cancel
Showing results for 
Search instead for 
Did you mean: 

Check flow rate of a multistreaming backup to lto2-drive

Stephan_Langbei
Level 4
My next question is, how can I check the flow rate (troughput) to a lto2 drive witch was written by a multstreaming backuppolicy (6 clients concurrently). At backupexec I have a logfile witch tell me the troughput for each save diskdrive, but at netbackup-enterprise I miss this logfile. Where I can check that 6 clients backup simmultanly enough to bring the drive in the streaming mode?

Also I miss a logfile like the backupexec 9.x-logfile.

Enviroment (Netbackup Enterprise 5.1 on W2K-Server, Advanced Clients (W2K, W2003E), STK L180)
4 REPLIES 4

Stumpr2
Level 6
This is from a FAQ on tek-tips:

http://www.tek-tips.com/faqs.cfm?fid=4865

How do you detemine the "Actual" backup speed?
faq776-4865
Posted: Feb 20, 2004

How can you determine the "Actual" speed of a tape device attached to your media server?
bperror -server -hoursago or ...
bperror -server -s -e

Using either of those commands - Output the results to a text file as in > c:\results.txt.
Use grep to organize and sort the results ...
grep Kbytes/sec c:\results.txt |grep -v duplicate |grep -v restore |awk -F\, '{print $NF}'|awk -f AWKFILE| awk '{print ($3)}'> c:\kbytes.txt
grep Kbytes/sec c:\results.txt |grep -v duplicate |grep -v restore |awk -F\, '{print $NF}'|awk -f $AWKFILE| awk '{print ($5)}'> c:\mytes.txt

I am piping the results of the awk to two additional text files - c:\kbytes.txt and c:\mbytes.txt - This simply lists a number in either kilobytes or megabytes.

The AWKFILE is a standard file containing the following:

BEGIN {bytes=0;thruput=0;items=0}

$2=="Kbytes" && $3=="at" {bytes+=$1; thruput+=$4;items++}
$1=="total" && $2=="Kbytes" {bytes+=$3; thruput+=$5;items++}
$2=="total" && $5=="Kbytes" {bytes+=$4; thruput+=$7;items++}

END {
# print "Total Volume: " bytes/1024 " Mbytes";
if (items>0) {print "Average Throughput: " thruput/items " KBytes/sec "int(((thruput/items)*60)/1024) " MBytes/min"}
}

That is how it tabulates the results. With these command you can easily compile one file to perform everything in one step, including getting the results for ALL your media servers in one go. I use this in a small VB application that I wrote up and it gives the user an option to select a media server, specify a time frame or a time period and then view the results which is in kbytes/sec or mbytes/min.

NOTE from Bob:
You can use the same theory and modify for whatever platform. There are a lot of scripts available in this community. I expect you will get lots of help on this topic :)Message was edited by:
Bob Stump

Stumpr2
Level 6
Another way would be to export the activity monitor to a spreadsheet and do the calculations for all the streams for total Kbytes/active elapsed time. The spreadsheet eill also allow you to do daily comparisons while changing max fragment size, number of streams, and the tuning of the BUFFERS.

This technote is old but the theory still works:
NET_BUFFER_SZ, SIZE_DATA_BUFFERS and NUMBER_DATA_BUFFERS - how they work and how to configure them
http://seer.support.veritas.com/docs/183702.htm

Methods to improve backup performance without adjusting VERITAS NetBackup (tm) buffer files
http://seer.support.veritas.com/docs/260719.htm

Performance Tuning Guides
There are two guides available for NetBackup performance tuning, one for UNIX and one for NT.

VERITAS NetBackup (tm) Performance Tuning Guide for UNIX Platforms: http://support.veritas.com/docs/240733

Tuning VERITAS NetBackup (tm) on NT systems for optimal performance: http://support.veritas.com/docs/248373Message was edited by:
Bob Stump

Stumpr2
Level 6
I realize I provided a lot of reference material. If you have had time to digest the information and if you have further questions, then please fire away.

Stephan_Langbei
Level 4
Sorry, but I have not tested your choice, because therefore I need a linux-PC (I will test it next week with a friend).