In Search of the Right Hardware for the Cloud

Arthur Cole
Slide Show

10 Steps to Increasing Data Center Efficiency and Availability

A systematic approach to building data center infrastructure management.

As I've mentioned often in the past, infrastructure still matters even in the age of the cloud.

That means issues surrounding processing power, throughput and capacity still reign in the physical world, and architectural issues that greatly influence the return on hardware investments are still legion.

One of the most basic, of course, is the long-standing toss-up between blade servers and standard rack-mounted designs. As IT consultant Ken Hess pointed out recently, there probably never will be a simple answer to this conundrum, considering the variables surrounding density, power consumption, workload distribution and management flexibility. When applied to multiple legacy systems across numerous data centers, these factors can easily shift the cost-benefit analysis one way or the other - and that's before you factor virtualization into the mix.

Industry-wide, the trend has been favoring blades for the past decade or so as enterprises embrace the efficiency and flexibility of distributed architectures over centralized processing. And this is likely to continue as both traditional enterprises and the new crop of cloud providers turn to low-cost, scalable, even modular, infrastructure.

In fact, the rise of PCIe-based Flash cache solutions in blade enclosures should accelerate this trend. LSI, for example, has teamed up with EMC and Cisco to provide an integrated solution that combines the UCS B-Series blade with the Nytro WarpDrive card, governed by the VFCache software stack. The intent is to provide a three-fold increase in mission-critical application performance without disrupting legacy infrastructure.

Regardless of which architecture you choose, however, performance across both blade and rack systems should increase courtesy of the newest generations of processors. HP's ProLiant, for example, will sport the Opteron 6200 and Intel Xeon E5-2400 and E5-4600 devices in the new Generation 8 line. This includes the rack mount DL360, 380 and 385 models, the BL420c and 465c blades and the ML350e tower.

Complicating matters even more is the advent of microservers - the new generation of ultra-small machines that now pack some of the most powerful processing engines available. Dell, for instance, has launched the PowerEdge C5220 built around Intel's E3-1200 V2 processor utilizing the so-called Ivy Bridge architecture. Not only will it draw less power than the previous Sandy Bridge models, but the 3D transistor design and 22 nm process is expected to increase processing power by half.

It would be convenient if deciding on a server architecture was a simple matter of weighing the pros and cons of various approaches and proceeding accordingly. That's a luxury that few CIOs have, however, as the needs of users, applications and data environments continue to shift with the tides.

Virtualization and the cloud have succeeded in increasing the flexibility of IT architectures beyond anyone's wildest dreams, but the drawback is that making hard and fast decisions on long-term infrastructure deployments has become more difficult. In the final analysis, then, it may prove wisest not to commit to a single hardware approach anymore.

Blade? Rack? Micro? Going forward, the correct answer might be "all of the above."

Add Comment      Leave a comment on this blog post

Post a comment





(Maximum characters: 1200). You have 1200 characters left.



Subscribe to our Newsletters

Sign up now and get the best business technology insights direct to your inbox.