Technology revolutions tend to happen in phases. When personal computers came to market in the early 80s most in business were actually used more like smarter terminals that could do some minor things typically done by typewriters (word processing) and calculators (spreadsheets) locally. Eventually they grew to be a full peer to servers and as increasingly mobile devices, laptops, performed alone. In that case hardware changed first and then software caught up.
With the Web, the opposite appears true; first web loads were put on existing hardware and then folks started redesigning the hardware making it more efficient and adding cores and the software around virtualization. It was only a matter of time before a major player decided to try something very different and HP Redstone is that big push.
Now Google is considered the leader in “the cloud” but they have found enterprise sales daunting and their model has run afoul of the law. Microsoft has moved to the cloud aggressively, their Azure platform is better regarded and they were the first of the two vendors to start talking about using something like ARM to redesign the data center.
Of course, this has implications for Intel as well because this may be the beginning of a revolution like the one that established Intel in the first place.
Meg Whitman’s HP – Jumping into the Lead
HP desperately needed to appear like they were capable of executing something other than executive boardroom drama and this fills that need nicely. The back story here is that this represents a strategy shift for HP from Carly’s consumer product focus where the firm lagged with TVs and Leo’s software focus where they lagged Oracle and IBM back to enterprise hardware where they are traditionally one or two in the market. This is the difference between fighting battles on someone else’s turf and fighting them on your own and the odds go up dramatically when you are fighting in an area you know.
As a leader in servers, HP has a much stronger chance to turn the market on their initiative than they ever did with TVs or software. This is HP trying to lead in a market they generally already lead and it is so much better to try to lead from the front than from the back of anything.
This move bodes well for the company.
New Generation for the Web
These new ARM hyperscale servers are designed specifically for web loads like video streaming, or data analytics and even web server hosting. Using technology from partner Calxeda, HP presented that a typical web hosting site with 400 servers on 10 racks using 20 switches and 1,600 cables could be consolidated onto their ARM based Redstone platform which, even though it would require 1,600 servers for the same load, would only take up a half of a rack, two switches and 41 cables. Savings is massive with power requirements dropping from 91 kilowatts to 9.9 kilowatts and costs for the system dropping from $3.3M to $1.2M.
This is impressive savings in a market that is extremely cost conscious right now and it is consistent with savings HP has been talking about on optical switches that are being developed on a different path and will be positioned against Cisco. All told, both efforts could result in up to a 90 percent reduction in power and purchasing cost for a qualifying load and numbers like that move markets.
However, the servers aren’t due to market in limited volumes until next year, likely more than a year out in final form, and the optical switches may be even further out, giving competitors a little extra time to get ready. But this should wake up the market.
Intel’s Problem - AMD’s Choice – NVIDIA’s Opportunity
For once, Intel and AMD have potentially different ideal paths to address this threat. Intel could recognize they have one real choice: do they let an ARM vendor cannibalize their server market or does Intel do it themselves? Given this will typically result in a 4x jump in processor volume even though it will also result in a net loss in revenue and profit, the end result is still vastly better if you are trying to keep your fabrication factories populated than running away from the business. AMD, on the other hand, might find it more interesting to embrace ARM and use their existing server experience to ride this wave rather than fight it. The result could be something far more competitive against an Intel solution than anything they have ever before brought to market.
Both AMD and NVIDIA have been positioning their graphics technology for Supercomputers and both could likely do an interesting ARM hybrid offering to expand these servers outside of web services and into other more processing intensive segments. The disruption provides an opportunity for both NVIDIA and AMD to move against Intel on a more level playing field though servers remain Intel’s market to lose.
A cautionary tale for Intel is that Steve Jobs wanted the iPad to run on Intel and felt that Intel refused to step up so went to ARM. Markets turn on decisions like this and another miss would be painful.
Wrapping Up: Google and Microsoft
Microsoft’s last decade has been plagued by ideas that they first had that Apple made better use of. They had the idea of a hard drive based portable media player before there was an iPod, they had two shots at tablets before there was an iPad, a they even had folks inside working on an iPhone like product long before Apple launched the iPhone. Google stomped them with Android and some think Google will win here as well. Microsoft first showcased the idea of a data center run by low cost processors years ago ironically using Intel Atom chips.
The question to be answered is whether they or Google will have a solution in the market that addresses this first. Windows 8, which shares code with Windows server, is compatible with ARM but most of the existing ARM code is Linux based as is Android and the Chrome OS. There is even a chance that a third vendor could use this transition to bypass both Google and Microsoft.
In the end, this week, HP gave the server, processor, and platform vendor’s one hell of a wakeup call. Now we get to see who was actually awakened. Runners take your mark, the race started some time ago…
This one little thing could win or lose wars cutting across every major vendor in technology. That’s how a leader leads and it’s nice to see HP step up one more time. Rob Enderle is President and Principal Analyst for the Enderle Group. To read more of his articles on TechZone360, please visit his columnist page.
Edited by Rich Steeves