This is the time of year to take stock in where high-performance computing (HPC) sits and where it is headed. That's because the SC09 conference is taking place in Portland, Ore., this week and it's the biggest HPC conference around.
SC is an odd duck as conferences go. Last year it had more than 10,000 attendees and, yet, it's a largely volunteer-organized event in a world where trade shows of this scope are packaged by conference specialists or some specific corporation. Think the much-renamed LinuxWorld (run by IDG) or VMworld (run by VMware).
"SC" comes from supercomputing. Today's large computer complexes are typically not supercomputers in the sense of a specialized architecture only suitable for a specific type of technical computing. Rather, as Ashlee Vance notes in The New York Times, "The supercomputing world was long dominated by systems that required specialized chips, memory systems and networking technology. But about 10 years ago, researchers realized they could link thousands of cheaper machines running on mainstream chips and achieve pretty solid performance."
Thus an HPC event is no longer about supercomputers per se (although the term is still used as a convenient moniker for a collection of resources managed as a single entity in a single location). Rather it's about the computing components, the interconnects, the storage, and the software that ties everything together and the applications that run on top.
The Top500 nicely illustrates the evolution of HPC over time. This list, released twice annually, ranks the largest publicly acknowledged supercomputers--as the term is used today--on the basis of a somewhat simplistic, but objective, benchmark. The Top500 entries are certainly not typical of mainstream HPC; they're the biggest of the big. But they nonetheless provide some quantitative insight into important trends.
The newest iteration of the list was released Friday. There were no striking departures from the trends of the last few years, but there was some continued evolution that's worth taking note of.
The continued rise of InfiniBand. InfiniBand is a system interconnect that offers a higher performance alterative to the ubiquitous Ethernet. Although its initial backers envisioned a broader role for the technology, it's settled nicely into HPC and, to a lesser degree, back-end commercial data center functions like database clusters where low latency and high bandwidth are also paramount. (The Sun/Oracle Exadata appliance uses InfiniBand for example.)
(Credit:
TOP500.org)
InfiniBand's initial growth in HPC wasn't so much about displacing Ethernet as it was displacing the fractured collection of high-performance interconnects that preceded it. Myricom's Myrinet and Quadrics' QsNet were the most common of these, but there were many. This year InfiniBand is deployed on 181 of the Top500, a 28 percent increase from a year ago.
That's a striking increase clearly. But what is perhaps more striking is that about half that increase came at the expense of Ethernet rather than mopping up a variety of older or proprietary connection technologies. This shift started between 2007 and 2008 but was even more pronounced this year.
It's certainly possible that the next 10GbE generation of Ethernet, which today is essentially absent from the list, could again push Ethernet's numbers higher. However, whatever the specific technology, the message that I take away is that large computer clusters are starting to favor more optimized interconnects even if they and the components they connect are largely off-the-shelf.
And we see an analogous trend with the proliferation of blade servers as well. Blades, a more modular and pluggable approach to system design, have proven popular in many enterprises and midmarket companies, in part, because they help bring together computing, storage, and networking technologies into a single integrated whole. That type of integration isn't of much interest in HPC. Rather, blades play to HPC by offering high densities and reducing cable count and complexity.
In fact, among x86 servers at any rate, dominance is not too strong a word to describe the presence blades in the Top500. Consider just one vendor, Hewlett-Packard. HP has 208 ProLiant systems on the list. A full 203 of these, almost 98 percent, are ProLiant c-Class blades.
Collectively, these trends suggest what might be thought of as a trend toward building optimization around standardization. In the main, especially as one moves down from the very top of the list, the Top500 is composed mostly of systems using mainstream technologies such as x86, Linux, and standard interconnects. Clusters are the dominant architecture.
But we're increasingly not seeing mere rackmount servers connected by Gigabit Ethernet. As the systems on the Top500 list grow in capability, we're seeing more focus on how they're packaged, powered, and connected.
SAN FRANCISCO--General manager of Intel Architecture Group Sean Maloney's announcement of a reference design for a "micro server" during his Tuesday afternoon keynote at the Intel Developer Forum brought me a sense of deja vu.
Intel's Sean Maloney holding microserver.
