We’ve been discussing the shift to I/O centric servers for a while, and much of our analysis was validated with HP’s latest server launch.  Dubbed Project Voyager, the new Gen8 servers promise to reduce operation costs, among other things, essentially making the management of servers an idiot-proof task.  Shortly after HP made their announcement at an event in Las Vegas this week, theCube rounded up the founding analysts at Wikibon to break down the marketing talk, taking a deep dive into what Gen8 really means for companies, HP’s business model and the IT job market.

Wikibon co-founder Dave Vellante kicks off the deep dive with a reference to HP’s mention of the data tsunami we’re all dealing with, asking fellow co-founder David Floyer if this is a change in HP’s business model.

“Yes, we’re shifting to unstructured data” Floyer says as adamantly.  “The amount of unstructured data in the servers are changing dramatically.  Moving from storage arrays to more storage in the server means they have to solve the problem of storage in the rack.”

Host John Furrier then runs through the three main pillars HP outlined during the press junket, noting the key areas the new Gen8 servers address.  Pillar 1 focuses on the dynamic workload, while pillar 2 is all about server management.  Pillar 3 covers the data center conditions, such as temperature control.  He then asks Floyer if HP’s done well with these areas of improvement, to which Floyer responds with a drawing out of which details are worth paying attention to, and which merely pump up HP’s latest announcement.

What’s new with Gen8?
Floyer purports that this is really a discussion of storage from an automation standpoint, and much of the other highlights mentioned during the press conference are notably new.  While HP would have you believe they’ve completely re-architectured the server, Floyer insists that they’ve really addressed the complexity of servers in the rack.  This is a core issue for server management, and Floyer goes on to point out the many human errors that could disrupt the ongoing care of a server.  “So [HP is] automating that, trying to simplify, take away the unnecessary, take all that data and make it a closed-loop feedback.”

“The architecture of capturing all this data,” Floyer goes on, “they can take it, give it to their partners and themselves, know what server is where.  This is incredibly useful in being able to actually service things, whether it’s a partner service, remotely or themselves.  They have the right data and the right architecture.   This will allow them, together with the consolidation of server and storage, to keep away from other vendors trying to put in large mainframe environments.”

At this point it’s pretty evident that HP’s making some important changes to how the server stack is run, but where’s the money?  It’s in the operation and services costs.  That’s the cost of maintenance and operations, but HP’s also promising savings in this area, roughly 50%.

The cost-savings behind Gen8

Vellante then pulls out several stats mentioned in the keynote, all listing the areas in operational costs HP’s new Gen8 server will save.  Asking Floyer if these numbers match up with his research, we gain additional insight to how a company must consider the true cost of running an IT department.

“Some are saving, but HP’s mixing apples and oranges,” Floyer says.  “In one part of their announcement HP says they could improve power by 70% per compute.  Well, the power of servers is getting more powerful thanks to Moore’s law, so only about half of that 70% is attributed to Gen8.”

Which brings us to the question of the bottom line.  Floyer presents this using Wikibon’s standard model as a point of comparison, aligning HP’s new product with the expectations of a mid-sized company worth $1 billion.

“If we look at it from the perspective of overall IT budget, my estimate is that this generation will help save around 4% of costs,” Floyer explains.  “If we take servers, they’re around 26% of the budget.  Of that 26% there’s a savings of 15% that can be attributed to Gen8.  Sometimes that will be indirect, with services and facilities a company would have to buy in or outsource.”

The next question Vellante brings up is whether or not Gen8 is a competitive differentiator.  Floyer would tend to agree.  “They put a lot of effort into addressing this,” Floyer starts.  “Other vendors like IBM have a good track record of innovation in this area, but HP has by far the biggest push in reducing operating costs in a greenfield environment by offering only two choices.  Given that HP has so many different servers, this is, in my view, the best.”

A new change in HP storage?

Vellante then brings up the fact that this is HP’s third major storage announcement in the past 100 days, following Project Moonshot and Project Odyssey.  It’s clear that HP’s after a new model here, and Floyer points out that HP’s investing strongly in the area of consolidating storage into servers.  The end result is reduced operational costs, all the while maintaining HP’s market share.  It’s a strong move on their part, inducing a heated rivalry with industry players like IBM, Cisco and Oracle.

Flash has become an important part of this I/O centric shift in server architecture, and the products emerging are heady on the Flash trend.  There was EMC’s recent launch of VFCache, while Fusion-io has been carving a path all its own.  Vellante ponders a single point of control for tomorrow’s servers, and if so, where should that point of control be?

Floyer observes two levels of the server/storage array, one at the active data level and the other for archiving, which can be in different places.  “Clearly big data is the way things are going,” Floyer says.  “The architecture will be top-down, you’ll develop apps together in the servers, bring the data down together so it can be provided in real time, and the management of that has to be much more server-centric, not storage-centric.”

What happens to “disk monkey” jobs?

As the panel comes to a close, Furrier reads a tweet from a SiliconAngle reader, who says “you can’t make the servers too smart.  We don’t want to be out of a job.”  The tweet was tongue-in-cheek, but brings up the very poignant point of server companies having to drastically change the IT environment to meet budget demands of a shaky market, making the server more efficient overall.

“The business value of big data applications, as I/O is taken away as a constraint…I see an increased spend in new applications and new ways of doing things with very high returns,” Floyer says.  “If you’re a disk monkey, then yes, your job is in danger.  But there’s so much opportunity to become partners with the apps to deliver the losest cost, highest availability performance of systems.

“Ask yourself, ‘what can I do to help applications run faster with less down time?’  That’s the way to keep your your job.”