Just yesterday someone asked:
"If you have a large data set in SSIS 2005 to process. Which way is faster to process the dataset in one data flow or if you can segregate the data flow in to 2 or 3 branches?"
What you are basically talking about is multi-threading.
SQL 2008 does a much better job of multi-threading than 2005. You really need to read this so that you understand what 2005 is really doing when you “think” you are spinning off multiple threads.
If you can bulk insert into an empty table for fast load then you might be better off taking this route with one data flow. The gains there are tremendous. Make sure you understand what fast load means and the requirements before you check the box and assume you will be doing fast loads.
In short there are possibilities that can speed up your load. Having plenty of CPU, IO, and memory (basically no hardware bottlenecks) is essential to see any performance gains with multi-threading.
"If you have a large data set in SSIS 2005 to process. Which way is faster to process the dataset in one data flow or if you can segregate the data flow in to 2 or 3 branches?"
What you are basically talking about is multi-threading.
SQL 2008 does a much better job of multi-threading than 2005. You really need to read this so that you understand what 2005 is really doing when you “think” you are spinning off multiple threads.
If you can bulk insert into an empty table for fast load then you might be better off taking this route with one data flow. The gains there are tremendous. Make sure you understand what fast load means and the requirements before you check the box and assume you will be doing fast loads.
In short there are possibilities that can speed up your load. Having plenty of CPU, IO, and memory (basically no hardware bottlenecks) is essential to see any performance gains with multi-threading.
Comments
Post a Comment