(Credit: Intel)He disclosed "a new ultra-low-voltage Intel Xeon 3000 series processor featuring a TDP (Thermal Design Power) of only 30 watts. To complement the broad range of dense and power-optimized platform offerings, Intel also demonstrated publicly for the first time a single-socket 'micro server' reference system which will help enable micro server innovation and future specification." Intel plans to ship the 30-watt dual-core chip in Q1 on 2010; a 45-watt quad-core version is set to ship immediately.
A reference system is primarily intended to demonstrate a concept. It provides a hands-on experience for partners and customers and therefore an opportunity to experiment with and fine-tune the basic approach. The microserver reference design will accommodate 16 server modules in a 5U-high (8.75-inch) chassis. The server boards are approximately 8-inches by 4.5-inches.
Jason Waxman, the general manager of high density computing at Intel, told me that they see the primary target for this class of system as "hosting companies that do a lot of white boxes." White boxes are systems that are often assembled in-house from component parts such as motherboards and cases. Waxman added that such companies nonetheless want many of the features associated with servers--such as memory with error correcting code (ECC).
In Intel's view of the world, microservers very much target service providers and companies that host busy Web sites and otherwise are associated with high-scale network computing. It sees this market as distinct from large high-performance computing (HPC) installations. Vendors such as HP tend to treat network computing and HPC as more of an overlapping customer group.
My deja vu when it comes to microservers relates to the fact that we've seen them before. They used to be called blades.
That's not to say that blade servers don't already exist today, but they've largely evolved into a much different concept from how they were initially conceived. The blades sold today by the likes of Cisco, Dell, HP, and IBM are about virtualization and integration. They pull together computing, networking, and storage and tightly integrate them both physically and through software. They are, in a sense, a form of scale-out consolidation.
Sun has largely eschewed this integration with their blade product line. However, Sun blades are heavily focused on high-performance computing--even to the point of integrating the HPC-centric InfiniBand interconnect on some of its products.
Rather, microservers hark back to the days of RLX Technologies, the company that did the most to promote blade servers during the Internet boom of circa 2000. Microservers are simply thin servers--compact, cheap, and simple. They provide cable simplification. They let hosting providers allocate low-cost physical servers to customers who don't want to share using virtualization.
Microservers bring blades back to their roots. Everything old is new again.
The "Dynamic Cube" is Fujitsu's first big server announcement since announcing sweeping changes to its corporate structure at the beginning of April.
By way of background, Fujitsu has been morphing into an organization that's about being an integrated worldwide business--rather than a Japanese company that opportunistically also sells elsewhere--for roughly a year now. In the case of North America, this meant creating an organization called Fujitsu America that consolidated the application services of Fujitsu Consulting, the system platforms of Fujitsu Computer Systems, and the retailing solutions of Fujitsu Transaction Solutions.
This reorganization also saw Fujitsu buying out its share of its 10-year-old joint venture with Siemens AG of Germany. The former Fujitsu Siemens Computers is now Fujitsu Technology Solutions, part of the global Fujitsu organization.
Fujitsu uses Dynamic Cube as a sort of informal moniker for its new BX900 blade server system. But it also applies to a broader collection of hardware and software components that virtualize servers and I/O--and manage those virtualized resources. These include:
- PRIMERGY BX900: This is Fujitsu's third generation of blade servers. Its 10U-high enclosure (vs. 7U for its predecessor) houses up to 18 Intel Xeon 5500 ("Nehalem") processor-based blades (BX920 S1). Fujitsu expects this new enterprise-focused chassis to be sold for at least five years before changes are needed to support future processor generations. It uses a fully passive midplane with 6400 Gbit/sec of total bandwidth.
- VMware ESX, Microsoft Hyper-V, and Xen (Citrix, Red Hat, Novell): Fujitsu relies on the standard x86 hypervisors to create a virtual server pool from the physical server blades in the BX900 chassis.
- ServerView Virtual I/O Manager (VIOM) is an abstraction layer between servers and the network and storage. It's conceptually similar to HP's Virtual Connect and IBM's Open Fabric. The basic idea is that it lets server administrators move around Ethernet and Fibre Channel connections within a blade environment without having to make changes to the broader data center network and storage infrastructure. Up to four enclosures can be part of a single VIOM domain.
