Advanced Backup to Disk Performance White Paper

Compressible (2:1) test data was saved with expected transfer rates as described in the drive’s
specifications:
• 35 MB/s with SDLT 320
• 60 MB/s with Ultrium 460
• 157 MB/s with Ultrium 960
For other test data, the Ultrium 460 tape drive showed exceptional results. The “NULL” test data was
saved with 112 MB/s which is very good compared to 60 MB/s for compressible and 30 MB/s for
native data.
Please check
Table1 and Table3.
MSA Read and Write Tests
The MSA read and write tests were executed using the HPReadData/HPCreateData PAT utilities,
which are downloadable from
http://www.hp.com/support/pat. The results provide information about
the disk array performance and finally good estimates for backups and restores.
MSA1000
The MSA 1000 RAID 1+0 configuration meets the requirements of production environments (fast and
secure). Therefore, only this setup was created and tested. Please check chapter
HP StorageWorks
Modular Smart Array 1000 (MSA1000).
For typical files, one single LUN was written with 76,53 MB/s and read with 45,70 MB/s. For 4
LUNs simultaneously, the read performance was 93,58 MB/s. The maximum read performance is a
good prerequisite for feeding two SDLT 320 (35 MB/s, 2:1 comp.) and one Ultrium 460 (60 MB/s,
2:1 comp.). But it shows also that this disk array configuration cannot completely utilize the high-
performance Ultrium 960 (157 MB/s, 2:1 comp.). Please check
Table4 and Table6.
For small files, one single LUN was written with 14,22 MB/s and read with 8,12 MB/s. For 4 LUNs
simultaneously, the read performance was 14,25 MB/s. These results show that the direct backup to
tape would suffer from the poor read performance. On the other hand, it is a good reason to use the
Advanced Backup to Disk feature. Please check
Table4 and Table6.
MSA1500
The MSA 1500 results show that not only different RAID levels configurations influence the
performance. In addition, the number of configured logical arrays and LUNs played a major role to
achieve a better performance. Some combinations were created and tested. Please check chapter
HP
StorageWorks Modular Smart Array 1500 (MSA1500), Table5 and Table7.
The maximal performance for writing was 73,35 MB/s and for reading 66,13 MB/s.
RAID 0 was tested only for determining maximum performance values and showing the impact of
configuring multiple logical arrays. The RAID 0 configuration showed a single LUN write transfer rate
of 73,35 MB/s with 1 logical array and 23,96 MB/s with 4 logical arrays. The big write difference
is based on the horizontal LUN organization. The configuration of 1 logical array distributes the LUN
across all physical 12 disks but the configuration of 4 logical arrays only across 3 physical disks. The
read transfer rate was almost the same (29,53 MB/s with 1 array and 29,26 MB/s with 4 arrays). If
simultaneously reading from 4 LUNs, the read transfer rate doubled (32,94 MB/s with 1 array and
66,13 MB/s with 4 arrays). This demonstrates that configuring single logical arrays with many
physical disks in RAID 0 could improve the write performance. It is important to understand how the
data is organized and when physical disks are accessed at the same time. Please note that RAID0 is
not recommended due to missing fault tolerance.
36