Re: AIX/ORACLE - Raid devices + async I/O
Date: 28 Oct 1994 09:02:47 GMT
Message-ID: <38qenn$1lor_at_thebes.waikato.ac.nz>
Anton Dischner (dischner_at_med.uni-muenchen.de) wrote:
> > > In 4.1.1 max filesystem size is up to 64 GByte (using NBPI ratio 4096)
> > ^^^^
> > At the risk of sounding dumb...
> >
> > Whats NBPI ratio?
> >
> >
> No it's not dump. (To be precise: I dont know it).
I can now anser that. I got my copy of 4.1 today, and is the ratio of inodes to disk. ie 4096=1 inode/4096 byes. 512=1inode for every 512 byte block. I'm running with 1024 and 512byte blocks at the moment on a test news spool, + LZ compression. looking good so far, some groups actually take up less space on disk than actual data (I didn't think it would be that mich difference. Heres some figures for the first few groups n my spool...
alt.1d : 203 arts 287828 Bytes 258560 Disk alt.2600 : 479 arts 932270 Bytes 810496 Disk alt.2600.hope.announce : 1 arts 613 Bytes 1024 Disk alt.2600.hope.d : 6 arts 9834 Bytes 8704 Disk alt.2600.hope.tech : 2 arts 953 Bytes 1024 Disk alt.2600hz : 6 arts 23326 Bytes 15360 Disk alt.266 : 0 arts 0 Bytes 0 Disk alt.3d : 140 arts 208473 Bytes 183808 Disk alt.abortion.inequity : 356 arts 1132482 Bytes 852480 Disk alt.abuse-recovery : 0 arts 0 Bytes 0 Diskalt.abuse.transcendence : 52 arts 163188 Bytes 126464 Disk
PS. Performance is quite good also. On an R24 the Kernel (jfsc) process (Which I assume is the backgrouund compression/decompression engine) uses on average < 1% CPU while I am tarring and untarring from the original spool area via NFS.
-- ====================================================================== | Hamish Marson | | Systems Programmer | | | Computer Services | INTERNET h.marson_at_waikato.ac.nz | | University of Waikato | PHONE +64 7 8562889 xt 8181 | | New Zealand | FAX +64 7 8384066 | ===========Disclaimer :- Remember. You heard it here first.===========Received on Fri Oct 28 1994 - 10:02:47 CET