The "data center in a box" approach to container computing has been controversial from the start. The merits and drawbacks of packing servers into shipping containers have been vigorously debated from day one (see Sun's Blackbox: Game Changer or Niche Product? from October 2006). Some of the industry's largest players are embracing the concept, especially Microsoft (MSFT).
But the skeptics remain many, and their critiques are summarized by ComputerWorld, which outlines six reasons containers won't work. Eric Lai interviewed data center professionals and consultants, and reports that they "were individually impressed with some parts of Microsoft's plan, (but) also expressed skepticism that the idea will work in the long term."
Here are the weaknesses in the container model cited in the ComputerWorld article:
- Managing power for a large container installation is more challenging than believed, and raises the risk of harmonics that can affect power quality.
- Servers in containers are more likely to be damaged in transit.
- The power and network connections represent a single point of failure, a point recently noted by the Uptime Institute when it gave Blackbox/Sun MD a Tier II rating.
- Microsoft's planned "lights out" operation and reliance on remote monitoring raises the likelihood that containers will fail and need to be replaced.
- Containers are a temporary solution, a "short, ephemeral model" as one source puts it.
- Microsoft's plans for air-side economization - using outside air to cool the data center - don't work well in all climates.
None of these issues are new, and Microsoft says it is aware of the challenges that accompany the "container farm" approach. "Half of the people say this is the greatest thing they'd ever heard. The other half say this will never work inside a data center," Manos told ComputerWorld. "But the fact of the matter is that this does work."