by Neil Pik » Thu, 18 Jun 1998 04:00:00
Dan,Quote:> Depending on you hardware and
> database setup (# of processors, separate index segments, dual-channel
> controllers, multiple controllers), you may be able to improve elapsed time
> by creating the non-clustered indexes in parallel.
Indices on the same table can't be created in parallel on SQL 6.5 due to
system table locks. (It works fine in SQL 7.0 and gives a HUGE improvement)
Neil Pike MVP/MCSE
Protech Computing Ltd
by Dan Guzma » Thu, 18 Jun 1998 04:00:00
This certainly makes sense, but I remember doing this a while back, maybe
under 4.21. Since I never benchmarked it, it's entirely possible the
subsequent index creates were blocked and actually ran consecutively.
Thanks for the correction.
Dan
Quote:>snip<
> Indices on the same table can't be created in parallel on SQL 6.5 due to
>system table locks. (It works fine in SQL 7.0 and gives a HUGE
improvement)
by Neil Pik » Fri, 19 Jun 1998 04:00:00
Dan,
That's right - under 4.x they ran consecutively (after each block was
released). With 6.5 I've actually had errors and subsequent indices
haven't even been created when I've run them in parallel.
Neil Pike MVP/MCSE
Protech Computing Ltd
by Lawrence M. Seldin, CMC, C » Fri, 19 Jun 1998 04:00:00
Sounds great, is SQL 7.0 available now?
- Larry
> Dan,
> Indices on the same table can't be created in parallel on SQL 6.5 due to
>system table locks. (It works fine in SQL 7.0 and gives a HUGE improvement)
> Neil Pike MVP/MCSE
> Protech Computing Ltd
http://ourworld.compuserve.com/homepages/seldin
by Neil Pik » Fri, 19 Jun 1998 04:00:00
Larry,
Nope - it will be in beta for quite a few more months. Release will
be by the end of the year.
Neil Pike MVP/MCSE
Protech Computing Ltd
by Lawrence M. Seldin, CMC, C » Sat, 20 Jun 1998 04:00:00
Neil,Quote:> Nope - it will be in beta for quite a few more months. Release will
>be by the end of the year.
Thanks for the info.
- Larry
Lawrence M. Seldin, CMC, CPC
Author of POWER TIPS FOR THE APPLE NEWTON and INTRODUCTION TO CSP
http://ourworld.compuserve.com/homepages/seldin
1. Brainstorm on making batch cycle perform better
Lawrence,
I'd do all the processing on a single fast machine - plenty of raid
with write-back caching controllers - with 4 processors. Probably the
slowest thing you're doing is transferring this data from machine to
machine over the network. As long as you keep the "batch" files and
the SQL files on separate spindles there won't be much contention.
Alternatively you could look at www.datajunction.com for a tool to do
the conversion and insert into SQL in one go.
Neil Pike MVP/MCSE
Protech Computing Ltd
3. Error Message for No Records
4. Using JOINS improve performance?
5. PRESS: Arbor announces Essbase Web Gateway
6. HOWTO: Improve SQL 6.5 Performance for very simple SQL
8. Improved performance with MS-SQL server?
9. How to improve view's performance?
10. improving performance thru indexing
11. Improving performance of Web server, SQL server
12. improving cube processing performance