We all read frequently about rising cooling and power demand in the data center. It seems like the trend is back to the mainframe world where computers will be housed in water cooled cabinets. My reaction to all of that is, who cares?
Of all the considerations for a data center deployment, power density has to be near the bottom of the list. Data center design, as measured by efficiency, is light years ahead of density. And itâ€™s not like density is difficult to achieve. So why are we constantly bombarded with these messages?
Iâ€™d really like to see people talking about a far more important criterion, namely â€œcriticality.â€ In my mind, consolidation and increasing criticality on a per site – or a per cabinet or per physical server basisâ€¦just pick the unit that matches your business â€“ is a far more important factor in understanding the underlying drivers in this business or where to spend your operational dollars.
For example, have you ever wondered why the level of redundancy of facilities is increasing from N to N+1 to 2N? Or why secondary deployments are becoming more common? The answer is simpleâ€¦when you can consolidate data center assets 10:1, using virtualization or utility computing technology, the dramatic rise in unit criticality is driving development, from ad hoc office data centers to purpose-built, scale data center developments as well as the development of secondary sites, now justifiable owing to the cost of outages or the reduction in locations.
Itâ€™s easy to see the many ramifications to consolidation and increasing density, whether itâ€™s i/o and network virtualization to increasing investment in back end infrastructure, but when the consequences of an outage have been multiplied 10x, criticality is the area to focus on if you want to keep your day job.