Amazon's EC2 SLA is extremely clear -
a given region has an availability of 99.95%.
If you're running a website and you haven't
deployed across across more than one region then,
by definition, your website will have 99.95%
availailbility. If you want a higher level of
availability use more than one region.
Good point.
let P(region fails) = 0.05% and let's assume (and hope) that the probability of failure of one region is independent of the state of the other regions.
P(two regions fail) = P(one region fails and another region fails) = P(region fails) * P(region fails) = 0.05% * 0.05% = 0.0025%
Making your availability = 100% - 0.0025% = 99.9975%
Ultimately it's more of a business decision if you want to pay for the extra 0.0475% of availability. I would think (or hope) that most engineers would want it anyway.
The numbers at this size appear insignificant. How would one (say an engineer) convince "the management" that the extra 0.0475% of availability is worth the investment/expense?
let P(region fails) = 0.05% and let's assume (and hope) that the probability of failure of one region is independent of the state of the other regions.
P(two regions fail) = P(one region fails and another region fails) = P(region fails) * P(region fails) = 0.05% * 0.05% = 0.0025%
Making your availability = 100% - 0.0025% = 99.9975%
Ultimately it's more of a business decision if you want to pay for the extra 0.0475% of availability. I would think (or hope) that most engineers would want it anyway.
The numbers at this size appear insignificant. How would one (say an engineer) convince "the management" that the extra 0.0475% of availability is worth the investment/expense?