Printable View
Hi all,
I need your advise about planning database storage.
We have the database with partitioning for IOT tables.
We used IOT for time-series data.
These tables belong one tablespace.The tablespaces was
split by 2000 MB data files and total size is 68GB ..
We are going to create new database with the same schema.
Should we create one or three big data files or create a lot small.
We will use Linux 64-bit.What is the better way for IOT with partitioning ?
Thanks,
Andrey
The size of your data files is more a factor of MTTR (mean time to recovery). Takes longer to recover a larger data file.
A couple weeks ago I heard of a guy with a tablespace consisting of one 500 Gigabyte datafile. (10g "bigfile") .. running on a mac.
Guess it's possible, but not something I would do.
gee whizz, bet that took some time to create.Quote:
Originally posted by KenEwald
The size of your data files is more a factor of MTTR (mean time to recovery). Takes longer to recover a larger data file.
A couple weeks ago I heard of a guy with a tablespace consisting of one 500 Gigabyte datafile. (10g "bigfile") .. running on a mac.
Guess it's possible, but not something I would do.
Also if you have separate disks, just made sure you spread it across all of them
Would that also apply with RMAN?Quote:
Originally posted by KenEwald
The size of your data files is more a factor of MTTR (mean time to recovery). Takes longer to recover a larger data file.
yes, if you need to recover a 10Gb from tape, will take longer than a 2gb file for example. If you are restoring a whole database wont make much difference
sorta of maybe - depeneds on how many blocks in the datafile ever contained dataQuote:
Originally posted by davey23uk
yes, if you need to recover a 10Gb from tape, will take longer than a 2gb file for example. If you are restoring a whole database wont make much difference
- so if you have a new 10gb df with zero blocks it could conceivably recover faster than a 2gb df all packed up with blocks that at some time had/has data