Why did anyone think HP was in it for public cloud?
Many have jumped on a recently published interview with Bill Hilf, the head of HP’s cloud business, as a sign HP is finally coming to terms with its inability to make a dent in Amazon’s public cloud business. But what had me scratching my head is not that HP would so blatantly seem to cede ground in this segment – but why many assume it wanted to in the first place.
For those of you that didn’t see the NYT piece, or the subsequent pieces from the hordes of tech insiders and journalists more or less towing the “I told you so” line, Hilf was quoted as candidly saying: “We thought people would rent or buy computing from us. It turns out that it makes no sense for us to go head-to-head [with AWS].”
HP has made mistakes in this space – the list is long, and others have done a wonderful job at fleshing out the classic “large incumbent struggles to adapt to new paradigm” narrative the company’s story, so far, smacks of.
I would only add that it’s a shame HP didn’t pull a “Dell” and publicly get out of the business of directly offering public cloud services to enterprise users, which was a good move. Standing up public cloud services is by most accounts an extremely capitally intensive exercise that a company like HP, given its current state, is simply not best positioned to see through.
But it’s also worth pointing out that a number of interrelated factors have been pushing HP towards private and hybrid cloud for some time now, and despite HP’s insistence that it still runs the largest OpenStack public cloud – a claim other vendors have made in the past – its dedication to public cloud has always seemed superficial at best (particularly if you’ve had the, um, privilege, of sitting through years of sermons from HP executives at conferences and exhibitions).
HP’s heritage is in hardware – desktops, printers and servers, and servers still present a reasonably large chunk of the company’s revenue, something it has no choice but to keep in mind as it seeks to move up the stack in other areas (its NFV and cloud workload management-focused acquisitions as of late attest to this, beyond the broader industry trend). According to the latest Synergy Research figures the company still has a lead in the cloud infrastructure market, but primarily in private cloud.
It wants to keep that lead in private cloud, no doubt, but it also wants to bolster its pitch to the scale-out market exclusively (where telcos are quite keen to play) without alienating its enterprise customers. This also means delivering capabilities that are starting to see increased demand among that segment, like hybrid cloud workload management, security and compliance tools, and offering a platform that has enough buy-in to ensure a large ecosystem of applications and services will be developed for it.
Whether OpenStack is the best way of hitting those sometimes competing objectives remains to be seen – HP hasn’t had these products in the market very long, and take-up has been slow – but that’s exactly what Helion is to HP.
Still, it’s worth pointing out that OpenStack, while trying to evolve capabilities that would whet the appetites of communications services providers and others in the scale-out segment (NFV, object storage, etc.), is seeing much more takeup from the private cloud crowd. Indeed one of the key benefits of OpenStack is easy burstability into, and (more of a work in progress), federatability between OpenStack-based public and private clouds, respectively. The latter, by the way, is definitely consistent with the logic underpinning HP’s latest cloud partnership with the European Commission, which looks at – among other things – the potential federatability of regional clouds that have strong security and governance requirements.
Even HP’s acquisition strategy – particularly its purchase of Eucalyptus, a software platform that makes it easy to shift workloads between on premise systems and AWS – seems in line with the view that a private cloud needs to be able to lean on someone else’s datacentre from time to time.
HP has clearly chosen its mechanism for doing just that, just as VMware looked at the public cloud and thought much the same in terms of extending vSphere and other legacy offerings. Like HP, it wanted to hedge its bets stand up its own public cloud platform because, apart from the “me too” aspect, it thought doing so was in line with where users were heading, and to a much more minimal extent didn’t want to let AWS, Microsoft and Google have all the fun if it didn’t have to. But public cloud definitely doesn’t seem front-of-mind for HP, or VMware, or most other vendors coming at this from an on-premise heritage (HP’s executives mentioned “public cloud” just once in the past three quarterly results calls with journalists and analysts).
Funnily enough, even VMware has come up with its own OpenStack distribution, and now touts a kind of “one cloud, any app, any device” mantra that has hybrid cloud written all over it – ‘hybrid cloud service’ being what the previous incarnation of its public cloud service was called.
All of this is of course happening against the backdrop of the slow crawl up the stack with NFV, SDN, cloud resource management software, PaaS, and so forth – not just at HP. Cisco, Dell, and IBM, are all looking to make inroads in software, while at the same time on the hardware side fighting off lower-cost Asian ODMs that are – with the exception of IBM – starting to significantly encroach on their turf, particularly in the scale-out markets.
The point is, HP, like many old-hat enterprise vendors, know that what ultimately makes AWS so appealing isn’t its cost (it can actually be quite expensive, though prices – and margins – are dropping) or ease of procurement as an elastic hosting provider. It’s the massive ecosystem of services that give the platform so much value, and the ability to tap into them fairly quickly. HP has bet the farm on OpenStack’s capacity to evolve into a formidable competitor to AWS in that sense (IBM and Cisco also, with varying degrees, towing a similar line), and it shouldn’t be dismissed outright given the massive buy-in that open source community has.
But – and some would view this as part of the company’s problem – HP’s bread and butter has been and continues to be in offering the technologies and tools to stand up predominately private clouds, or in the case of service providers, very large private clouds (it’s also big on converged infrastructure), and to support those technologies and tools, which really isn’t – directly – the business that AWS is in, despite there being substantial overlap in the enterprise customers they go after.
However, while it started in this space as an elastic hosting provider offering CDN and storage services, AWS, on the other hand, has more or less evolved into a kind of application marketplace, where any app can be deployed on almost infinitely scalable compute and storage platforms. Interestingly, AWS’s messaging has shifted from outright hostility towards the private cloud crowd (and private cloud vendors) towards being more open to the idea some enterprises simply don’t want to expose their workloads or host them on shared infrastructure – in part because it understands there’s growing overlap, and because it wants them to on-board their workloads onto AWS.
HP’s problem isn’t that it tried and failed at the public cloud game – you can’t really fail at something if you don’t have a proper go at it; and on the private cloud front, Helion is still quite young, as is OpenStack, Cloud Foundry, and many of the technologies at the core of its revamped strategy.
Rather, it’s that HP, for all its restructuring efforts, talk of change and trumpeting of cloud, still risks getting stuck in its old-world thinking, which could ultimately hinder the company further as it seeks to transform itself. AWS senior vice president Andy Jassy, who hit out at tech companies like HP at the unveiling of Amazon’s Frankfurt-based cloud service last year, hit the nail on the head: “They’re pushing private cloud because it’s not all that different from their existing operating model. But now people are voting with their workloads… It remains to see how quickly [these companies] will change, because you can’t simply change your operating model overnight.”