home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: biz.sco.general
- From: fred@genesis.demon.co.uk (Lawrence Kirby)
- Path: sparky!uunet!pipex!demon!genesis.demon.co.uk!fred
- Subject: Re: SCO Xenix w/SCSI Performance Testing/Results
- Distribution: world
- References: <74@consult.UUCP>
- Organization: BT
- X-Mailer: Simple NEWS 1.90 (ka9q DIS 1.19)
- Lines: 63
- Date: Fri, 1 Jan 1993 16:26:35 +0000
- Message-ID: <725905595snz@genesis.demon.co.uk>
- Sender: usenet@demon.co.uk
-
- In article <74@consult.UUCP> bob@consult.UUCP writes:
-
- >
- >> OLD MACHINE NEW MACHINE
- >> 386/33 w/8mb memory 64kcache 486/DX266 w/16mb memory 256k cache
- >> Ultrastor 12F ESDI controller Adaptec 1542 SCSI controller
- >> Maxtor 338mb drive (~ 16ms) HP 667 SCSI-2 drive (~13ms)
- >> CMS Enhancement 250mb tape Archive ST-525 mb tape (SCSI)
- >> SCO Xenix 2.3.2 SCO Xenix 2.3.4 GT
- >> Tape backup results when using CTAR
- >> Typical: 2.3mb/min Typical: 5.8mb/min
- >>
- >Results:
- > -----Old System------ -----New System------
- >DEVICE BLOCKSIZE COUNT REALTIME USER KERNEL REALTIME USER KERNEL
- >====== ========= ===== ======== ==== ====== ======== ==== ======
- >rhd0a 2k 8192 25.4 0.2 5.0 38.8 0.2 1.6
- >hd0a 2k 8192 38.3 0.5 15.1 59.9 0.0 26.7
- >rhd0a 256k 64 25.6 0.0 3.8 36.8 0.0 1.3
- >hd0a 256k 64 25.6 0.0 12.9 59.6 0.0 26.5
- >rroot 2k 8192 26.8 0.3 5.0 38.8 0.1 1.9
- >root 2k 8192 34.4 0.5 15.6 59.6 0.1 26.9
- >rroot 256k 64 25.9 0.0 3.8 36.9 0.0 1.8
- >root 256k 64 25.9 0.0 13.0 59.6 0.0 26.7
- >rhd00 2k 8192 25.4 0.5 4.7 38.8 0.1 1.8
- >hd00 2k 8192 34.7 0.5 14.5 59.6 0.1 27.0
- >rhd00 256k 64 25.7 0.0 3.7 36.9 0.0 1.9
- >hd00 256k 64 27.0 0.0 13.1 59.7 0.0 27.0
- >
-
- I did some similar tests on 2 SCO Unix 3.2.2 systems.
-
- SYSTEM 1 SYSTEM 2
- 486/33 8MB memory 64Kcacke 486/25 8MB memory, no external cache
- Ultrastor 12F ESDI controller Ultrastor 14F SCSI controller
- Maxtor 338mb drive (~ 16ms) Quantum 425S FAST SCSI-2 drive (256K
- cache)
-
-
- ------SYSTEM 1------ ------SYSTEM 2-------
- DEVICE BLOCKSIZE COUNT REALTIME USER KERNEL REALTIME USER KERNEL
- ====== ========= ===== ======== ==== ====== ======== ==== ======
- rhd0a 2k 8192 25.0 0.2 12.3 40.8 0.6 15.5
- hd0a 2k 8192 25.4 0.2 13.4 33.3 0.5 10.9
- rhd0a 256k 64 23.8 0.0 7.8 20.3 0.0 4.1
- hd0a 256k 64 25.2 0.0 11.5 33.2 0.0 7.8
- rroot 2k 8192 25.3 0.2 12.4 42.1 0.6 15.7
- root 2k 8192 24.9 0.1 13.1 33.1 0.5 10.8
- rroot 256k 64 24.5 0.0 7.3 19.3 0.0 4.3
- root 256k 64 25.0 0.0 11.7 35.0 0.0 7.7
- rhd00 2k 8192 25.5 0.2 12.0 41.3 0.4 16.1
- hd00 2k 8192 24.8 0.1 12.9 34.9 0.4 10.9
- rhd00 256k 64 23.9 0.0 7.8 20.1 0.0 4.3
- hd00 256k 64 24.8 0.0 11.7 33.3 0.0 7.5
- file 2k 8192 23.5 0.1 12.8 18.3 0.5 14.1
- file 256k 64 23.5 0.0 11.3 16.3 0.0 10.5
-
- 'file' is a large, contiguous file residing on an AFS filesystem. What's
- interesting is that REALTIME is always better using AFS than using a disk
- device directly, especially for small block sizes. Perhaps AFS has some
- inherent lookahead. AFS doesn't seem to impose a measurable overhead
- compared to block devices.
-
- On system 2 there seems to be a big overhead for small block sizes whereas on
- system 1 block size only seems to make a significant difference on raw
- devices.
-
- I did do some informal comparisons between the 14F and 1542B some time ago.
- The 14F does appear faster, especially with a FAST SCSI drive (the 1542B
- doesn't support FAST).
-
- -----------------------------------------
- Lawrence Kirby | fred@genesis.demon.co.uk
- Wilts, England | 70734.126@compuserve.com
- -----------------------------------------
-