[ZODB-Dev] FileStorage database unrecoverable?
Toby Dickenson
tdickenson@geminidataloggers.com
Fri, 28 Mar 2003 08:22:51 +0000
On Friday 28 March 2003 12:58 am, Greg Czajkowski wrote:
> This brings up a question of mine, does packing time
> increase with the size of FileStorage? If so, is it
> linear?
Packing time starts out linear. memory usage is linear too, so packing time
will drop below linear for reasonably large data sets.
> Now how does PartitionedFileStorage compare?
There should be no substantial difference.
> The reason I am asking is my database will be growing
> to over 10GB within the next six months, it stores one
> OOBtree which consists of
> strings,lists,tuples,longs,ints, and dicts.
> I have been trying to find the best Storage solution
> in terms of first size, then speed of analyzing every
> single BTree item. (I dont need undo support, just
> read/write/del)
For a 10G FileStorage you might need up to 1G of ram to hold FileStorage's
index. If thats OK, then FileStorage will be hard to beat.
> Berkeley was out of the question because database grew
> way too large. Not enough memory for MappingStorage.
> Haven't tried DirectoryStorage.
> Read CompressedStorage
> has bugs,
Do you have evidence that CompressedStorage would help you much? I considered
adding native compression to DirectoryStorage, but my data sets showed little
to be gained.
> so I stuck with FileStorage.
--
Toby Dickenson
http://www.geminidataloggers.com/people/tdickenson