How will Fusion and Portworx Data Services impact the future direction of Pure Storage?

Pure Storage has announced two new services at Pure//Launch. What are the implications for the company and the industry as Pure further adopts the “as a service” model for solutions delivery to the enterprise?

Background

Pure Storage has announced two new services at the Pure//Launch event on 28 September 2021. Fusion is a SaaS management plane for the administration of Pure Storage hardware and software resources and Portworx applications. Portworx Data Services (also delivered as SaaS) extends the capabilities of PX-Enterprise to automate and manage the deployment of popular relational and NoSQL database solutions.

What will these SaaS solutions offer, and how do they compare to the market?

Storage Resource Management

History is awash with the attempts of vendors to build practical and efficient SRM (storage resource management) tools. Almost all have failed, typically because they try to “boil the ocean” by supporting all vendor hardware or are overly complex to install and manage. Most SRM solutions risk becoming the “tail wagging the dog” as the ability to upgrade primary storage becomes dependent on the capability of the SRM tool to support new platform features and functionality.

The public cloud has provided storage vendors with an excellent opportunity to solve the issues of SRM software while implementing significant value-add in a re-invention of the definition of SRM itself. The beginnings of this trend have been in place for nearly a decade, with the introduction of cloud-based analytics. Pure Storage introduced Pure1 in 2015, with Pure1 META adding deeper analytics in 2017. Nimble Storage introduced Infosight in 2013 and was acquired by HPE in 2017. HPE subsequently rolled out the Infosight platform across its storage and server portfolio.

NetApp has collected telemetry data since the genesis of ONTAP, initially through AutoSupport and now with ActiveIQ. Dell Technologies platforms collect and analyse data through CloudIQ. All these solutions demonstrate value-add by using the collective “wisdom of the storage crowd”, typically enabled by machine learning algorithms.

Proactive

Monitoring storage platforms is a relatively benign aspect of storage management, whereas proactive optimisation is an entirely different proposition. However, the automation of workload placement is not without precedent. When VMware introduced DRS (and Storage DRS), the idea of the platform deciding the right location for virtual machines within a cluster (and making the move dynamically) was quite a leap of faith. Today, we don’t question workload placement, either on-premises or in the public cloud. The adoption of containerised applications and Kubernetes pushes this paradigm one step further, with little or no control of workload placement available to the user. In systems that work well, this evolution is a good thing, as it releases the operations teams to do more valuable work.

Fusion

Pure Fusion implements the proactive storage operations and optimisation tasks as a management plane in the public cloud. As a SaaS offering, the requirement to upgrade is outsourced to Pure Storage. The customer only needs to provide API access for the supported platforms (a feature already in place with Pure1). It’s unlikely Pure Fusion will ever support third-party storage, so upgrades to the Fusion functionality can easily be kept in lockstep with releases of Purity (the FlashBlade/FlashArray operating system).

All the challenges of scalability are removed with a SaaS solution, as the vendor handles these tasks. SaaS management has the benefit of using knowledge from the entire install-base of systems when deciding how to optimise resources.

Policy

Remote platform management is a helpful tool for vendors and customers alike. However, the more significant benefits are gained through the ability to apply business-specific policies to data placement. This process can mean optimising for cost, performance, capacity, or availability. Policy-based management also enables the automated protection of data, implementation of ransomware features and data mobility.

The transition to policy-based management is the biggest step forward from traditional SRM functionality. It provides the degree of abstraction needed that breaks the physical link to the hardware. As a result, customers can move to a consumption-based billing model, leaving Pure Storage to decide on the most optimum hardware to meet customer requirements. This capability is essential because it enables Pure Storage to address other customer challenges, such as rack density, power consumption and meeting green agendas.

Hardware

While the move away from hardware specifics is important for the customer, the capabilities of the hardware are still critical to delivering a cloud-like experience. Pure Storage has built features into FlashArray and FlashBlade that offer seamless upgrades and capacity increases. ActiveCluster enables data to be moved around a metro complex of arrays to optimise data (for example) between FlashArray//X and FlashArray//C hardware.

The move to consumption-based on-premises deployments of storage cannot be efficiently delivered without features that dynamically and seamlessly rebalance data across the physical estate of hardware.

Storage vendors that haven’t modified their solutions to be cloud-friendly are either leaving resources unused in the data centre or compromising their own margins. At worst, some vendors could require customer involvement and scheduled downtime to deliver the optimisation process.

Portworx Data Services

As we’ve explained in previous blogs and podcasts, we believe that container-attached storage will become an abstraction layer from physical storage in the same way that datastores provided pooled capacity for VMware virtual instances without continual interaction with the storage platform.

Pure Storage has already started the Portworx/FlashArray/FlashBlade integration with features introduced into PX-Enterprise 2.8. Portworx Data Services provides further value-add to this process by enabling customers to deploy and operate managed databases within a Portworx cluster.

Much of the groundwork for this new feature has been in place for some time. Portworx has comprehensive documentation on how to deploy popular database solutions with the PX-Enterprise platform. Portworx Data Services takes this process one step further by automating the process through a SaaS portal.

Developers now have access to complex storage constructs in the form of databases through an API call in the same way as block, file and object resources are provisioned. Portworx Data Services essentially makes databases a “first-class citizen” and aligns with capabilities already provided by the public cloud vendors.

Detail

The success of Portworx Data Services (PDS) will be in both the depth and efficiency of the managed services offered. Initially, PDS will offer 1-click deployment, sizing, and availability management. Data protection will be integrated through PX-Backup. Database images are taken from standard repositories such as Docker Hub. However, this needs to be expanded to include custom-builds where required.

The Architect’s View™

The introduction of Pure Fusion is an essential step forward to delivering a hybrid cloud experience for Pure Storage customers. Other vendors are already on this journey. HPE introduced similar functionality to Fusion earlier this year with the announcement of the Alletra platform and Data Services Cloud Console (DSCC). Alletra essentially obfuscates the underlying hardware platform, while DSCC automates deployment and management. Dell Technologies has introduced APEX, although we’re yet to see whether this service will offer full platform automation or is just a portal for storage purchases. NetApp has a comprehensive cloud portal with direct integrations into public cloud platforms.

The success or failure of any of these solutions will be dictated by the depth of features and the ability to make on-premises consumption-based storage as (close to) seamless as the public cloud. This is a big task for any vendor, however, Pure Storage seems to have all the pieces in play to make this work.

Portworx Data Services appears, at this point, to be an outlier in the market, with no other equivalent offering. The nearest similar solution is perhaps Amazon ECS Anywhere, which offers the capability to run ECS and applications on customer-owned infrastructure.

The key value proposition for Portworx Data Services is not the automation part of the deployment process but the data management capability. Data will far-outlive the lifetime of the application delivery package (whether a VM or container). As such, Portworx and the data services offered are in a prime position to be the data custodian for the customer, delivering deployment, protection, and ongoing management of the data resources.

The long-term prize for Pure Storage must be the data management of the customer’s most valuable asset, whether on-premises or in the public cloud. Data is perpetual, while everything else around it is ethereal. We will be assessing Pure Storage data management capabilities versus the rest of the industry as the storage management discussion transitions to one of data management.