Compaq ProLiant 1000 Compaq Backup and Recovery for Microsoft SQL Server 6.X - Page 32

Disk Controller Buses - saturation of Fast-Wide SCSI-2 buses 10/20 MB/sec

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 32 highlights

Compaq Backup and Recovery for Microsoft SQL Server 6.x Page 32 Compaq 4.3 GB Fast-Wide SCSI-2 disk drives (7200 RPM). Compaq External Storage cabinets for Fast-Wide disk drives. The system was set up as follows: All system boot and binary files were installed on a single, internal hard drive on the integrated SCSI-2 controller. The test database was created on a dedicated logical volume (array) on the Smart-2 controller to isolate the I/O occurring to and from the data files, and spanned both SCSI channels on the controller. RAID level 5, which stripes the data across the disks in the volume along with parity information, was used for fault tolerance on this array. A smaller RAID-1 volume held the 'master' database and the transaction log file for the test database. Note that the transaction log dump was not measured as part of the database dump tests (see section on Calculating Database Dump Throughput). All 15/30-GB DLT Tape drives used for testing database dump to tape devices were attached either to the integrated Fast-Wide SCSI-2 controller or to additional Wide-Ultra SCSI-2 controllers45, with no more than 2 drives per controller. All 35/70-GB DLT Tape drives used for the tests were attached to Wide-Ultra SCSI-2 controllers, 2 drives per controller. Bottlenecks that can impede overall throughput in the system model include: Source Side (Server): CPU - insufficient CPU cycles to process I/O requests or other backup related activity. System Bus(es) - saturation of peripheral bus-to-memory path when multiple high-throughput peripherals are used. Possible saturation of processor-to-memory path on SMP machines in CPUintensive environment (not usually a concern in backup environments). Peripheral (expansion) Bus - saturation of EISA (16.6/33 MB/sec46) or PCI bus (66/132 MB/sec) interfacing to peripherals (disk controllers). Disk Controller Bus(es) - saturation of Fast-Wide SCSI-2 bus(es) (10/20 MB/sec47) on Smart-2 controller which contains the database volume. Disks - insufficient spindles in database volume to service required number of read requests from application. Software - serialization or contention points in application code, insufficient threads in application, etc. Destination Side: 45As mentioned in the section referenced by the previous footnote, the integrated Fast-Wide SCSI-2 controller may be used in conjunction with additional, Wide-Ultra controllers without suffering an overall loss of throughput. 46 Numbers in parenthesis indicate maximum regular and burst mode data transfer rates. Actual rate at saturation is usually lower. 47 Numbers in parenthesis indicate maximum asynchronous and synchronous data transfer rates. Actual rate at saturation is usually lower.

  • 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
32
Compaq 4.3 GB Fast-Wide SCSI-2 disk drives (7200 RPM).
Compaq External Storage cabinets for Fast-Wide disk drives.
The system was set up as follows:
All system boot and binary files were installed on a single, internal hard drive on the integrated
SCSI-2 controller.
The test database was created on a dedicated logical volume (array) on the Smart-2 controller to
isolate the I/O occurring to and from the data files, and spanned both SCSI channels on the
controller.
RAID level 5, which stripes the data across the disks in the volume along with parity
information, was used for fault tolerance on this array.
A smaller RAID-1 volume held the
‘master’ database and the transaction log file for the test database.
Note that the transaction log
dump was not measured as part of the database dump tests (see section on
Calculating Database
Dump Throughput
).
All 15/30-GB DLT Tape drives used for testing database dump to tape devices were attached
either to the integrated Fast-Wide SCSI-2 controller or to additional Wide-Ultra SCSI-2
controllers
45
, with no more than 2 drives per controller.
All 35/70-GB DLT Tape drives used for the tests were attached to Wide-Ultra SCSI-2 controllers,
2 drives per controller.
Bottlenecks that can impede overall throughput in the system model include:
Source Side (Server):
CPU - insufficient CPU cycles to process I/O requests or other backup related activity.
System Bus(es) - saturation of peripheral bus-to-memory path when multiple high-throughput
peripherals are used.
Possible saturation of processor-to-memory path on SMP machines in CPU-
intensive environment (not usually a concern in backup environments).
Peripheral (expansion) Bus - saturation of EISA (16.6/33 MB/sec
46
) or PCI bus (66/132 MB/sec)
interfacing to peripherals (disk controllers).
Disk Controller Bus(es) - saturation of Fast-Wide SCSI-2 bus(es) (10/20 MB/sec
47
) on Smart-2
controller which contains the database volume.
Disks - insufficient spindles in database volume to service required number of read requests from
application.
Software - serialization or contention points in application code, insufficient threads in application,
etc.
Destination Side:
45
As mentioned in the section referenced by the previous footnote, the integrated Fast-Wide SCSI-2 controller
may be used in conjunction with additional, Wide-Ultra controllers without suffering an overall loss of
throughput.
46
Numbers in parenthesis indicate maximum regular and burst mode data transfer rates.
Actual rate at
saturation is usually lower.
47
Numbers in parenthesis indicate maximum asynchronous and synchronous data transfer rates.
Actual rate at
saturation is usually lower.