NetApp E-series: Not part of the big message, but here to stay, says CEO


Everyone knows the “these go to 11” scene from Spinal Tap.

Another great moment is when the band gathers around as a US radio station plays one of their old numbers. Looks of pride disappear from their faces as the presenter declares the band to be “currently residing in the where-are-they-now file”.

Also in the where-are-they-now file at NetApp’s Insight 2024 event, it seems, was the company’s E/EF-series high-performance storage. Its problem is it just doesn’t fit with NetApp’s message of Ontap as a single operating system (OS) for everything.

It was a big theme at this week’s event in Las Vegas, not least because the company announced – and went very big indeed in messaging terms – new artificial intelligence (AI)-focused data discovery, classification and management capabilities for Ontap.

And no wonder. It’s a powerful set of features that will help curate AI datasets in customer workloads. 

But it also felt like the E-series was akin to an embarrassing relative who didn’t quite fit in. It runs a totally different operating system from most NetApp products and can’t share in the new developments made in the Ontap platform.

Mentioned by some executives in presentations, all knowledge denied by others, it started to look like the E-series might be at risk, and perhaps NetApp wasn’t being entirely candid about the product or its future. Are its days numbered? Or will it be replaced by something in the Ontap camp?

We set out to find out, and found NetApp committed to the E-series, as we shall see. But it took some digging. 

E-series arrays arrived when NetApp bought Engenio in 2011. They ran the SanTricity operating system and were spinning disk only, with flash added later as NetApp (a flash sceptic at first) adapted them for its first foray into flash storage in 2013.

Today’s EF-series are built for all-out speed, multi-petabyte scale and targeted at high-performance computing (HPC) workloads, which include AI use cases. 

But where does the E-series fit, given the overwhelming NetApp message based on Ontap as the single, unified data management platform across most of its product lines?

First, we interviewed Jonsi Stefansson, NetApp’s cloud chief technology officer, who was candid about the inconsistencies in messaging.

How does E-series fit with the messaging of Ontap as a unified data management platform aimed at AI workloads?

E-Series is great from a performance perspective, but it doesn’t have any of the data management. And the biggest issue in today’s world, with AI, is all these silos being created that have no connectivity to the enterprise or business-critical workloads. I compare E-Series to Vast Data or DDN, you know, sort of dumb storage devices. 

But when E-series gets used in AI workloads, how do things play between E-Series and Ontap? 

It doesn’t. So we are almost guilty of creating a silo there. In all honesty, it’s a great, powerful, performant storage layer. E-Series is fantastic on that.

You can use something like CloudSync or XCP to copy the data over. But is that really what you want to do? No, you want to bring the right datasets at the right moment to the AI environment. And then you want to be able to optimise it for cost. 

There is no [cloud connectivity] in E-Series. That’s the problem … but that’s the reality of it.

Where does the incompatibility lie between Ontap and E-series? 

For example, if you’re moving and migrating, then you’re dealing with a very heavy data gravity issue.

You know, if you want to do [NetApp copy tool] XCP or XCOPY or whatever over the wire, it’s going to take a long time. And there is no continuous synchronisation, or you have to buy some synchronisation tool or third-party tool to do this for you. But it doesn’t retain deduplication, compression and encryption during transfer.

So Ontap-to-Ontap is always the best setup you have.

We do not talk a lot about E-Series. And that is because we are kind of breaking our own word by saying, hey, unified storage across everything and that means Ontap across everything. 

Do you think E-Series will be replaced by something more in the Ontap area? 

That’s not my decision. But no, I think [E-series] has a purpose for HPC workloads [where] you don’t need any of the data management capabilities.

Should E-Series customers running workloads on it have any concern that it’s a risk? About support from NetApp? 

No. I mean, it has its purpose. E-Series is fantastic. It’s just doesn’t have the enterprise data management capabilities that I think are crucial. I mean, the same thing would apply to DDN or Vast or Pure or anybody.

E-Series will always have a place at NetApp
George Kurian, NetApp

To be fair to Stefansson, he was coming at things from the perspective of a cloud-focused executive with a decent amount of Ontap experience within NetApp.

Later, NetApp CEO George Kurian made things clear, and said the E-series is here to stay within its very specific set of workloads.

What is the future for E-Series? Will it be replaced? What’s the roadmap? Should customers that use E-Series be a bit concerned? 

E-Series will always have a place at NetApp.

It is focused on extreme performance environments where the storage system requires very limited data management, but it’s really, really focused on serving data at a very predictable, very low latency. 

The second is environments where the customers want a dense repository of data, typically backing our object store. So our storage grid products are built on top of the E-Series platforms.

There are environments, for example, where the data management sits at the application layer. So it could be high-performance computing where you’ve got a parallel file system that has all the data management associated with it. Or like in object storage, the data management is in the storage grid software, but what it requires out of storage is just performance and reliability.

So there we have it. NetApp’s E-series is going to stick around. But it doesn’t fit with the company’s single, unified OS messaging.



Source link