Posts Tagged ‘server utilization’

• Data Center Virtualization and Cost Reduction

  On 03.01.10, In Data Centers, GreenIT, IT Asset Management, By Rajesh Kadam

  GreenIT has brought significant cost saving opportunities by way of increased productivity.  Data Centers have been quick to grab on to this . According to the Green Grid, nearly 60% virtualized their IT equipment. Another 47% consolidated and optimized their servers, 46% bought computer equipment that uses low power / low wattage processors and 40% make full use of power management tools. Virtualization and consolidation of servers and storage devices is a no-brainer—a way to reap productivity gains without sowing investment. The House of Representatives, for example, reduced its energy consumption its data centers by 50% and lowered its installed base of servers! With the state of IT Asset Management today, it’s not uncommon for server utilization to be as low as 2% to 5%. Companies can make-do with fewer servers, increase their rate of utilization and save the energy costs that are otherwise incurred on operating the surplus servers before virtualization reduces their numbers. A growing number of options exist to lower costs—blade servers which save space and energy costs, switch to cloud computing and desktop virtualization. For efficient utilization of energy, the devil is in the details of comparative data of consumption by each piece of equipment and the angel in the overarching view of consumption patterns. Companies need to automatically discover their many assets and all the details of their versions, brand names, parameters of power consumption, etc. They also need to categorize the data for it to be intelligible and map the relationships in the asset classes, compare their baseline consumption with the benchmarks in order to detect opportunities for cost reduction while maintaining service levels. Manual data crunching with Excel sheets is way too cumbersome and slow. Software tools for IT Asset Management are needed to automate data gathering, analysis and visualization for identifying the sources of excess costs and means to lower them.

  0 Comments

• Datacenter Consolidation: Management of Latent Inefficiencies

  On 02.16.10, In Data Centers, GreenIT, IT Asset Management, By Rajesh Kadam

  Over the past few years, datacenters bought a grab bag of servers and other IT equipment in a bunch as if from a clearance sale. But then Datacenter managers woke up to the fact that they were running up soaring bills for energy, software licensing, maintenance, duplicate applications, higher cost of space and staff to manage the assorted equipment. A garage sale of servers followed as applications were consolidated on fewer servers and data center managers grabbed the low hanging fruit of higher server utilization. Now the time for a slog has come---the gains from server consolidation and virtualization in recent quarters have diminished. Global sales of virtualization software fell 18.7 %, according to IDC data, in the second quarter of 2009. Analysts at IDC and Gartner agree that this was not due to the economic recession alone. The Gartner analyst concludes, “data center managers are still not paying sufficient attention to the process of measuring, monitoring and modeling…….”. Currently, equipment in datacenters is largely monitored manually with excel spreadsheets which is something like highway police tracking traffic without radar. A Computerworld survey revealed that 63% of enterprises rely on excel spreadsheets to assign IP addresses. As the size of the datacenter increases, the excel plodding gets more excruciating and the single largest component of the operations cost according to IDC. In consolidated and virtualized dataceters, the price of any outage is jolting. A single outage in Amazon was estimated to have cost $1.86 million per hour. More applications are deployed on fewer but more powerful servers which consume more electricity. If they are underutilized, the loss from malfunctioning of applications and loss of electricity is a whole lot more than with older servers. Virtualization lowers visibility into performance of individual units of equipment as capacity is assigned randomly by unseen electronic means to whichever unit working at less than full capacity. Manual monitoring of capacity use would be like recording the movement of a submarine with binoculars. Capacity assigned by electronic means is best read automatically by software able to sense the signals on the network. A consolidated and virtualized data center subsumes the previously siloed physical pieces of equipment and depends on on its many inter-connected moving parts working together for optimal performance. Any chink anywhere in the network has a cascading effect on the entire system and brings it down like a house of cards. Automated discovery and monitoring is like a medical device reading the pulse of the body and forewarns against failure and finds the root cause if the damage is already done.

  0 Comments