a popular option, managed service
providers are using Cloudion Hyperstore
as their object storage solution to resell to customers due to its cost,
flexibility, and use of the S3 API.
Another significant use case is in media and entertainment,
where we have many customers using Cloudion Hyperstore to store and
manage large amounts of video and audio files.
Lastly for this slide, primary storage archiving.
As mentioned earlier, information lifecycle management is a key component
for major storage vendors
who understand that the value of data changes over time.
And having the flexibility of moving from a highly performant block
storage device,
where data is frequently accessed and modified to a lower cost archive
tier,
when that data is no longer actively being read or changed.
Some other common use cases are file, sync, and share, like storage made
easy.
Higher archical storage management, like HP's Store Next,
email archives such as Veritas Enterprise Vault, and big data machine
data activities like Splunk.
What does this all mean?
With Cloudion, you can think big.
We enable hyperscale storage for limitless capacity growth.
It provides infinite storage with infinite possibilities.
It is easy to manage and grow and has some of the richest feature sets
available in the marketplace
with multiple petabytes deployment worldwide.