Earlier this week Intel had a major press and analyst event in San Francisco to talk about their vision for the future of the data centre, anchored on what has become in many eyes the virtuous cycle of future infrastructure demand – mobile devices and “the Internet of things” driving cloud resource consumption, which in turn spews out big data which spawns storage and the requirement for yet more computing to analyze it. As usual with these kinds of events from Intel, it was long on serious vision, and strong on strategic positioning but a bit parsimonious on actual future product information with a couple of interesting exceptions.
Content and Core Topics:
No major surprises on the underlying demand-side drivers. The the proliferation of mobile device, the impending Internet of Things and the mountains of big data that they generate will combine to continue to increase demand for cloud-resident infrastructure, particularly servers and storage, both of which present Intel with an opportunity to sell semiconductors. Needless to say, Intel laced their presentations with frequent reminders about who was the king of semiconductor manufacturing.
Server technology - One of the clear themes of the presentations, which Forrester has written about for some time now, was resources tailored to specific workloads, and Intel made prominent mention of several large partners for whom they designed customised versions of their products. My suspicion is that they are far more interested in customised silicon since AMD snapped up two immense (albeit low-margin) deals to ship modified version of their CPUs for both of the leading consumer gaming platforms.
For “smaller” (code word for things that an ARM or low-end AMD CPU might run) workloads, Intel spent considerable time discussing the future Atom SOC products, specifically the Avoton chips which will be released this year as an up to 8-core low power server processor, with a claimed 50 design wins to date for new products.
At the high end, the ubiquitous Xeon was lurking with a full-court press of low power 22nm Haswell parts coming this year. To add some spice, Intel gave out a little teaser about their future 14 nm Xeon SOC, with a hint at a 2014 delivery.
Racks and data centres
Moving further up the stack, Intel spent time discussing its Open Compute Alliance projects, specifically its OCP rack servers and their rack-level photonics interconnect.
The OCP rack servers continue a trend already established by vendor such as IBM and HP of packaging rack servers with shared rack-level infrastructure such as power and cooling. The OCP servers take it step further by also sharing PCIe aggregation and reducing cable connections for connections to external network and storage modules, a step I expect HP, IBM and Dell will also take. The rack-level photonics interconnect was introduced in January at the OCP conference, and consists of four bundled 25 Gb links that can be terminated in a storage module or pretty much anything else, for that matter, but storage is the logical candidate for a big fat pipe that connects to a collection of servers.
While “interesting”, this is just the tip of the iceberg. These photonics interconnects are standard photonic technology, and as such remain expensive and interesting only for connecting large discrete chunks of infrastructure. The real breakthrough will be in the area of silicon photonics, which Intel, IBM and others are pursuing. When the photonics are integrated into the silicon at the chip or SOC level, then we will see an explosion of ultra-high speed and power efficient interconnects that will enable new system architectures. My expectation is that we will begin to see practical silicon photonics in the 14 to 10 nm generations of parts.
Storage and big data
As a technology, storage interests Intel for two reasons; first as a consumer of CPUs, potentially with custom accelerators attached and second because of their IP portfolio around solid state storage as the industry moves increasingly toward flash storage in pursuit of both performance and power efficiency. Big data interests because of the fact that it drives massive purchases of combined storage and servers. Much was made of Intel’s investment in optimised Hadoop and other software, but in the end Intel will make its money on semiconductors. I don’t expect that Intel will ever become a major big data software player, but their efforts to optimize the software stack will be a boost for the whole ecosystem.
HPC, for years a passion of Intel since x86 clusters began to displace traditional HPC systems in the Top 500 benchmark list, was also prominently featured, but this time with a bit of a spin – in addition to legacy HPC, which Intel will continue to pursue, HPC is now the computing back end for big data. This is both a clever bit of marketing spin and also true. Clustered compute architectures for HPC that have simple and non-latency sensitive interconnect requirements are in act similar to many Hadoop and other big data and analytics clusters, and Intel currently enjoys a commanding lead as an infrastructure provider.
This event is well timed on Intel’s part. Legacy IT processing is not an issue. For all practical purposes, Intel owns this space. But the emerging worlds of cloud, big data, and the Internet of things may have some surprises left as they develop. This event allows Intel to highlight its successes and lay out strategy for what will be the fastest growing segments of the infrastructure business, and also ones where Intel may actually face competition from emerging ARM alternatives and an intensely focused AMD, which has put a lot of its muscle behind cloud, mobile and low-power semiconductors and has, as noted, managed to snatch up a couple of highly visible CPU contracts. As noted, these will be low margin business, but they will keep fab volumes high, which can help the yields and costs for other AMD products.
In summary, this was a powerful event for Intel, serving notice that they seem to be continuing their legacy of constructive paranoia, despite their dominant position in the market.
Richard Fichera is a Forrester analyst serving Infrastructure & Operations Professionals.This piece was originally published on the Forrester blog network. Republished with permission.