ARM Community: Server Ecosystem: A Fine Kettle of Fish - ARM Community

What would servers would look like in the future if they utilized low power, lower performance system-on-chip devices? This reminds me of a conversation I had about eighteen months ago, when I was talking to a company in Taiwan about ARM’s thoughts in the server area.

They told me...
1. The processor has to be 64-bits
2. We must immediately support virtualization software to enable the CPU compute on hand to be divided up across many tasks
3. We must deliver offer QPI-based solutions to support off-chip coherency

Hmmm. I am 0 for 3! So should I just fold up my tent and go home?

Well, ARM firmly believes the answer is no. Clearly, there are server application workloads where the three ingredients outlined above are absolutely the right approach. But I challenged this person (without much success actually) to think not about how servers exist today, but what ultimate low power could enable for servers. If a workload can run on a 32-bit incredibly efficient processor such as the Cortex-A9 or the Cortex-A15 there will be meaningful power and cost savings over a 64-bit platform. One of our partners refers to this as “32-Bit and Proud!”

A year ago, when we were outlining our initial software plans for deployment on 32-bit ARM processors, my suggestion of Ubuntu was met with various levels of derision. They were seen as a very small player. But fast forward to today, and the release on October 13^th of the Ubuntu Server product running on ARM, things have changed. This release has received some good coverage, and with IBM and Dell platforms gaining Ubuntu certification and HP’s recent announcement that Ubuntu will be its primary OS running its Cloud Service, this Linux distribution has improved its profile and credibility.

Based on feedback I got to my initial server blog, I am going to endeavor to incorporate an English expression that ties to the key theme of the blog. So for this one, I picked the phrase “a fine kettle of fish”. The exact history of this one is a little unclear, but it basically means a chaotic situation. You can liken it to the US phrase “can of worms” I guess.

The reason I picked this one, is that it hints at disruption. And we see exactly this in the server value chain right now for emerging cloud-based applications. As I noted above, traditional servers are different. I can cover that another day. Many cloud based services are based on Linux and use a mixture of open source application, internally-developed applications and of those, an increasing number of which are Java based. Some Taiwan Original Design Manufacturers (ODMs, like Quanta, Wistron and Mitac) are seeing this as an opportunity to deliver technology directly to end users. On the other hand, Original Equipment Manufacturers (OEMs, like HP, IBM and Dell) have a great opportunity to innovate their systems. How could a server look and behave? It does not need to look like it does today.

ARM’s business model is fundamentally about licensing the same processor technology to many companies in order to save those companies development time and cost. This allows the silicon companies to focus on areas of value added technology outside the processor core itself. Additionally, we believe about choice. We want OEMs to be able to select from multiple silicon components and from multiple software solutions. We want end users to have a range of platforms from which they select the one that best fits their specific needs. This path has served us well for twenty-one years in the mobile industry and we wish to bring those exact benefits to the server arena. We actually think that this market will go in a similar direction to what we have seen in the mobile market, where the devices will become customized to perform specific server functions or classes of functions. You have seen from this week’s announcement of the Cortex-A7 processor, and the announcement of the big.LITTLE processing concept that the drive to deliver increased functionality inside a tight power envelope requires a solution other than simply more gigahertz. I see future server devices looking like advanced heterogeneous computing nodes with a mix of general-purpose compute, application-specific compute and hardware-accelerated engines. Of course, time will tell if this vision turns out to be accurate.

Our EVP of Sales challenges his staff to think about whether they are leaping out of the shower in the morning to start their work day. Life is short and if you are not passionate about what you are doing, go find something that does do it for you. I am excited about my job. Some people say that it is because ARM has 0% market share in the server market today so anything I do is on the upside! However, my excitement is in imagining what types of platforms will exist in a few years time as companies, empowered by choice and limited only by their own imagination get creative again in the server domain.

More soon…

More blogs on ARM Cortex-A7

• Combining large and small compute engines - ARM Cortex-A7
• Mobile benefits of ARM Cortex-A7
• The ‘big’impact of ‘LITTLE’ processing to electricity bills everywhere

Ian Ferguson, VP Segment Marketing, ARM, has spent years fighting from the corner of the underdog. Most of those scars are healing nicely. Ian is particularly passionate about taking ARM technology into new types of applications that do not exist or are at the very formative stages. After driving ARM’s server program for five years, Ian now leads ARM’s vertical marketing organizations, that supports ARM’s partners to grow their business across a wide range of applications with a view to reinvent the way the server function is implemented in networks as opposed to simply replacing incumbent platforms.