Follow Datanami:
April 24, 2013

Implementing an On Demand Data Center

Ian Armas Foster

As the data processing and storage needs change over time for businesses, so too must the nature of their data centers. However, that transformation process can take a considerable amount of time to complete, which is unacceptable in an environment where the demand for results approaches real time. As such, Jon Hudson and Pete Moyer of the Brocade Global Solutions Architecture Team say they are working toward an on-demand, faster and more responsive data center.

According to Hudson and Moyer, plenty of technologies exist to reduce data center response time, including x86 virtualization and Brocade-supported OpenFlow. However, implementing those technologies introduces a certain element of risk, as unpredictability continually enters the equation the faster datacenters are overhauled. If that risk isn’t nullified or reduced significantly, it wouldn’t be worth the upgrade in response time.

Three sequential levels and one overall orchestration level exist for Hudson and Moyer when it comes to adjusting the data center’s reaction. The first subset, servers, represents the meat of the data center whose fix is relatively straightforward with the implementation of x86 virtualization. The second, L2-7 devices, can have their response time reduced by a combination of virtual routers and virtual ADX’s that quickly connect operators to their facilities.

The third section, routing and policy, generates faster response times when OpenFlow is implemented.

“You’re reducing the overall sequential tasking from days to minutes,” said Moyer on what happens after some combination of these technological upgrades is instituted.

Again, these specific technologies, while reducing response time, introduce a level of risk that at a certain point becomes unsustainable, leading Hudson to rhetorically ask the question, “How do I take advantage of these technologies without introducing that risk?”

The answer lies in the fourth, overarching problem of orchestration. The ability to monitor and control these upgrades is paramount, and, according to Hudson and Moyer, OpenStack serves to take care of that in a safe and efficient manner. “Even if I’m implementing OpenFlow on existing hardware, I can now do it safely,” Hudson said on the effect of applying OpenStack to orchestrate the data center.

Of course, it is worth noting that, like OpenFlow, Brocade supports OpenStack. However, making quick and safe changes to data center hardware is becoming increasingly important, and if the OpenStack system can indeed orchestrate the various other facility changes, the promise of a faster, more responsive (or on-demand) datacenter may soon be realized.

Related Articles

Giving Big Data a Sporting Chance

IBM Weaves Brocade into Big Data Fabric

Wolfram: Software the Hard Part