Telefonica Digital has launched a cloud service from servers in Madrid and London, using kit from Joyent and promising faster scalability and greater reliability than the competition. Instant Servers promises 99.996% uptime and backed cash compensation. Telefonica also reckons its clouds can quadruple in capacity instantly, and …
99.996% uptime means no more than 3.456 seconds per day downtime. Believable?
If they do it like the telcos do, they could have 10% of their systems down for 34.56 seconds, or 1% down for 345.6 seconds, and still get an "average" of 3.456
The key thing (which we're not told in the article) is over what period you measure it. My guess is a month, so you're allowed 1 minute 45 secs of downtime without incurring penalties - this would be close enough to 100% for most commercial purposes. But it could be measured over a year, permitting a single outage of 20 minutes, which might not be acceptable.
If you want to read the Ts&Cs, scroll down to Schedule II Para 3, and you shall find:
Portal Availability: percentage of time in which the Telefónica Digital Cloud Services Portal is running and giving access to the Client. The objective value is 99.996% yearly, 99.95% monthly.
Always read the small print.
It all depends on waht you consider downtime
If it means "server processing available", that is quite different from "server available from outside".
Also, it may say that "maintenance" does not count in availability.
So, read the small print..
That's a sex thing surely?
"busty" applications ?
I'm obviously still a bit confused about this cloud stuff - I'd have expected 5 9s availability to be the starting point for commercial contracts. I assume data is stored in multiple locations, so that a data centre failing doesn't cause more than a short pause in service ? I mean, thats 1980s technology.