Does heat kill your hard drive? Most of us think, temperature impact life span of hard drives more than anything.
This has been a persistent myth for years within the storage industry. Until a team of researchers from Rutgers University, GoDaddy and Microsoft shatter the myth.
The researcher team led by Ioannis Manousakis and Thu D. Nguyen, of Rutgers, Sriram Sankar of GoDaddy, and Gregg McKnight and Ricardo Bianchini of Microsoft studied the impact of temperature and humidity variations on hardware reliability in datacenters.
I will quickly summarize the overview of the study:
A recent study estimates that data centers consumes roughly 2% and 1.5% of the electricity in the U.S and worldwide, typically a single hyper scale accounts more than 30MW.
To reduce data center energy consumption, the techniques involved increasing the hardware operating temperature and reducing the need of cooling air inside data center.
Although lowered cooling cost seems clear win but this may have severe consequences like decreased hardware reliability.
None of the prior research have yet addressed the tradeoffs between cooling energy, datacenter environmental conditions, hardware component reliability, and overall costs in modern free-cooled datacenter.
Environmental Conditions and Free cooled Reliability in Free-cooled Datacenters clears the picture of environmental conditions impact on reliability of hardware.
The study was carried out at 9 Microsoft datacenters around the world for 1.5 years to 4 years and data were collected.
Based on the data they derive a new model of disk lifetime as a function of both temperature and relative humidity.
Then the researchers quantify parameters between energy consumption, environmental conditions, component reliability, and costs.
Their Key Findings Include:
- On average 89% of component failures account for disks, regardless of the environmental conditions.
- Relative humidity have a much stronger impact on disk failures than absolute temperature in current data center operating conditions.
- Temperature variations and relative humidity variation are negatively correlated with disk failures
- Disk failures rates increase significantly during periods high relative humidity
- Disk controller/ connectivity failures increase significantly when operating at high relative humidity
- Server designs that place disks in the back of enclosures can reduce the disk failure rate significantly (in high relative humidity datacenters)
- Employing software techniques to mask disks significantly reduces infrastructure and energy costs.
This research is an important breakthrough for the enterprises who maintain data centers. Now they can focus their efforts to maintain humidity level rather than employing techniques for cooling.