Using this as a data source:
and selecting Google Mail and Postini Services at the items to measure, he arrives at this:
|4/9/2010||Postini||2:24||Unspecified Emergency Maintenance|
|4/7/2010||Core Gmail||11:20||HTML mode email down for "a small number of users"|
|4/2/2010||Postini||2:08||Failed Postini update|
|3/16/2010||Core Gmail||9:51||Inbound/Outbound Email was not routing|
|3/15/2010||Core Gmail||4:38||Users unable to access gmail accounts|
|3/10/2010||Core Gmail||0:50||Users unable to access gmail accounts|
|3/8/2010||Postini||0:29||Anti-Spam not anti-spamming|
|3/4/2010||Postini||0:58||Anti-Spam not anti-spamming|
|2/25/2010||Postini||7:46||Users unable to send email|
Total runtime (2 services) 2304
48 days, 24 hrs day, 2 services
I have not done my own math here to verify, just thought it would be interesting to share. Note that he gives them twice as many runtime hours since he's counting two services. I would tend to halve that, resulting in double the downtime percentage.
I thank Google for publishing their outage information, by the way.
Just a data point for the next time someone is asking you for more nines than is reasonable.