informa
/
Database Security
Commentary

Power Rationing--Green Gets Serious

As part of my normal routine I try to speak with as many data center managers as possible. A trend has appeared lately that I believe we are on the front end of. I am calling the trend power rationing. We have been told several times now over the past few weeks that data center managers are being given a hard limit as to how much power they can use. This is a shift from the more common "Reduce power consumption by x%" to "You can use X watts of power."
As part of my normal routine I try to speak with as many data center managers as possible. A trend has appeared lately that I believe we are on the front end of. I am calling the trend power rationing. We have been told several times now over the past few weeks that data center managers are being given a hard limit as to how much power they can use. This is a shift from the more common "Reduce power consumption by x%" to "You can use X watts of power."I explain the difference like this; no one is very happy with the cost of a gallon of gas right now, especially us Texans with our big V-8 gas-guzzling trucks, but at the end of the day we can still get all the gas we want. We don't have to change our driving behavior; we just have to pay more for it. That said, because of the increased gas cost we're trying to cut back on travel as much as possible. This is similar to today's data center -- you are trying to cut power consumption as you go through your normal product rotation cycles, and while it may take you longer to get power and it is definitely costing you more, you can still get it, eventually.

Rationing changes that. If suddenly you were only allowed 20 gallons of gas a week, you would have to instantly change your driving habits. If this power rationing trend continues, you will have to instantly reduce your power consumption in the data center. No longer will gradual upgrades work; you must be able to turn things off, completely.

Instant power savings can be achieved quickly in a few areas. First is data archiving, getting old data off of your arrays. Disk archiving is ideal for data that still needs to be accessed occasionally, meaning that disk is still needed, but can be moved from these active disk areas to systems like Copan's File Archiver and NexSAN's ATABeast, that can turn off the drives attached to it. The remaining active data set should be managed as well with data compression. Companies like Storwize can deploy an in-line compression technology to compress all of your active file system data, even network-mounted databases with no performance impact. With old data off the arrays and active data being compressed, file systems can be consolidated to fewer storage systems, and those arrays can be turned off.

Solid State Disk solutions like those from Texas Memory and Solid Data Systems play a key roll in turning things off. If you have a database application that has a significant I/O workload, you're probably striping that data across multiple disk drive spindles to be able to deliver the required performance. High drive count also means high power and cooling costs compared with a single Solid State Disk that can deliver even better performance. This again allows you to reduce that drive count and turn arrays, or at least drive shelves, off. Further, SSD performance can also mean fewer servers needed to handle database I/O requests. Those servers that are now no longer needed also can be turned off.

Lastly, infrastructure virtualization solutions like those from Scalent Systems and Unisys play a key roll here. Unlike server virtualization technologies that require physical servers to be powered on and running, infrastructure virtualization can actually power physical servers on and off. It also allows servers to be repurposed as needed. Instead of having a physical server set up as a backup to your virtualization environment, you could have it powered off until needed. Then, when there's a failure, power the box on and route it to the correct storage and IP networks when it's needed. Many physical servers sit powered on but idle in a data center in case something goes wrong. With infrastructure virtualization, you can re-deploy a failed service in the time it takes to power on and boot the server. For services that can afford a few minutes of downtime, this can lead to a huge cost savings.

As green gets serious and power restrictions less flexible, green is more than just increasing power efficiency; it is reducing consumption in real time.

George Crump is founder of Storage Switzerland, an analyst firm focused on the virtualization and storage marketplaces. It provides strategic consulting and analysis to storage users, suppliers, and integrators. An industry veteran of more than 25 years, Crump has held engineering and sales positions at various IT industry manufacturers and integrators. Prior to Storage Switzerland, he was CTO at one of the nation's largest integrators.

Recommended Reading:
Editors' Choice
Kirsten Powell, Senior Manager for Security & Risk Management at Adobe
Joshua Goldfarb, Director of Product Management at F5