I read that:
+"Oracle’s largest in-house database is a 101-terabyte database running Beehive, the company’s in-house email and collaboration software, running on nine Oracle Exadata boxes."+ and I'm just wondering how big the largest known implementations are outside of Oracle Corp.
The URL for the above quote is:
Turkcell is a big one. Just google "Turkcell Exadata" and you should find plenty of info. Ferhat has a blog where he's discussed various issues. Tanel Poder also blogged about the initial migration. Turkcell was an early adopted and moved a 100T db on to a V2 full rack. They later bought 2-X2 full racks and created a 16 node cluster. Not sure how big the DB's are now (there are multiple). HCC compression has been used heavily so the uncompressed size would be very substantial. Ferhat did at least one presentation at Open World this year as well.
Kevin clossen has posted some interesting hypotheses about that largest oracle database where the published i/o load of the system seems to be very low :) so in this case it might have been a case of a bunch of Exadata machines were available and so Oracle put its internal email system on them for the publicity.
You can find the information on his blog .
On the other hand there is a list of customer references "Customers Achieve Extreme Performance at Lowest Cost with Oracle Exadata Database Machine" on oracles site it includes Turkcell. According to that article Turkcell has 40 Tb of data compressed into 10Tb of space so they should not really need more than a single full rack.
According to the slide the old V2 Rack was supporting 250T of uncompressed data and the new 2 rack X2 cluster was supporting about 600T of data (this was as of last Sep. so it's probably somewhat larger by now).