Advanced Backup to Disk Performance White Paper

Figure 36. Advanced Backup to Disk - Average CPU Utilization with 10 GB File Depot Size
Restore
The data was restored from the file library (MSA1500) or tape (SDLT/Ultrium) to the original location
(MSA1000).
The typical file restore of four LUNs (concurrency = 4) showed a transfer rate of 42,52 MB/s for the
file library, which was beaten by double SDLT 320 (43,12 MB/s) and one Ultrium 960 (51,00
MB/s).
The typical file restore of one single LUN (concurrency = 1) resulted in 23,37 MB/s for the SDLT 320,
35,97 MB/s for the Ultrium 460 and 51,80 MB/s for the Ultrium 960.
The typical file restore of one single LUN (concurrency = 4) was only tested for the Ultrium 960 to
demonstrate its fast speed. The Ultrium 960 showed exceptional 34,23 MB/s, which is more than
half of the full restore performance of 4 LUNs (51,00 MB/s). This can be explained by the maximal
Ultrium 960 performance of 160 MB/s (drive specification, 2:1 comp.), which theoretically provides
40 MB/s for 1:4 multiplexing.
The small file restore of four LUNs (concurrency = 4) resulted in 4,66 MB/s for the SDLT 320 and
9,12 MB/s for the Ultrium 960. Please note that this small file restore suffers from the slow Windows
NTFS. Tape drives cannot stream and must heavily enter the start/stop mode.
The restore of single 8 KB files from the file library was tested multiple times. This was executed due to
the fact that different files have different positions in file depot segments which affect the runtime. All
restores lasted less than 30 seconds each and were even possible within 10 seconds. Multiple single
file restores (one after the other) strongly benefit from these little runtimes. Single restores of small files
are much faster from disk than from tape. In this scenario, tape drives cannot compete due to the
time-consuming tape load/unload and positioning procedure.
40