Category

Scaling the RAC

Russ Feed Scaling the RAC

Pure //Accelerate 2019 – Cloud Block Store

September 19, 2019

Pure //Accelerate 2019 has come and gone with the theme of delivering “The Modern Data Experience” in a “Simple, Seamless, & Sustainable” manner and of course… several new announcements including:

  • Flash Array //C 
    • Cost-effective capacity optimized flash storage with the same features, reliability, and API’s as all of the FlashArrays before it. 
  • Pure as-a-Service
    • A rebranding of the “Evergreen Storage Service” in an effort to encompass everything that Pure is now offering both on-prem and in the cloud.
  • AI Data Hub
    • Quickly deploy AI workloads by unifying data that was previously stranded by multiple application and storage silos. 
  • Cloud Block Store for AWS
    • The Purity OS being leveraged natively in AWS to create more simple, reliable, and consistent operations across on-prem and cloud deployments.

The key to these announcements is the general availability of all of them, but there were plenty of other announcements that were included in the middle of the four that are highlighted above. Some examples would be topics involving DevOps, Kubernetes (K8s), and  leveraging vVOLs across the FlashArray //X and the //C in order to create policy based storage classes through SPBM similar to the way we defined storage classes for automated persistent volumes in Kubernetes. Do not think for one second that I am dismissing or glossing over any of those announcements because they are HUGE and I definitely want to come back and revisit vVOLs in a discussion involving K8s as well. 

However….

This post is a quick perspective piece that I want to focus in on for one singular announcement. 

Cloud Block Store (CBS)

I have had and listened to several conversations around all of the announcements, but none more than CBS, which has some obvious use cases such as:

  • Lift and Shift workloads to the cloud (make sure you are using vVOL’s if you are going to do this with VMware workloads)
  • Dev/Test
  • Disaster Recovery
  • Snap & Rep as a PIECE of your backup strategy (DO NOT make it your only backup strategy)
    • Worth noting that this can also be accomplished with “Cloud Snap” as well, which does not require CBS.

All of those things are great, and some may even say they are obvious (I agree). That being said, the prevailing theme I heard and the primary reason for this post is that the perspective of cloud block store is that it is “just a virtual storage array running in AWS”.

Sure, technically, that is accurate but I want to present a bit of a different perspective…

When we design and build architecture we do so in a way to account for a number of different variables such as reliability, durability, scalability, performance(ability), availability, manageability, and any other “abilities” you can imagine. In particular when thinking about this in the context of designing around all of the services inside of a public cloud (or really anywhere) it reminds me of the old “Build vs Buy” conversation that started with HCI (Hyper-Converged Infrastructure).

To me, CBS is more of a best practices design blueprint that enables me to CONSUME storage resources as opposed to BUILD & MANAGE them. 

This is not about “inserting Purity into AWS”, but instead deploying an extremely intelligent and native architecture (I would post a picture, but I don’t have a good one and am unsure if I am allowed to. Will update if this changes) to automatically provide you with storage resources that meet the required “abilities” mentioned above without you needing to either custom design and build them yourself (Ops) or to design that into the application (Dev). 

There is so much to unpack and discuss on that topic, but the essence is that you now have an option to provide consumable resources to the teams that need them without those teams worrying about anything other than consuming them. Perhaps we could even start calling it “Declarative Storage” for AWS that looks something like this:

apiVersion: storage.AWS.io/v9000
kind: StorageClass
metadata:
name: PureCloudBlockStore
annotations:
storageclass.AWS.io/is-default-class: "true"
labels:
addonmanager.AWS.io/mode: EnsureEntAbilities
provisioner: CBS.io/aws-purity
parameters:
type: purity-enterprise
allowVolumeExpansion: true

That is probably a bad joke, but you get the idea. Cloud Block Store is less of a “storage product in the cloud” and more of a “design blueprint” to automatically deploy, scale, protect, and ultimate service storage resources that also happens to enable data mobility and consistent operations both on-prem and in the cloud!

Best,

Russ


Scaling the RAC

Welcome to RAC Scale, the work in progress blog!

November 28, 2016

Russ here!

Thanks so much for visiting my new blog. As you can likely tell, this site has a long way to go (bare with me), but I wanted to go ahead and get this site published, and the first few posts that introduce myself. I figured that while the site is in its early days, it may make sense to post what my earlier days were like as well.

But first, what is this site all about?

Myself and my friends began making Short Educational Videos around data center technology. They are high energy, fun, and sometimes plain stupid videos that hit different technologies from a high level. The videos have been released to a wonderful reception, but the common feedback we receive is the desire for written content around the videos.

What I hope to do is provide some write-ups and references for those videos, as well as, write about a great number of other topics.

Topics ranging from: General Technology, Cloud, Data Center, Virtualization, Sports, eSports, and the business around these topics. Hopefully you will join in on the conversation as time goes on, and are looking forward to the journey. I know I sure am!

P.S. The name comes from my initials combined with a play on Rack Scale. Russell A Cantwell (RAC) Scale… I know, it’s clever! 😉

Best,

Russ