Very Computer

One simple question : is SQL Server 7 possible to work with five hundred
millions of records (500 000 000). Record has cca 10 fields and most of them
are integer. Server is strong enough (2G RAM, dual processor P3, lot of HDD
space).

Thanks,
Drago. :-)

Hi!

Quote:> One simple question : is SQL Server 7 possible to work with five hundred
> millions of records (500 000 000). Record has cca 10 fields and most of
them
> are integer. Server is strong enough (2G RAM, dual processor P3, lot of
HDD
> space).

Of course it's possible, number of rows is limited only by available space.
Just be sure to have proper indices, a table scan could take quite a long...

HTH,
Dejan Sarka

Do you have such experience (with one table with so much records) ?
I need info to be able to make right decision on current project. Thanx.

Drago. :-)

Quote:> Hi!
> > One simple question : is SQL Server 7 possible to work with five hundred
> > millions of records (500 000 000). Record has cca 10 fields and most of
> them
> > are integer. Server is strong enough (2G RAM, dual processor P3, lot of
> HDD
> > space).
> Of course it's possible, number of rows is limited only by available
space.
> Just be sure to have proper indices, a table scan could take quite a
long...

> HTH,
> Dejan Sarka

Quote:> Do you have such experience (with one table with so much records) ?
> I need info to be able to make right decision on current project. Thanx.

i have over 4 mln rows in one table with a couple of indexes, total space is
about 500MB. table contains historical data over 2 yrs period. report is run
against this table covering about 1000 rows at a time and takes about several
seconds to complete.

BUT eventually (after a sudden power off) database became corrupted, with a lot
of extents allocated to nowhere and indexes out of order. dbcc checkdb said that
it can't repair without first repairing some error which it can't repair
either... so i deleted almost everything fom that table (4.6 mln rows in 1 hr)
and run dbcc again. this time it succedded and everything else seems to be OK.

server Compaq PL1600, 1x PII/450, RAID 5 (9.1GB x 6), 320MB

hope it gives you some clue

Quote:> Do you have such experience (with one table with so much records) ?
> I need info to be able to make right decision on current project. Thanx.

I worked on db optimization project for the biggest mobile phone services
provider in our country. All records from phone switches are imported to SQL
Server for billing and analysis and we are talking about 100 000 000 records
per month. And BTW, those records are much longer, not just ten columns
(from your description I suppose you are talking about 50GB of data, I had
nearly 100GB per month).
To persuade the customer that SQL Server is the right solution, I tested the
speed you can reach for inserts. On my desktop, one processor, one disk,
over LAN, nothing done for HW optimization, I succeeded to insert 2 000 000
rows in 10 minutes on SQL 7.0. So I suppose with better HW, using filegroups
etc., it is possible to do it ten times faster.
Anyway, be aware that dealing with so huge databases is not fun.

HTH,
Dejan Sarka