Many in the storage industry look to Pure Storage to lead in all-flash innovation through its flagship FlashArray product line. In that respect, Pure Storage certainly delivered at its Accelerate conference held this past week in Austin, TX. It announced increased capacity, storage class memory, and analytics for its FlashArray product line.
Yet its most notable concerns its Cloud Block Store announcement. On the surface, this may look like a “me-too” offering. It is anything but. Rather, it represents a “me-first” moment for Pure Storage. Cloud Block Store creates a new benchmark for measuring how vendors deliver their storage volumes in the cloud.
Notable Pure Storage Announcements
- Its new FlashArray//C Capacity Storage line of capacity-centric all-flash arrays. These arrays will use TLC and QLC flash to deliver high levels of capacity in a small footprint. Organizations should use //C arrays for secondary workloads such as application tiering, backup targets, and policy-driven replication.
- DirectMemory Cache. Pure Storage brings a new Tier 0 tier of flash to its FlashArray//X storage line. The DirectMemory Cache uses Storage Class Memory (SCM) to achieve latencies as low as 100 microseconds. Notably, organizations may non-disruptively introduce this class of storage into its existing FlashArray//X arrays.
- Pure1 VM Analytics. This software tool increases retention of analytics data from 7 days to up to three years. This better equips organizations to perform capacity and load forecasting, plan for workload growth, and do migration scenario planning.
Yet among these announcements, Pure Storage’s Cloud Block Store announcement caught my attention. On the surface, it seemed routine: Pure Storage now offers a FlashArray storage volume in the AWS cloud. This gives it a “me-too” component as both NetApp and VMware already deliver similar functionality in the AWS cloud.
The Me-first Component of Cloud Block Store
One only observes the “me-first” component of Cloud Block Store when one examines its underlying architecture. Its competitors deploy cloud-based storage volumes in the AWS cloud on Elastic Compute Cloud (EC2) instances that use Elastic Block Store (EBS). Organizations may then attach these storage volumes to their cloud applications. Using these EBS-based cloud storage volumes, they get the same storage management features in AWS that they obtain on-premises.
The Pure Storage Cloud Block Store architecture follows this approach to a point. It also deploys its Purity software on EC2 instances. However, its Cloud Block Store architecture differs in a substantial way from how its competitors deploy their respective cloud volumes.
Pure Storage wanted to deliver the same levels of availability and redundancy in the AWS cloud as it does with its on-premises storage arrays. To accomplish this, Pure Storage did the following.
It created what it refers to as a virtual shelf. This virtual shelf consists of seven EC2 instances each with its own NVMe-powered instance stores. Each of these seven EC2 instances functions as a virtual disk drive. Together, they function as a single virtual storage shelf.
Taking this approach, Pure Storage Cloud Block Store adheres to the practices associated with AWS’ placement groups. Each instance resides on fault domain isolated hardware. In the event one of these EC2 instances fails, the virtual shelf itself remains up and operational.
The two EC2 instances running the Purity Software stripe the data across these seven EC2 instances in the virtual storage shelf. This configuration becomes a highly available virtual storage volume that organizations may use and present to any EC2 instance in the AWS cloud.
In the event one of the two EC2 instances running the Purity software fails, the application remains unaffected. Within the Cloud Block Store volume, the other EC2 instance in this grouping takes over the application workload to provide high availability. Further, both EC2 instances with the Purity Software connect to the virtual storage shelf to provide high availability within an AWS availability zone.
Data Durability in Cloud Block Store
Pure Storage extended the capabilities of its Cloud Block Store in yet another innovative way. Pure Storage recognized that companies often use the cloud for test and dev. This use case may prompt companies to start and stop applications in the cloud or even turn them off. This can result it the data in the application’s underlying storage volumes being deleted even when they may need that data again.
To preserve this data in the event they turn the application on again, Cloud Block Store copies the data to AWS S3. Copying the data to S3 provides three distinct benefits:
- Data durability. S3 has eleven 9’s of data durability. By copying all data stored in the Cloud Block Store volume to S3, organizations can have a very high degree of confidence about the data’s durability.
- Data recoverability. Should an organization need to turn an application off for a time, it can recover the data copied and stored in S3. Cloud Block Store automatically performs this task by recovering the data from S3. This shortens the time to restore or recreate the data since Cloud Block Store has high bandwidth connectivity to S3.
- Cost-effective. Data stored in S3 is one-tenth or less of the cost of storing data on EBS. During the time an application is turned off, companies retain data they may need again at a fraction of the cost. During the time the application operates, the S3 cost overhead is nominal for the benefits this approach offers.
Cloud Block Store: Simple Data Availability, Durability & Performance in the Cloud
Organizations are told they get all the same benefits in the cloud as they do on-premises. In many, if not most cases, this premise holds true. Yet with block storage volumes in the cloud, this was not the case. Organizations would have to sacrifice availability, durability, and performance when hosting mission-critical applications in the cloud. This often led them to keep these applications on-premises because of this cloud shortcoming.
Pure Storage’s Cloud Block Store addresses these long-standing data availability, durability, and performance issues for block storage in the cloud. Pure Storage Cloud Block Store creates the equivalent of a virtual storage array in the cloud using readily accessible cloud resources. In so doing, organizations get the block storage experience they expect the cloud to deliver. More importantly, Pure Storage delivers it in a manner in which organizations want to consume it in the cloud, simply and with the click of a button.