Can your system deliver the required performance – HP StoreEver Ultrium Tape Drives User Manual
Page 47
Can your system deliver the required performance?
NOTE:
The compression ratio for LTO-6 is 2.5:1. For all earlier models the compression ratio is
2:1.
•
The HP LTO–6 Ultrium 6650 tape drive can write uncompressed data at up to 160 MB/sec
(576 GB/hour).
•
The HP LTO–5 Ultrium 3280 tape drive can write uncompressed data at up to 140 MB/sec
(504 GB/hour).
•
The HP LTO–4 Ultrium 1840 tape drive can write uncompressed data at up to 120 MB/sec
(432 GB/hour).
To obtain this performance it is essential that your whole system can deliver this performance. In
most cases, the backup application will provide details of the average time taken at the end of the
backup.
Typical areas where bottlenecks can occur are:
•
Disk subsystem
A single spindle disk may not be able to deliver good data throughput at poor compression
ratios. Best practice to ensure good throughput is to utilize multiple disk spindles or data
sources.
•
System architecture
Be aware of the architecture of your data protection environment.
The aggregation of multiple client sources over a network provides a good way of delivering
good performance, but anything less than Gigabit Ethernet will limit performance for LTO
Ultrium tape drives.
Some enterprise class backup applications can be made to interleave data from multiple
sources, such as clients or disks, to keep the tape drive working at optimum performance.
•
Tape media type
The data cartridge should match the specification of the tape drive. A lower specification will
have a lower transfer speed (see
). Use:
◦
HP LTO–6 Ultrium 6.25 TB RW or Ultrium 6.25 TB WORM cartridges with HP LTO–6
Ultrium 6650 tape drives
◦
HP LTO–5 Ultrium 3 TB RW or Ultrium 3 TB WORM cartridges with HP LTO–5 Ultrium
3280 tape drives
◦
HP LTO–4 Ultrium 1.6 TB RW or Ultrium 1.6 TB WORM cartridges with HP LTO–4 Ultrium
1840 tape drives
•
Data and file types
The type of data being backed up or restored can affect performance. Typically, small files
incur greater overhead in processing and access than large files. Equally, data that is not
compressible will always limit the speed at which the drive can write/read data. You will
achieve no more than native rates with uncompressible data.
Examples of files that compress well are plain text files, spreadsheets; those that compress
poorly are those that are either compressed as part of their format (such as, JPEG photographic
files) or stored as compressed (such as, .ZIP files or .gz/.Z files on UNIX platforms).
Performance Assessment Tools
47