- This announcement also sees the global debut of ServerView Resource Coordinator (SVRC) Virtual Edition. It unifies virtual and physical management. It's not really a new product--internally its version number is 13.3. However, it was formerly part of Systemwalker, a large, monolithic management tool that Fujitsu only offered in Japan. Over time, Fujitsu plans to bring in other pieces from Systemwalker, such as storage and network management, while packaging them in a way that doesn't require the adoption of a complete Fujitsu management framework.
The relatively low profile of Fujitsu in North America has never especially been a product issue. Fujitsu customers with whom I've met generally speak warmly of Fujitsu quality, capabilities, and support.
Rather the issue has been more one of Fujitsu's spotty approach to being a worldwide technology vendor. They've had pockets of success and investment but it's long seemed a bit ad hoc. With a clear corporate push behind getting more global, it's at least thinkable that we could start hearing more about Fujitsu in the U.S. And the BX900, together with its associated software, is a nice addition to its enterprise IT portfolio.
Few things are more annoying to PR folks than being all revved up for a big announcement, and then some big event comes along and sucks all the air out of the tech news cycle for the day.
That's what happened to Hewlett-Packard on Monday. The company announced its BladeSystem Matrix. And Oracle announced its decision to acquire Sun Microsystems.
Guess which won the coverage game. (And, yes, I'm one of the guilty parties.)
On a normal day
But on a normal day, Matrix would have, or at least should have, generated a lot of interest. Here's why.
First, a little background about server blades. They were initially pitched back around 2000 as a hardware-focused approach that disaggregated processing from the other components of a computer system and physically consolidated it. We still see remnants of that vision when blades are used in high-performance computing and some high-transaction Web sites.
But they've largely evolved in a different direction. Software (for management and virtualization) and a new style of integration are now the most important elements of mainstream commercial blade products. Many of the virtues extolled early-on for blades--fewer cables, better density, less physical redundancy--remain to be sure. But even if they're necessary, they're not sufficient.
Which brings us to HP BladeSystem Matrix. It's very explicitly an integration play. As HP put it to me: "Matrix is the box, the operating environment, the partners. The overarching environment. It's much more deeply integrated than just a bundle."
The box is HP c-Class blades (both Integrity and ProLiant) including HP's Virtual Connect 8Gb Fibre Channel and Flex-10 Ethernet modules. Virtual Connect essentially virtualizes LAN and SAN connections--allowing them to be switched among server blades inside the box in response to changing resource needs.
The software is something HP calls the Matrix orchestration environment. It brings together several components that HP already sells (and can also ship separately): Insight Dynamics-VSE, Orchestration, Recovery, and Virtual Connect Enterprise Manager (EM). Collectively, this software provides automated provisioning, capacity planning, and disaster recovery.
Finally, partner integration goes beyond the usual "we work with partners." Everyone does that. Specific application templates done in conjunction with the likes of SAP provide standardized (but customizable) ways of setting up complex software environments from a self-service portal.
Down this road before
HP has been down this road conceptually before.
In 2001, it offered up the Utility Data Center (UDC) as the ultimate adaptive, dynamic data center infrastructure--in other words, the sort of attributes it now associates with Matrix. But UDC was ahead of its time. It was tied to a lot of expensive, dedicated components. And it was not especially integrated with software outside of HP. Matrix--as well as other takes on dynamic data centers such as that from VMware--may incorporate proprietary technologies, but it's much more rooted in industry standards and standard components than UDC was. (CNET's Dan Farber has a 2004 take on UDC's demise .)
But the UDC vision and the things that HP learned from the experience clearly informed its work on Matrix.
One last note. I've seen some commentary in the vein of this announcement being a reaction to Cisco Systems. Cisco's Unified Computing System has justifiably grabbed a lot of attention. After all, Cisco is an industry heavyweight. And if Cisco is indeed serious about becoming a data center player beyond networking gear, that's big news.
However, Cisco is just starting out. HP is on its second major generation of blades and is building on software and earlier projects that, in some cases, go back a decade or more.
So to call Matrix a reaction to a newcomer gives Cisco more credit than it deserves at this point. Rather, I view this as HP continuing to build on existing plays that have already led its blades to the No. 1 spot.
- prev
- 1
- next
