torsdag den 30. oktober 2008
En af Googles succesparametre ... oppetid.
A Google parameter of succes ... reliability.
The reliability of cloud computing has been a hot topic recently, partly because glitches in the cloud don't happen behind closed doors as with traditional on-premises solutions for businesses. Instead, when a small number of cloud computing users have problems, it makes headlines. As with most things at Google, we are fanatical about measuring the availability of Gmail, and we thought it best to simply share our reliability metrics, which we measure as average uptime per user based on server-side error rates. We think this reliability metric lets you do a true side-by-side comparison with other solutions.
We measure every server request for every user, every moment of every day. Any millisecond delay is logged. Over the last year, Gmail has been available more than 99.9 percent of the time — for everyone, both consumers and business users. The vast majority of people using Gmail have seen few issues, experienced no downtime, and have continued to have a great Gmail experience, with exception of an outage in August 2008. If you average all these data together, including the August outage, across the entire Gmail service, there has been an aggregate 10-15 minutes of downtime per month over the last year of providing the service. That 10-15 minutes per month average represents small delays of a couple of seconds here and there. A very small number of people have unfortunately been subject to some disruption of service that affected them for a few minutes or a few hours. For those users, we are very sorry. And for Google Apps Premier Edition customers, we have extended service level agreement credits to them.
So how does greater than 99.9 percent reliability compare to more conventional approaches for business email? We asked some experts. Naturally, the normal caveats apply for on-premises solutions, since each individual business environment will vary, depending on server reliability, staff response time, and actual maintenance schedules for each application.
According to the research firm Radicati Group, companies with on-premises email solutions averaged from 30 to 60 minutes of unscheduled downtime and an additional 36 to 90 minutes of planned downtime per month.1
Read more: http://googleblog.blogspot.com/2008/10/what-we-learned-from-1-million.html