[Progress Communities] [Progress OpenEdge ABL] Forum Post: RE: Surprising performance inconsistency (adding index)

Status
Not open for further replies.
D

dbeavon

Guest
In my case I was initially adding inactive indexes and then using "idxbuild". You are right that it was very fast and it fully utilized my CPU and disk. (I think the bottleneck was actually on the CPU cores, given the number of threads that I had requested in the proutil command.) Unfortunately we will NOT be able to do the "idxbuild" in production where we have replication enabled. My recollection is that the idxactivate didn't seem much faster than simply making the indexes initially active in the first place (via SQL92). I may have to test that again to be sure.... It is too bad that these types of schema operations are limited by the clockspeed of a single core. It has been quite a long time since multi-processor servers became normal. At these moments I realize how fortunate I am to do much of my development work on SQL Server, where parallel execution plans are generated on a regular basis for all types of different SQL queries, and DDL operations.

Continue reading...
 
Status
Not open for further replies.
Top