 |
» |
|
|
|
|
|
|
|
|
HP Tape Backup - HP StorageWorks DAT, DLT, DLT1, DLTVS80 and DLTVS160 Performance (Capacity and Speed)
Block size
All DLT products are optimized for use with large block sizes. Block size specifies the size of the chunks of data written onto tape.
ISSUE:
Block sizes of less than 64 KB for DLT1/DLT VS80/DLT VS160 and 32 KB for all other DAT and DLT drives can drastically increase the backup/restore time and severely affect the performance of the drive.
SOLUTION:
Most backup applications allow viewing and adjusting the block size used for a particular device. See below for advice on how to achieve this for CA ARCserve, Veritas Backup Exec and Tapeware.
CA ARCserve (for Windows NT)
See CA Technical Support Article "NT-AS6: Title: Changing the block size for a DLT drive":
http://support.cai.com/techbases/asnt/NASNT123.html
ARCserve for Windows NT version 6.x defaults to a write-block size of 16k for DLT drives. In order to increase performance, make the following changes in the Windows NT registry, using REGEDT32.EXE:
HKEY_LOCAL_MACHINE \SOFTWARE \Cheyenne \ARCserve \CurrentVersion \TapeEngine \Device# (The device number corresponds to the tape drive that will require a change in block size, this is located in the Device Manager)
Click Edit
, then select ADD
VALUE
: ValueName=DefaultBlockFactor Data type = Reg_Dword Data: 6, Hex
In addition to 6, the following values can also be selected:
0 = 512 (default) 1 = 1,024 (1 KB) 2 = 2,048 (2 KB) 3 = 4,096 (4 KB) 4 = 8,192 (8 KB) 5 = 16,384 (16 KB 6 = 32,768 (32 KB) 7 = 65,536 (64 KB)
After making the change, stop and start the ARCserve Tape Engine, then try another backup.
Veritas Backup Exec (for Windows NT and 2000)
Start Backup
Exec
.
Select the Properties
of the DLT1 tape drive in the Devices tab.
Click Configuration tab
.
Change the Block
size
to 64
KB
for DLT1 (See Figure 1) or change the Block size
to 32 KB
for all other DAT and DLT drives.
Figure 1: Block Size
Repeat individually for other tape drives.
The process would need to be repeated if the hardware set up is changed (for example: SCSI ID altered, new tape drives added)
Fbackup (HP-UX backup utility)
Fbackup uses the blocksperrecord option to specify the number of logical blocks that define a Fbackup record. Using the default blocksperrecord of 16, effectively 16k, may cause performance issues with products such as the DLT.
To increase the value change the blocksperrecord parameter from 16 to at least 32 (optimally 64) in the fbackup_config file
Tapeware
There are two factors to keep in mind regarding block size. The first is implementation of One Button Disaster Recovery (OBDR) requires a block size 2048K. If altering the block size from this value, then disaster recovery will not be restorable. The second is that different product types see different benefits from changes in block size.
DDS drives can adapt to block sizes with little or no performance impact; thus, increasing the block size would yield minimal improvements and render a disaster recovery unusable. For this reason the block size for a DDS drive is default of 2048K and HP suggests leaving it.
DLT tape drives operate most efficiently with a block size of 32 or 64K. In addition, there is no implementation of disaster recovery on the DLT products, so changing the block size from 2048k is not an issue.
TapeWare, by default, uses a blocksize of 32K for all DLT tape drives. If experiencing performance problems with a DLT tape drive, it may be possible to increase the backup speed by increasing the blocksize to 64K. To do this, close TapeWare and stop the TapeWare service if it is running. Then, edit the TWTAPDEV.INI FILE in the following way:
Open the file in a text editor, such as Notepad, and find the following section:
[dltDevices]
readToWriteDelay=50
deviceBlockSize=32768
mediaClass=4
developmentLevel=3
Change the line deviceBlockSize=32768
to deviceBlockSize=65536
Save the file and restart TapeWare.
LTO devices are also limited to 2048K, if wishing to implement disaster recovery. However, the LTO products handle data differently than DLT and DDS, so increasing the block size would not improve performance very much.
Network versus local backup
ISSUE:
Backing up across the network will almost certainly cause the speed of backups to be reduced.
SOLUTION:
Backing up local data (data residing on disk drives attached to the same machine as the tape drive, but not necessarily the same SCSI bus) will always improve this. External tape drives can be moved between machines to achieve this or dedicated tape drives should be attached to each.
Slow or overloaded server
ISSUE:
Tape drive performance will be impaired if the server is slow or overloaded.
SOLUTION:
Ensure as few processes as possible are running while a backup is being performed.
|
NOTE:
|
Backup operations and virus scans should not
be performed at the same time. |
Disk drives on the same SCSI bus as the tape drive
Be very careful when mixing different types of SCSI devices on the same bus. The chances are that newer hard drives will be a LVD device, and when attached with an older tape devices it’s performance can be halved.
In addition, having the Tape device and a hard drive on the same SCSI bus can potentially cause problems, as both will contend for bandwidth.
Each device on the SCSI bus has its own unique number (
Click here for more information on SCSI ID
). These are numbered either 0 to 7 or 0 to 16, depending on the type of SCSI used. The priority that a device has on the SCSI bus is based on its ID number. For the first 8 IDs, higher numbers have higher priority, so 7 is the highest and 0 the lowest. For Wide SCSI, the additional IDs from 8 to 15 again have the highest number as the highest priority, but the entire sequence is lower priority than the numbers from 0 to 7. So the overall priority sequence for wide SCSI is 7, 6, 5, 4, 3, 2, 1, 0, 15, 14, 13, 12, 11, 10, 9, 8. The priority levels are used to determine which device can have control of the SCSI bus. If more than one device wants control at the same time, the higher-priority device will "win", while the lower-priority device will have to wait.
Hard drives tend to monopolize the bus as they are always being accessed. If the hard drive has a high priority then it may effectively deny the tape drive the information that it needs causing performance and read/write issues.
The disk could be set to a lower priority SCSI ID such as 0 or 1 but even then the issues with priority may cause problems with the tape drive. A high priority hard drive will generally result in a low performance tape backup and a high priority tape drive will result in a low performance system, especially during backups.
The best solution, in all cases, is to put the tape drive onto its own SCSI bus with its own host bus controller.
Push agents
Are push agents (also called agent accelerators) being used? Push agents maximize back-up performance by providing source level compression and distributed processing to reduce network traffic. Push agents can usually be purchased as an option from the manufacturer of the back-up software.
Virus scans
Virus scans can have very real impact on the speed of a backup and possibly cause contention errors should both the backup application and virus software require access to the same files at the same time.
While real time scans can potentially be a problem, a file by file scan of the entire system can slow any backup down considerably and cause streaming problems.
These system wide scans are often scheduled to run during “off-peak�? times, which is often the same time that backup jobs are started.
It is advised that virus scans and tape backups be scheduled for different times.
Media usage
ISSUE:
If all else fails, it is possible the media could be at fault.
SOLUTION:
Try replacing the media.
|
|
Printable version
