Founded in 2011 by Facebook, Intel and Rackspace, the OCP’s official mission is to apply the benefits of open source to hardware and to ‘rapidly increase the pace of innovation in, near and around the data centre and beyond’. Heady stuff.
Whilst the OCP membership includes what might be termed the ‘usual suspects’ – companies such as Facebook and Google, who are looking to improve data centre performance – there are also companies who, you might think, would be looking to protect their market position against open source.
The recent OCP US Summit saw some interesting announcements. But the one which has caught almost everyone’s eye was that of Microsoft, which is using Qualcomm’s ARM based Centriq server processor in Project Olympus. This project is looking to create modular and flexible hardware for cloud computing and is said to be attempting to apply to hardware the same open source principles found in the software domain.
ARM recognised the potential of the data centre some years ago and essentially laid its cards on the table with the launch of the 64bit ARMv8 architecture.
That was picked up by a number of companies, including Qualcomm, which announced in November 2014 plans to develop ARM based chips targeted at data centres. The first such device – Centriq 2400, featuring 48 Falkor cores, developed using its ARMv8 architectural license – came in December 2016.
While Qualcomm’s deal with Microsoft won’t see ARM based chips storming the barricades, it marks a pretty big step forward for the ARM architecture in a sector where Intel has dominated.
While Qualcomm and ARM got most of the attention, Project Olympus featured another interesting name – AMD, with its Naples processor. AMD has been the poor relation over the years, but industry observers believe Naples could provide serious competition for Intel’s Xeon processors. And one of the main areas of competition will be price.
Inevitably, with ARM and AMD getting higher profiles in data centre technology – particularly when backed by Microsoft’s muscle – people have begun to wonder whether Intel might just start to lose its stranglehold. The answer to that one is ‘don’t bet the house on it’. There’s a very good reason why Intel bought FPGA company Altera and the benefits of that deal will be seen in such areas as search acceleration in data centres.
A saying from years ago was ‘nobody gets fired for buying IBM’. You could adapt that to buying Intel for data centres. But one of the ideas behind open computing is ‘horses for courses’. While Intel chips will continue to handle particular tasks better than other technologies, open computing provides the opportunity for the ARM architecture and for companies like AMD to get a bigger slice of the action.
