12-22-2015 04:34 PM - edited 12-22-2015 04:35 PM
I have SAS/ACCESS to PostgreSQL licensed.
I want to use PostgreSQL as a data store for my ETL jobstats.
I want this data store to reside on my compute server.
SAS provides a bundled instance of PostgreSQL, which it uses for the WIP database on that same compute tier.
I would like to hear opinions on whether to use the SAS-provided PostgreSQL instance for storing my own ETL jobstats data.
Or do you find I should install a separate instance of PostgreSQL for this data?
12-22-2015 04:39 PM
My gut feeling (not fact based!) is to use a separate instance.
The bundled one is a part of the SAS product/configuration, and could in some future change with a new SAS release, a dependancy your might don't want.
Also, if you need to trouble shoot some SAS content behaviour, you don't need to rule out explicitly your own use of the same database.
01-06-2016 10:20 AM
I personally agree completely with LinusH. I would use a different instance as well.
The reason is that the WIP db it has a performance tuning (and encoding, usually just UTF-8) configured for the kind of data and transactions it is expected to be handled by the shared services, VA and so.
Given this scenario, as I want the SAS system to work as optimal as possible, I would not include external factors or change the performance settings.
In other hand, I would chose to optimize the performance, encoding and colate settings for my ETL data and transactions too. Therefore I would create a different instance, with better conditions for my ETLs.
I am feeling curious too about other perspectives.