Compaq ProLiant 1000 Compaq Backup and Recovery for Microsoft SQL Server 6.X - Page 34

Compaq Backup and Recovery for Microsoft SQL Server 6.x, randomly

Get Compaq ProLiant 1000 PDF manuals and user guides View all Compaq ProLiant 1000 manuals
Add to My Manuals
Save this manual to your list of manuals

Page 34 highlights

Compaq Backup and Recovery for Microsoft SQL Server 6.x Chart 3 - Dump to Single Tape Device Page 34 Looking at the result sets, the steepest differences can be seen when comparing 35/70-GB DLT performance to 15/30-GB DLT performance: the 35/70 is almost 4x faster. This vast difference is understood when considering the hardware enhancements in the 35/70 vs. the 15/30, such as twice the number of simultaneous read/write channels (4 vs. 2), increased tape speed, increased tracks-per-inch on the media, enhanced interface (Fast-Wide SCSI-2), and a larger on-board buffer cache (8 MB vs. 2 MB). Other differences in performance are seen when comparing SQL Server 6.0 results with SQL Server 6.5 results. The explanation for this essentially comes down to the issue of tape block size (please see the earlier section entitled Block Size). The larger read/write blocks specified by SQL Server 6.5 allow the DLT to record the data at a much faster rate - especially in the case of the higher-speed 35/70-GB drives where SQL 6.5 performs twice as fast. The smaller block size used by SQL Server 6.0 also prevents it from realizing any benefit from the hardware compression feature of the DLT drives. With SQL Server 6.5 we can realize the benefits of hardware compression. From the above results however, these benefits seem limited since the performance with compression only slightly exceeds that without compression (the non-compressed throughput of the DLT is specified at 4.5 GB/hr and 18.0 GB/hr for the 15/30-GB and 35/70-GB drives, respectively). We must remember however, that the data used for our testing lends itself to compression less than would the typical numeric and textual data found in most production databases (our database was loaded with randomly generated numbers and text), therefore providing very conservative results. The compression achieved by the tape drive was measured to be around 15% (a 1.17:1 ratio); a greater increase in both capacity and throughput could be expected with a more compressible data set. Users should test the "compressability" of their own data using a representative data set52. 52 See the earlier section on Estimating Hardware Compression Ratios.

  • 1
  • 2
  • 3
  • 4
  • 5
  • 6
  • 7
  • 8
  • 9
  • 10
  • 11
  • 12
  • 13
  • 14
  • 15
  • 16
  • 17
  • 18
  • 19
  • 20
  • 21
  • 22
  • 23
  • 24
  • 25
  • 26
  • 27
  • 28
  • 29
  • 30
  • 31
  • 32
  • 33
  • 34
  • 35
  • 36
  • 37
  • 38
  • 39
  • 40
  • 41
  • 42
  • 43
  • 44
  • 45
  • 46
  • 47
  • 48
  • 49
  • 50
  • 51
  • 52
  • 53
  • 54
  • 55
  • 56
  • 57
  • 58
  • 59
  • 60
  • 61
  • 62
  • 63
  • 64
  • 65
  • 66
  • 67
  • 68
  • 69
  • 70
  • 71
  • 72
  • 73
  • 74
  • 75
  • 76
  • 77
  • 78
  • 79
  • 80
  • 81
Compaq Backup and Recovery for Microsoft SQL Server 6.x
Page
34
Chart 3 - Dump to Single Tape Device
Looking at the result sets, the steepest differences can be seen when comparing 35/70-GB DLT
performance to 15/30-GB DLT performance: the 35/70 is almost 4x faster.
This vast difference is
understood when considering the hardware enhancements in the 35/70 vs. the 15/30, such as twice the
number of simultaneous read/write channels (4 vs. 2), increased tape speed, increased tracks-per-inch
on the media, enhanced interface (Fast-Wide SCSI-2), and a larger on-board buffer cache (8 MB vs. 2
MB).
Other differences in performance are seen when comparing SQL Server 6.0 results with SQL Server
6.5 results.
The explanation for this essentially comes down to the issue of tape block size (please see
the earlier section entitled
Block Size
).
The larger read/write blocks specified by SQL Server 6.5
allow the DLT to record the data at a much faster rate - especially in the case of the higher-speed
35/70-GB drives where SQL 6.5 performs twice as fast.
The smaller block size used by SQL Server
6.0 also prevents it from realizing any benefit from the hardware compression feature of the DLT
drives.
With SQL Server 6.5 we can realize the benefits of hardware compression.
From the above results
however, these benefits seem limited since the performance with compression only slightly exceeds
that without compression (the non-compressed throughput of the DLT is specified at 4.5 GB/hr and
18.0 GB/hr for the 15/30-GB and 35/70-GB drives, respectively).
We must remember however, that
the data used for our testing lends itself to compression less than would the typical numeric and textual
data found in most production databases (our database was loaded with
randomly
generated numbers
and text), therefore providing very conservative results.
The compression achieved by the tape drive
was measured to be around 15% (a 1.17:1 ratio); a greater increase in both capacity and throughput
could be expected with a more compressible data set.
Users should test the “compressability” of their
own data using a representative data set
52
.
52
See the earlier section on
Estimating Hardware Compression Ratios.