January 30, 2011 at 11:04 am
HI Im working as a SQL DBA where im currently supporting a lage volume of data migration from a legacy system to a brand new product.
One of the migrations consist of a large volume of data with billions of rows of data in some staging tables. I have a couple of questions here which im hoping the experts can contribute to in terms of performance tuning.
1)We have created a number of indexes to improve the read time of some queries. However these tables then have large inserts going into them. Does indexing not slow down the insert.
2)This specific migration is going to be running consecutively for over 24 hours. As a test I did a rebuild and reorganise half way through and this helped. Is this something valid to suggest in production.
Thanks
January 30, 2011 at 10:37 pm
avinash jobanputra (1/30/2011)
1)We have created a number of indexes to improve the read time of some queries. However these tables then have large inserts going into them. Does indexing not slow down the insert.
Yes. Indexing is a tradeoff. Test and see what kind of hit you take on the inserts. Usually a small number of indexes is fine, but do test.
2)This specific migration is going to be running consecutively for over 24 hours. As a test I did a rebuild and reorganise half way through and this helped. Is this something valid to suggest in production.
Absolutely. You should be doing regular index maintenance. Not rebuild and reorg, the latter is redundant, but something. This script's good if the DB is large. http://sqlfool.com/2010/04/index-defrag-script-v4-0/
Gail Shaw
Microsoft Certified Master: SQL Server, MVP, M.Sc (Comp Sci)
SQL In The Wild: Discussions on DB performance with occasional diversions into recoverability
January 31, 2011 at 2:43 am
When you are transfering large data to new table; I suggest you create just basic structure of the table i.e. without indexes, rules etc.
Once transfer is done you can create the required indexes depending on your queries.
This will be much faster.
HTH
---------------------------------------------------
"Thare are only 10 types of people in the world:
Those who understand binary, and those who don't."
January 31, 2011 at 5:57 am
free_mascot (1/31/2011)
When you are transfering large data to new table; I suggest you create just basic structure of the table i.e. without indexes, rules etc.Once transfer is done you can create the required indexes depending on your queries.
This will be much faster.
HTH
Maybe, maybe not. Inserts into tables with a well-constructed clustered index are faster than inserts into a heap table. It's all the additional non-clustered indexes and foreign key constraints that can slow down inserts. But even these may benefit inserts depending on how the inserts are being done. You really have to address each situation individually and not use a blanket approach.
"The credit belongs to the man who is actually in the arena, whose face is marred by dust and sweat and blood"
- Theodore Roosevelt
Author of:
SQL Server Execution Plans
SQL Server Query Performance Tuning
January 31, 2011 at 7:45 am
Given that you have massive volumes of data to work with and yet are asking basic questions I really must encourage you to get a professional to help you with your migration efforts.
I will throw out that it can often be much more efficient to drop (some or all) indexes on tables that are going to have large amounts of data added to them in an ETL process and then rebuild them after load completion.
Best,
Kevin G. Boles
SQL Server Consultant
SQL MVP 2007-2012
TheSQLGuru on googles mail service
Viewing 5 posts - 1 through 4 (of 4 total)
You must be logged in to reply to this topic. Login to reply