I’m sitting here in the blogger lounge at HPE Discover 2016, in London (while also keeping an eye on the news coming from AWS Re:Invent in Las Vegas)… and the only thing that comes to my mind at the moment is something I read on twitter a few days ago: “the software is eating the hardware and the cloud is eating the software”.
More Focus, but…
HPE is much more focused than a year ago (thanks to the split from HP Inc.) and are trying to remain relevant in a world that has been seeing a major change on how datacenters are designed and infrastructures are consumed. They’re trying to remain part of the conversation and bring credible solutions, but it’s quite obvious they are struggling.
Take Synergy for example – a brilliant idea….had it been launched 5/7 years ago. Now, we have the cloud, the “Nutanixes” and all the software-defined-ish stuff you may need… why should someone buy a configurable piece of hardware from a company that doesn’t even own the whole stack and can’t give an iPhone-like experience? If only the rumours were true and Simplivity were the target of an acquisition… that would be something to talk about (and, probably, it would be as good as the 3PAR acquisition has been).
Server? Storage? What else?
Yes, HPE is the worldwide leader for server revenues and shipments. But this is a declining market. It’s not about the quality of the server it’s about the cloud eating up everything.
The biggest infrastructure investments are made in large datacenters now, with SMEs looking to offload as much as they can to the cloud. Large datacenters are all about efficiency, they’re trying to squeeze everything they can from any resource and this is why they are doing more with fewer servers. Even more so, they don’t want any form of lock-in – hyper-scalers are designing their own equipment to take efficiency to the next level but they also want to make hardware vendors irrelevant. In fact, if you own the specification, and it’s an open standard, there is no way the hardware manufacturer will make a difference, if not in the price…
And what about Storage?…it’s even worse. Even though 3PAR is still doing well (in comparison to the rest of external market), external enterprise storage shipments and revenues are seeing trends which are comparable to servers. And so is the cause. Big datacenters look for very large, and cheap, capacity-driven infrastructures (massive scale-out infrastructures) and very low latency solutions (they look for CPU-storage vicinity).
The HPE 3PAR team is working on a very nice/cool and seamless adoption of 3D Xpoint (as everyone else in this industry?!). It will be delivered as a cache first and then as a tier in the future (but isn’t that what happened with Flash adoption a few years back? Yes, it is!). They are still evolving the same platform, which is still good for traditional applications and workloads, but not enough for next generation infrastructures. A 3D Xpoint powered 3Par can give 250 microseconds latency, while a 3DXpoint PCI card can give you 1/10th of that!
Storage as we know it today, will be less and less of interest and less relevant. In-memory (and CPU-driven), ephemeral or persistent storage is the future. Several startups are already working on it, most of them are delivering software solutions (not hardware, software!) And when hardware comes into play, it will be commodity stuff. Yes, again, HPE has shown us a prototype of what they call “The Machine”, but it’s just another science project made out of components that won’t be available for years… while large end users are looking at these problems now.
I won’t comment about networking, not my field of interest, but I suspect it’s not that distant from what I’ve seen in other fields.
And what about the cloud?
Well, the fact that HPE has invested so much in OpenStack and now de-investing just as quickly, is sad. They say they’re going to partner with Suse to build a common platform, the truth is that they’re selling everything to Suse (Helion, Cloud Foundry stuff, engineers and all). And Eucalyptus has disappeared too, although it had some potential to help with this “Hybrid IT” that HPE loves to mention in every speech here… but I wonder if Eucaplyptus could have kept up with the incredible pace set by AWS in terms of innovation, new services and improvements. Having to chase after someone else is always a tough job.
HPE is partnering with everyone but Amazon… and even though one of the partners is Docker (which is everybody’s partner now), I can’t see a lot here… Yes, HPE servers are shipped with Docker on board but I don’t think anyone cares about it.
Closing the circle
HPE is now more focused than in the past. It has a leaner organization, but it’s also clear that a lot has to be done for it to become fancy again… and I certainly wouldn’t want to be the one in charge of having to come up with the right strategy to achieve that…
I’m not saying that HPE will disappear anytime soon, but the difference between a Server from HPE or Dell (or anyone else) is minimal and doesn’t justify all that effort, especially from large datacenter customers… and the same goes for storage and, probably, networking. Everything is moving up in the stack.
Software could make a difference, a huge difference, but HPE has never been as successful as it thinks about software.… And they are totally missing that HCI is 100% software. Embracing the cloud (I mean embracing it seriously) could be too much for HPE, they aren’t ready and they might not be able to cannibalize themselves to become a totally different company (and the IBM story could tell us a lot about this). Options do exist though, and I think that HPE should try to repeat the 3PAR story with Cloud instead of storage this time round (with Simplivity perhaps).
I’m afraid I don’t have a formula or suggestion for HPE…….it’s just that we’re living in a different world now and even though the products they make are good and still at the core of our infrastructures, they are no longer relevant in the ongoing conversations about the future of IT.
Disclaimer: I was personally invited to attend HPE Discover, with HPE covering my travel and accommodation costs. However I was not compensated for my time. I am not required to blog on any content; blog posts are not edited or reviewed by HPE before publication.