- 1 Vapor IO Brings OpenDCRE to General Availability
- 2 VMware Takes the Wraps Off vRealize Automation and vRealize Business
- 3 Microsoft Previews Hyper-V Containers for Windows Server 2016
- 4 Mirantis Led FUEL Project Gets Installed Under OpenStack Big Tent
- 5 Red Hat Enterprise Linux 7.2 Adds Security, DR Features
IBM Brings Power Gauges to Mainframes
The latest move in IBM's Project Big Green is what it calls a "gas gauge" for its mainframes, allowing customers to monitor their systems' precise energy consumption in real time.
|The latest addition to Project Big Green is a "gas gauge" for its mainframes that will enable customers to monitor the precise energy consumption of their systems in real time.|
By doing so, mainframe users can get a far more accurate picture of how much power they are actually using at any given time or over the course of a day. This will allow for more precise planning for workloads and making decisions regarding deployment of other equipment.
IBM also plans to publish typical energy consumption data for the IBM System z9 mainframe, based on the responses of around 1,000 customer machines to determine average watts per hour consumed.
While a mainframe does have a listed power draw, called a "label rating," it often doesn't match the real draw, since every machine has different workloads. Rick Lechner, vice president of IT optimization at IBM, told InternetNews.com that the typical energy use can be as low as 60 percent of the labeled power draw.
"It's helpful for customers to know that, so they can see how much power they are consuming and do a better job of planning for power consumption," he said. "What it demonstrates is the mainframe is leading the way here. This is the first server to meet the EPA's request for a gas gauge, if you will, to let the client know what the actual power consumption is instead of the label power."
The mainframe has always had the ability to report on such information. It's just that up to now it has not been needed or wanted, said Lechner. The "gas gauge" monitors a mainframe's actual energy and cooling statistics as collected by internal sensors and presents them in real time on the System Activity Display.
Statistics can be monitored in real time or gathered for statistical reports run daily, weekly or whenever the customer wants. IBM is also offering a Power Estimator Tool for future planning, based on the findings of the gauge. The Estimator Tool calculates how changes in system configurations and workloads can affect the entire energy envelope of a datacenter.
"Because the gauge doesn't just tell you how much power you use but it also correlates the power to the workload, customers can see how much utilization of this IT asset they are getting for this much power consumption. So they can compare the efficiency of different servers and make more intelligent decisions on the placement of workloads," said Lechner.
In addition to the z9 mainframes, it will also monitor the power draw of storage devices. Lechner said this type of monitoring will be on all IBM servers eventually. The System p, which uses IBM's Power6 processors, has a similar monitor which can be set to throttle back the processor during times of low workload, thus saving money.
IBM's Green Grid project is part of a major effort to help data center customers get a handle on their power, cooling and space issues. Just as Microsoft uses itself as a guinea pig to test its own software first, IBM is eating its own dog food and making itself the first test target for these ideas.
Since 1997, the company has consolidated 155 datacenters worldwide down to seven. It evaluated 11,000 servers deployed company-wide and has launched on a project to consolidate 3,900 of those servers onto 30 highly virtualized mainframes running Linux.
The company noted a side benefit, too. In addition to all the power and space saved by such a consolidation, a lot of software is priced on a per-processor basis. Going from 3,900 machines to 30 will cut down not just on the electric bill but on the software licenses as well.
This article was originally published on InternetNews.com.