![[*]](dodeca3_100.gif) | Ned Batchelder : Blog | Code | Text | Site Top ten databases » Home : Blog : December 2003 |
Top ten databasesFriday 12 December 2003 Winter Corporation conducts a survey of large databases, ranking the largest in a number of categories. The top by total size this year is France Telecom with a 29-terabyte Oracle database. Most rows goes to AT&T with half a trillion rows in a Daytona database. There are lots of other categories. Surprising to me was how many off-brand databases are in the lists. Daytona? Teradata? | |
Comments
On the AT&T website it says that Daytonal manages a 94 terabyte data warehouse. Larger than what was listed.
Note also that the web sites makes a distinction between databases used for data warehousing and those used for OLAP. In the latter case the top database sizes are a bit smaller and the DBMSs used are DB2 and IDMS.
What does SMP mean?
Symetric Multi Proccessor
And MPP is "Massively Parallel Processing".
Welcome to the world of "true" enterprise (distributed computing) -- as opposed to the stuff that makes up the typical blog threads around soap, xml, rest, etc.
Now imagine what it would be like to have a multi-terabyte database that had to handle 40,000+ transactions per second. This is the world of the "off-brand" databases.
FYI -- I was involved (slightly) with a 52TB database over 5 years ago. This survey is clearly incomplete -- and not surprising -- the size of these databases, the transactional throughput, the operating characteristics, etc. are competitive advantages to the enterprises that own them. And these enterprise might easily spend > $100M/year maintaining and updating these databases.
Add a comment: