On 11/1/06, mos wrote:
At 02:27 PM 11/1/2006, Jochem van Dieten wrote:
What is the big deal of a TB? Now, if you get past 20 TB you might
want to team up with one of the commercial PostgreSQL supporters
(Fujitsu, EnterpriseDB, Greenplum etc.), but Sun even sells appliances
for 100 TB PostgreSQL databases.
There is a big difference between a 20 TB database and a 20 TB
table! Unless you're storing huge blobs, a table of over 1TB will have
hundreds of millions of rows (billions?), and that means huge index trees
that need to be maintained.
Indexes scale with ln(O). The difference between 100 million rows and
1 billion rows is maybe 10%. And if you are worried about your indexes
getting asymmetric use a hash index instead of a B-tree. Though
realistically you would partition the data and then your indexes get
partitioned too.
If PostgreSQL can put 20 TB into a table and
still have reasonably fast inserts and queries, then I'll take my hat off
to them.
It can if you design your queries to make use of the indexes and the
partitioning.
But first I need to see proof that they can accomplish this. So if
you have any sites or white papers you'd like to share, go ahead. Keep in
mind we're talking about TB tables here, not databases.
Google the PostgreSQL and PostGIS mailinglists.
Jochem
--
MySQL General Mailing List
For list archives: http://lists.mysql.com/mysql
To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED]
