Reduce Your Software’s Carbon Footprint

Posted By Lev on December 21, 2010 | Software Quality

As you read this, thousands of data centers across the country are using up something around 1.5% of US electricity consumption. That number is projected to increase by 70% according to a recent congressional report. There’s quite a lot of coverage about software that helps manage power use when computers are not fully utilized, virtualization technology saving on hardware use, Cloud performance testing, and more recently about data center metrics like CUE and PUE.

Most of our datacenters are running IT software that supports business processes like Trading systems in financial services, Order Entry, Billing and Provisioning in telecom companies, and ERP, customer accounts, and other transactional systems that are part of any business. Many of these systems are custom developed or heavily customized by IT departments.

Sometimes this customization makes the software run more efficiently. But all too often, corners are cut and software performance problems are solved brute force by throwing more hardware at the problem. Hardware is cheap, but its carbon footprint adds up quickly.

Why is the brute force solution prevalent? More than training, the problem is time pressure. Most IT software developers are not properly trained in the software engineering techniques of optimizing performance. But even when they are, they don’t have the time to think about that as they rush to get projects done on schedule. Most managers have no way to influence their IT development organization towards writing more efficient software.

Some companies track power consumption so they can mind their PUEs and CUEs. We believe that while measuring the size of the problem is a good start, if you want to actually fix the problem it’s better to measure its source.

Over the years we’ve seen many companies, including HSBC, WellPoint, Allianz, among others, analyze the structure of their business applications and reduce hardware resource use by 10-15%. It’s even better when that analysis takes place all through development to ensure that the resulting product is always well engineered, runs efficiently, and is reliable.

Maybe instead of riding their bikes to work once a week, green-minded IT professionals should think about analyzing the structure of their biggest applications.

Tags: efficiency, Green, software engineering