Jeffrey, For dynamic files with records where the record size is large relative to the key size, I would go with KEYONLY rather than KEYDATA. We have had clients whose files (as KEYDATA) have split repeatedly with large records, so that a file that was 7 gig on Monday would be 19 gig by Thursday, which made their backups take a huge amount of time. Setting the split type as KEYONLY, the same files increased in size, but nowhere near a gig, let alone 12 gig.
Susan Lynch ----- Original Message ----- From: "Jeffrey Butera" <[EMAIL PROTECTED]> To: <[email protected]> Sent: Wednesday, August 24, 2005 10:18 AM Subject: [U2] Dynamic files in Unidata > I have a few questions for those more comfortable with dynamic files in > unidata. Currently I have a statically hashed file: > > File name = H08.CR.FF.TEXT > Number of groups in file (modulo) = 13331 > Static hashing, hash type = 0 > Block size = 16384 > Number of records = 88030 > Total number of bytes = 119079718 > > It's got a 16K blocksize because the size of the records is all over the place > (from a few words to many pages of free form text). We're dumping in about > 50000 new records per year so the file size isn't bad now but in the next few > years I anticipate growth (1Gig+) that will most likely lead me to a dynamic > file for this application. > > My questions: > > 1) Can anyone shed insight on 'reasonable' split/merge loads? By reasonable, > I guess I mean I'd like something a little more aggresive than the defaults: > > SPLIT_LOAD=60 > MERGE_LOAD=40 > KEYDATA_SPLIT_LOAD=95 > KEYDATA_MERGE_LOAD=40 > > I've had someone in the past suggest a split/merge load of 20/10, but have no > basis for that suggestion. > > 2) Thoughts about KEYDATA/KEYONLY? I've read the documentation, but I'm > looking for real-world insight from those who have used dynamic files in > unidata. > > 3) Anything else I'm not thinking about but should? (Arguments to not use a > dynamic file are welcome.) Currently, my biggest headache is that resizing > is taking about 45 seconds - I can only imagine how long it would take if I > had 1,000,000 records in there. > > -- > Jeff Butera, Ph.D. > Administrative Systems > Hampshire College > [EMAIL PROTECTED] > 413-559-5556 > > "...our behavior matters more than the beliefs that we profess." > Elizabeth Deutsch Earle > ------- > u2-users mailing list > [email protected] > To unsubscribe please visit http://listserver.u2ug.org/ ------- u2-users mailing list [email protected] To unsubscribe please visit http://listserver.u2ug.org/
