Planned Obsolescence Architecture?
By jasoncatsun on Oct 28, 2008
Increasingly we're talking to our customers about "fail in place" architectures -- certainly something that the "cloud" can help facilitate. This architecture promotes the idea that if something fails, leave it be, the service continues. Perhaps something "fails in" to provide that capacity, or perhaps things are degraded for a period of time (varies with the SLO, may mean a loss in service capacity or not.) All possible in today's world.
In contrast, many IT architectures do not follow this concept. Often degraded means "emergency!" Emergency changes are made to try and bring the service back to full capacity and degraded equals a loss in service capacity. Does it have to be? With virtualization, are we leaving a little "spare" capacity for this type of event? An extra blade on the chassis unallocated?
Now back to components -- what if you knew that a piece of hardware was going to fail every three years, but you planned around it -- architecturally. The cost of that item should certainly be less. I would bet your overall SLOs would be better!