Your budget is more in line for a 99.9% uptime. If you need a 99.99% you will need to rethink your budget.
Consistent 99.99% uptime is difficult enough to achieve for one of the many points of failure (server, power, network, DNS, etc.). It is not at all trivial to achieve that level of uptime for all of them. Add them all together and the costs become exponentially higher.
You should ask yourself if you really need "less than 5 minutes downtime a month" (many major banks have more downtime than that) or do you simply want to be reimbursed for any downtime over 99.x%?
The only host I know that is 99.9% uptime is GoDaddy. Except they don't offer Fantastico (or cPanel)..
GoDaddy is certainly a host that "guarantees" 99.9% uptime with some sort of SLA, doesnt mean they actually achieve that. Many hosts on this forum have a far superior uptime record compared to GoDaddy.
Also in a clustered server, because what it does change is that you have different services in different servers, if your httpd goes down, it doesn't matter if your email or mysql server is up.
Jedito, why do we keep meeting like this
Though I agree, 99.99% is not likely in a services cluster alone over a prolonged period of time, it certainly is "more" likely then in a single server architecture environment. Though I agree that a service is down if a service is down, and even though other services (email, etc) are still up isnt relavent (at least in this conversation) - you do dramatically decrease your probability of having a service go down in a cluster formation. Additionally, time to recover in case of disaster is significantly reduced. Combine these 2 factors, and all else being equal - a services cluster will have a better uptime record vs a single server architecture model. (having said this, a well maintained environment will always do better then a half hazardly maintained environment - architecture aside)
Last edited by cartika-andrew; 04-28-2006 at 12:03 AM.