Tom, We have just upgraded from V2.7 to V3.2.0.6 in the weekend (I guess your 3.4.0.6 is a typo). Our HW is similar with Gbit (one), AIX 4.3.3 and ESS. We am not using the new option SORT_FILE. I tested with different MULTIPLEXING and ended with 5 as the best value for performance (probably depends on the layout of the filesystem in the ESS). 2 streams.
2 x 3590 E Tape drives (probably faster then yours but we only use 2), TSM 4.1 server, 4.2.1 client and rl_compression ~1.4 BKI1215I: Average transmission rate was 210.716 GB/h (59.937 MB/sec). BKI0020I: End of backint program at: Tue Jan 22 02:10:11 2002 . BKI0021I: Elapsed time: 01 h 50 min 51 sec . BKI0024I: Return code is: 0. Or old backups was slower but we have also moved to new HW and ESS in connection with the upgrade which all improved the performance. A backup test with the old SW and the new HW gave use ~2.5 hours but that also depends on which tape drive it hits (2 are Ultra SCSI attached, 1 is FW SCSI). Regards, Jan Norback Tivoli Certified Consultant: ADSM/TSM IBM Cert. Adv. Technical Expert RS/6000 AIX Atos Origin-IT (MS/DS/OSS Unix) VA-106, PO-box 218, 5600 MD Eindhoven, The Netherlands Groenewoudseweg 1 5621 BA Eindhoven, The Netherlands Email: [EMAIL PROTECTED] Tel: +31 (0)40-2780289 Mobile: +31 (0)6-12994492 Fax : +31 40 2783962 -----Original Message----- From: Kauffman, Tom [mailto:[EMAIL PROTECTED]] Sent: Monday, January 21, 2002 5:14 PM To: [EMAIL PROTECTED] Subject: Has anyone else seen a performance hit going from backint 2.4.17. 3 to TDP/R3 3.4.0.6? We migrated from backint 2.4.17.3 to Tivoli Data Protection for SAP just after the first of the year; no other changes in the supporting infrastructure. I'm seeing a 19 MB/sec drop in throughput. Has anyone else run into this? Is this just a stupid tuning parm thing I missed? We're running the same init<SID>.utl we ran prior to the upgrade. Short overview of config - SAP/R3 database on ESS, spread evenly over 3 fiber channels (not using SSD). DB server is an S7A, AIX 4.3.3-06; TSM server is an S7A, AIX 4.3.3-06, TSM 3.7.4.0. Running five seperate strams, two on one gigabit ethernet, three on the second gigbit ethernet. TSM configured to write the data directly to tape, IBM LTO in a 3584 library -- using five of our ten drives. We are using TDP/backint rl_compression; we are not using TSM client compression. This is from the 2.4.17.3 run: KI0020I: End of backint program at: Sun Dec 16 02:28:46 2001. BKI0022I: Average transmission rate was 251.035 GB/h (71.406 MB/sec). BKI2019I: Compression ratio for backup was 1.23. BKI0021I: Elapsed time: 02 h 25 min 35 sec. and this is from the 3.4.0.6 run: BKI1215I: Average transmission rate was 184.671 GB/h (52.529 MB/sec). BKI0020I: End of backint program at: Sun Jan 13 03:30:03 2002 . BKI0021I: Elapsed time: 03 h 27 min 14 sec . BKI0024I: Return code is: 0. This is a weekly off-line backup. We also run nightly on-line backups, to four tape drives. I'm seeing roughly the same hit, down from 60 MB/sec to 51 MB/sec. As can bee seen, I no longer get any benefit from running the additional tape drive for the off-line backup. So -- is this real, or is this a stupid config/tuning issue? Tom Kauffman NIBCO, Inc
