02-26-2012 04:25 AM
Backup ratio is very slow , most of the clients are giving only 10MB/Sec. I am having 2 media servers and master servers are in cluster.Media servers is having 10G ethernet cards, clients having 1Gb cards.
Slowness is more in windows clients. Some of the linux clients are giving 1Gb also.
Network drivers are updated version. What could be the problem.
Solved! Go to Solution.
02-26-2012 05:22 AM
Take a look in the "Symantec NetBackup™ 7.0 - 7.1 Backup Planning and Performance Tuning Guide". There is chapter about backup tuning, this will provide some tools to get you going.
http://www.symantec.com/docs/DOC4483
PS: Have you done the buffer tuning in Netbackup ?
http://www.mass.dk/netbackup/guides/49-netbackup-buffer-tuning.html
02-26-2012 05:22 AM
Take a look in the "Symantec NetBackup™ 7.0 - 7.1 Backup Planning and Performance Tuning Guide". There is chapter about backup tuning, this will provide some tools to get you going.
http://www.symantec.com/docs/DOC4483
PS: Have you done the buffer tuning in Netbackup ?
http://www.mass.dk/netbackup/guides/49-netbackup-buffer-tuning.html
02-26-2012 06:30 AM
>> What could be the problem.
Network! Have you tried FTP'ing a large file from a slow client to the media server and time how long it takes. Then work out how many Mb/s or Gb/s it is transferring at?
02-26-2012 09:24 PM
The doc that Nicolai referenced above will tell you to start with testing client read speed from disk. Use bpbkar on client for this test. Details in the pdf.
More info in this TN: http://www.symantec.com/docs/TECH8326
02-27-2012 08:12 AM
First test the speeds that your media servers can handle if you are writing to tape drives run a dd test to measure the speeds, this will remove netbackup from the equation just use:
#Mount a tape from a media server:
tpreq -m $TAPE -a r -d hcart -f /tmp/$TAPE
#Write data to the tape:
dd if=/dev/zero of=/tmp/$TAPE bs=256k count=100000 & >> $LIBNAME.dd_write.ksh
#Read the data from the tape:
dd if=/tmp/$TAPE of=/dev/null bs=256k count=100000 & >> $LIBNAME.dd_read.ksh
To measure the speed only open a second screen and start a iostat:
iostat -exnDl1 <rmt/#> 1
replace the rmt/# with your drive address and you will see the speeds.
This will tell you if the bottleneck is at netbackup or drives/san levels.
Regards.
03-09-2012 12:44 PM
Hi Rajeev,
What is the Network Setting for those client which is have low through-put.
Check the Network setting for those client having low through put on OS level as well as Switch Level.
If not then make it 100 mbps/Full duplex on both side and check backup speed.
Thanks
Bharat
03-10-2012 10:13 AM
That’s slower than a snail with arthritis. Turns out, NetBackup uses certain default values for shared memory buffer sizes
To increase the size of the buffers and the number of buffers, you have to create the following two files:
/usr/openv/netbackup/db/config/SIZE_DATA_BUFFERS
/usr/openv/netbackup/db/config/NUMBER_DATA_BUFFERS
Simply fill in the values that you need for each. A good number for SIZE_DATA_BUFFERS would be 131072 (128KB), or, if your tape drive supports it, 262144. You can also choose to have 16 such buffers by typing in this number in the NUMBER_DATA_BUFFERS file. Note that you may have to increase the shared mem limits on your system before doing this.
see below reference for setting Buffers.
http://www.symantec.com/business/support/index?page=content&id=TECH1724