12-18-2013 10:43 AM - last edited on 12-18-2013 05:08 PM by Lisa198503
Previously my database restores on Data Protector v6.0 using a MSL6060 tape library with LTO3 tape drives would take about 4.5 hours. The same database restore now takes over 9 hours to complete on a newer Data Protector Cell Manager running v7.0 on a HP-UX server that is SAN attached to a MSL6060 tape library with LTO4 tape drives. I have been working with HP Support for several weeks trying to find out what is causing the problem and thought it might be useful to ask the community if anyone else has seen this issue. The problem with database restores is not confined to a particular db server, nor to a particular backup tape, nor to a particular LTO4 tape drive in the library. Its very odd that now that we are on a newer Cell Manager and a tape library with faster tape drives that our recover window doubled in length.
The most obvious difference is that the restores have 50 second pauses beween reading the data on the tape. The writes from the tape are varied between 10 and 20 seconds long but the pauses are very consistent.
In November 2013 I performed a Disaster Recovery test at Sungard where the same database restore on a IBM 3584 tape library only took 2 hours to complete. As I had recovered the Data Protector Cell Manager and the database/app servers from ignite backups and a copy of the DP filesystems, the configuration settings were the same at the DR test as they are in production (with the exception of course of having a different tape library at Sungard). The restore at the DR test did not exhibit any of the pauses. I feel if we can solve the mystery of the restore pausing for 50 seconds then we fix the issue. Any ideas?
P.S. This thread has been moved from Tape Libraries and Drives to Data Protector Support and News Forum. -HP Forum Moderator
12-18-2013 01:03 PM
Update to this problem. After making the following change recommended by HP Support, the time out periods are varying in length. There are still many 50 second pauses between restore writes, but now there are some 10 and 20 second pauses occurring.
Enable tape drive lock down and try restoring from the tape:
# scsimgr set_attr -d estape -a norewind_close_disabled=1