on 2016 May 30 10:07 PM
hi folks, What is the reason why using a CREATE EVENT...BACKUP DATABASE uses the entire memory on a server vs the dbbackup command which uses only minimal RAM to perform the operation (SQLAnywhere 17), on the same server.
thanks much for any guidance, Cosmin
Request clarification before answering.
Another thought. You mentioned a scheduled backup.
When you schedule this, are you using a maintenance plan? Maintenance plans can do a fully validation operation and that can take lots of memory.
Again ... just a guess ... and if this does not help you may need to contact Product Support.
{alternatively providing the exact details of the scheduling mechanism and event/operation syntax used might show enough details so some one can do real diagnosis}
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
I can't think of any reason for that difference and I didn't find any bug reports that sound similar ...
Being full backups the two should behave almost identical unless the target device indicates that the server can run parallel backups. You could try adding "AUTO TUNE WRITERS OFF" to your backup statement to see if that has any effect.
If disabling parallel backups doesn't affect this then you may need to work through this with Product Support. It sounds like something that will need to be recreated to see what is really happening.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
If both are performing the same type of backup (full vs incremental, for example) then they should behave similarily. The main difference I can think of is that dbbackup can perform a client side backup, which may not benefit from the parallelism avaiable to server side backups; the latter could use more cache but probably not the whole machine's worth.
But that is not a very informed understanding about what you are seeing. A fuller description of the specific natures of the 2 different backups you are performing and some idea about how much memory is your "entire memory on a server" might reveal more sources of differences. [... for example is your server side backup doing something specific with the checkpoint log? ... or auto tuners? ... or writing to different media? ... or is being ran at a heavy transaction time?]
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
hi Nick, thanks a big bunch for the fast reply. Both backups are full. The server has 128gb RAM and rarely goes over 5-6gb of RAM used, both while in use or when executing the dbbackup command. When scheduling the BACKUP DATABASE, also full, the command invoked starts taking up all the available RAM. What should I be looking at to limit the RAM used?
thanks again, Cosmin
User | Count |
---|---|
60 | |
10 | |
8 | |
8 | |
7 | |
6 | |
6 | |
5 | |
5 | |
4 |
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.