Main content

Pure Storage enters new era of dataset management with Fusion and AI Copilot

Derek du Preez Profile picture for user ddpreez June 19, 2024
Summary:
As Pure Storage advances its Fusion platform and introduces AI Copilots, this could mean a step change in how enterprises manage their data.

An image of Pure Storage CEO Charles Giancarlo
(Image taken by diginomica)

Pure Storage has been a leader in flash storage arrays for years now, proclaiming the benefits of performance at scale for enterprise buyers. For a long time during this period, however, the cost of hard disk meant that hard disk still won out in certain purchasing scenarios - that is until last year when Pure finally proclaimed that the two had finally converged on price. This was a defining moment in the vendor’s history, having said consistently that this would happen, predicting what it marks a trend towards the end of hard disk sales in the enterprise market. 

And as we gather in Las Vegas for Pure Storage’s annual Accelerate event, there’s been a noticeable change in tone in terms of how the company is thinking about its go to market strategy. Last year the benefits of flash versus hard disk were still the main focus of discussion, pointing to not just total cost of ownership advantages, but also improvements in reliability and power/space efficiency. However, this year during a press conference with media and analysts, the leadership team at Pure spent their time focusing more on the future of data and storage management - with simplicity, accessibility and ease of use all at the forefront. A subtle shift away from the hardware benefits towards the software capabilities. 

It feels like somewhat of a perfect storm in terms of timing. On the one hand the storage market macro trends, as mentioned, have turned in favor towards flash being the ‘obvious choice’ in many respects. And on the other hand, enterprises are thinking about their data strategies as a priority given the intense pressure to invest in new Artificial Intelligence technologies. Storage is going to play a key role in this area, given the heavy duty workloads required for AI. 

And two of the key announcements out today - updates to Pure Fusion and the introduction of a Generative AI Copilot for storage - provide a compelling proposition for buyers thinking about the management of their datasets. As Pure CEO Charles Giancarlo said: 

I think we're really unveiling our new positioning, which I've been talking about for some time, in terms of us providing a data management and storage platform for our customers. 

The future is Fusion

It’s worth outlining Fusion first, as this is the enabler for the Generative AI Copilot and the broader management story. Fusion was announced a couple of years ago, but was focused primarily on greenfield arrays. Pure has now made it so that Fusion is backwards compatible and upgradeable on the previous investments customers have made in Pure arrays. 

So, what is it? Fusion essentially is a management platform that unifies arrays and optimizes storage pools on the fly across structured and unstructured data, on premise, and in the cloud. Now fully embedded in the Purity operating environment, the next generation of Fusion will be available to all global customers. 

Giancarlo described Fusion as the equivalent platform to how users think about storage platforms in the consumer world: 

For all of us, as consumers, we've moved on to so-called cloud based storage, if you need more, you just pay for that. And you can share it, share it with others, which makes it much easier to be able to share files and be able to get access to things that are not on your own computer. Well this is what Fusion does for the enterprise. We will be releasing software which is backwards compatible, upgradable. 

Now arrays will be networked with one another, you will be able to share capacity, you will be able to share performance. And customers will be able to set up policies whereby the data that is stored in those environments can be non-disruptively moved, or what’s called ‘load balanced’ across their environment. 

There are a couple of benefits to this, as Pure sees it. Firstly, instead of buyers looking at individual arrays and managing them individually, Pure customers can effectively look at their arrays as private cloud storage. They don't need to know what application an array is on, or where it is, but they can set up policies so that those applications can be communicating with a ‘storage class’. And those policies take care of managing that storage class. 

What this means, in turn, is that the application always gets the performance and the capacity that it requires, according to Pure, even if that means moving things around. But other benefits include things like geofencing - so that certain data can stay within certain country boundaries, if needed. 

Equally, Fusion also means that developers get a new level of self-service. A gripe amongst developers is that if they are developing a new application, and they need storage, they often have to go to their IT department to provision it alongside their performance and capacity requirements. Giancarlo adds: 

In some cases they can set it up by going into their management system, but often they have to acquire new storage capability and set it up specifically for that. With Fusion, IT can now set up a set of APIs, storage classes, and developers - just by programming their own applications - can use an API for a storage class and it's provisioned immediately. So again, it provides cloud-like services into the enterprise environment. 

Complementary outcome

So understanding that Pure Storage now has a unified management platform across all of its arrays with Fusion, this in turn has implications for the vendor’s AI ambitions on its platform. As Shawn Hansen, VP of Pure’s Core Platform Business Unit, said: 

The advantage of having a single unified operating system and a single unified control plane is clean metadata. We generate hundreds of petabytes of metadata every year, across tens of thousands of customers. And so as we began building AI inside, we asked ourselves, what are the complex kinds of questions that we can ask that you cannot ask of a single box?

And this is where Pure’s Generative AI Copilot comes in. Dubbed as the ‘first’ AI copilot for storage, the generative AI tool provides new opportunities for enterprises to manage and protect their data using natural language. Hansen said: 

It's not possible without having the kind of simplicity that Pure has in place. It's a radically new way of managing large fleets of storage using natural language. We think this will make a ‘decent’ storage admin ‘good’. And it’ll make a ‘good’ storage admin ‘amazing’. 

The idea is that because of Fusion and the collected metadata, customers will be able to ask the Copilot questions that they used to have to go and ask customer support - for example: why is there a latency problem? This is particularly difficult to root out when there’s a problem at scale: 

A standard, regular storage administrator will struggle to be able to find that problem at scale. And we have that information. 

So now you can simply ask the question: my Oracle database is slow, here's my volume ID, what's going on? AI does an amazing job of looking across billions of rows of information - and now in a short amount of time will be able to find that problem. 

So we will automate some of the most difficult needles in a haystack to find. These are complex performance problems. These are difficult security issues. 

Down the line Pure expects that it will also be able to show customers how they rank against other companies in their industry. So, for example, if there are 1,000 customers in your sector and your organization ranks 265 against 1,000, the Copilot will also be able to tell you how to improve your ranking by 10% and give you the exact steps to go and do that. Hansen said: 

We think that the idea of an AI copilot at scale fundamentally changes the way that you simplify managing storage. Instead of having to search through 10 different screens, you can ask the question in your own language. We've had the ability for a while to read the information, but with Fusion we can change the configuration, we can automate across the entire environment. 

My take

Last year I wrote that Pure Storage was hoping to make enterprise storage strategic with its platform and offering. And this year it feels like that vision has more meat on the bones, thanks to the updates to Fusion and the explanation of its AI Copilot vision. Pure Storage has always had business priorities of making storage easy to use, accessible and cost effective. With enterprise buyers facing an uphill challenge of getting to grips with their data environments, thanks to advancements in AI, the infrastructure running these environments becomes a top priority. Understanding a latency problem, as one example, when the issue could be occurring across any number of arrays isn’t an easy challenge to solve today. Running logs on various arrays and trying to understand the issue is time consuming and often not accurate. If Pure’s AI Copilot, combined with the use of Fusion, can provide that information at speed - that’s a very helpful resource for companies. Providing automated fixes, or at least guidance on how to carry out a fix, could well be the next step. It’s only day one of Accelerate, so I’m sure there will be plenty more to come, but it’s been a compelling start and an interesting shift in gears for Pure Storage so far. 

Loading
A grey colored placeholder image