There is a new post on Data Center Knowledge regarding tiered pricing by Equinix. To quote:
Equinix (EQIX) said last week that it expects to implement tiered pricing that acknowledges the broader range of facilities and markets the company now serves.
This is a good idea because there is a diversity of needs for datacenter space across different geography at different levels of quality. Not every app needs to be supported by a tier 4 datacenter. But some apps do need to be local to a given geography. Offering a mix of environments across a wider footprint is a good move.
However I was disappointed to see a rationalization by Jarrett Appleby (normally a person that I think highly of) that potentially sets unreasonable expectations:
when it comes to reliability, he said not many changes are needed in the Switch and Data facilities. “We’re getting 5 nines (99.999 percent uptime) out of these sites,” Appleby said. “So from a performance standpoint, they’re already there.”
When it comes to something like reliability of complex systems, past performance is not an indicator of future performance. Systems break over time; components wear out. A newly built datacenter might operate at 100% availability but that doesn’t mean that you should expect consistent 100% uptime for the foreseeable future. Rather, it’s reasonable to set long-term performance expectations based on analysis of the components’ availability and redundancy. This is what the “tier” nomenclature refers to.
So, if Equinix wants to sell lower-tier datacenter facilities then good for them. Seriously, it’s a good idea. But customers should be aware of what they’re buying into, and not misled about the quality of a given facility. Your app may need to be hosted in multiple datacenters if the uptime requirements are greater than a single facility can provide. Marketing doesn’t change that fact.
