Green is the New Black

Black is the fashion industry’s never-fail color. For every other industry, green is the color that can’t lose—environmental green, that is. Once little more than window dressing for mollifying the environmental lobby, green has become big business. Despite the some governments’ resistance to environmental warnings, the business community and the public have caught green fever.

Businesses are paying attention to green and clean. “Clean tech” companies raked in more than $800 million of venture capital funding in Q3 2007, an 80% increase over Q2, according to PricewaterhouseCoopers. Wal-Mart’s latest “better living” survey records steady increases in environmentally friendly product sales and predicts that 43% of Americans will classify themselves as “very green” within five years.

But should CIOs care? True, as top officer at their companies, CIOs have a share of responsibility for the companies’ public images, which these days must include green elements. IT’s data centers consume huge amounts of electricity, which drops a major green issue right into the CIO’s wheelhouse. At the end of the day, however, CIOs are evaluated on whether the databases are accessible, the applications are responding and the network is fast. If forced to take a hit in one realm or the other, right now it may be in green friendliness.

The good news for CIOs is that it isn’t always a choice between good green or good business. There are technology strategies that deliver IT and green benefits at the same time, most notably server virtualization. Data centers use a lot of servers and servers use a lot of electricity. Most of it is generated by coal-fired power plants, a constant target for regulators and environmental advocates. The state of Kansas just denied permits for two coal-fired power plants, citing their environmental impact. That underscores the trouble utilities will have building generating capacity to meet the seemingly endless growth demand.

Power consumption by the world’s data centers doubled between 2000 and 2005, a 16% annual worldwide increase according to a report by Lawrence Berkeley (Calif.) National Laboratory staff scientist Jonathan Koomey. Koomey estimated that worldwide data center power consumption will increase 40% by 2010 if current trends continue. If it keeps up, he wrote, the world will need 14 new 1,000 megawatt power plants to meet the demand, most of them fossil-fueled.

A key reason servers consume a lot of electricity is because they use it so inefficiently. Google’s decision to retail its in-house AC/DC power converter technology highlights server inefficiency. The Google converters work at 90% efficiency in data center servers. Wasting 10 percent of the electricity that flows through the server isn’t great, but it’s a lot better than the standard 60-to-70 percent efficiency of most data center servers.

Server virtualization’s main goal is to use server capacity more efficiently, but in doing that it also reduces power consumption. Those benefits are fueling virtualization’s growing popularity. Probably fewer than 10% of data center are virtualized today, but IDC predicts that more than 75% of all large companies (more than 500 employees) are deploying virtual servers.

There is one significant limiting factor CIOs must consider before turning to virtualization: reliability. For all of its faults, the brute force approach minimized the impact of individual server crashes. In a virtualized environment, a server crash will take down several important applications. Clusters and fault-tolerant technology are the two most commonly employed technologies for increasing availability, and they work in virtualized environments as well. The problem is they consume roughly equal amounts of electricity, so the green quotient is a wash.

Clustering also is a complicated failure-recovery technology to implement and manage. Since simplifying the IT infrastructure is a goal in virtualization projects, extensive use of clusters wouldn’t generate a good long-term return. Fault-tolerant technology prevents failures from occurring in the first place, eliminating unplanned downtime and data loss. Fault tolerant servers are also easier to use and simpler to manage than fail-over clusters.

The most successful crash-resistant virtualized data centers will probably use a combination of standard, clustered and fault tolerant servers to support a range of uptime availability requirements. Fault-tolerant servers will protect the application environments that can tolerate no downtime and require continuous availability.

Regardless of which solution IT managers choose, security and reliability complete the blueprint for “greening” the data center through virtualization. It all goes back to the old saw about putting your eggs in one basket. If an IT manager has to do that by deploying several key applications on one server, then he has to think of a color other than green – white, the color of titanium, because his server “basket” has to be that strong.

Greg Enriquez is senior vice president, worldwide sales & field operations at Stratus Technologies, a Maynard, Mass.-based developer of continuous availability technology.