BS ArcServe 9.0 Backup slowdown!!!! 1

[NYCamc]

System Setup:
BS Arcserve 9.0 build 2000
Windows Server 2000 SP3
Dell Poweredge 2650
Adaptec 29160 SCSI adapter
ADIC Scalar 100 AIT 3
Intel PRO/1000 Gigabit Server adapter (Duplex-Full)

Backups were running great last week. Throughput was very high. I come in on Monday and the Saturday Full is still running! Throughput was reduced to almost nothing. Nothing changed in the environment from the Thursday differential to the Saturday Full. What has caused this slowdown? Anyone else experience dramatic decreases in throughtput after a couple of weeks of use. The stats for one of servers on the weekly job are below to illustrate the slowdown. Note the time and avg. throughput on the last job compared to the others. Local backups are just as slow. Drivers and firmware for all hardware are up to date. CA is no help whatsoever! Any ideas? Thanks for the help!






Saturday Full
03/01/2003
Totals For................... DOMINO1 ( 10.1.1.11 )
Total Session(s)............. 4
Total Directories............ 2,507
Total File(s)................ 34,706
Total Skip(s)................ 0
Total Size (Disk)............ 93,214.16 MB
Total Size (Media)........... 93,319.75 MB
Elapsed Time................. 1h 43m 44s
Average Throughput........... 899.58 MB/min



Monday Diff.
03/03/2003
Totals For................... DOMINO1 ( 10.1.1.11 )
Total Session(s)............. 4
Total Directories............ 2,323
Total File(s)................ 2,734
Total Skip(s)................ 0
Total Size (Disk)............ 87,720.58 MB
Total Size (Media)........... 87,735.43 MB
Elapsed Time................. 2h 55m 19s
Average Throughput........... 500.39 MB/min



Tuesday Diff.
03/04/2003
Totals For................... DOMINO1 ( 10.1.1.11 )
Total Session(s)............. 4
Total Directories............ 2,327
Total File(s)................ 2,800
Total Skip(s)................ 0
Total Size (Disk)............ 88,520.65 MB
Total Size (Media)........... 88,535.43 MB
Elapsed Time................. 1h 28m 1s
Average Throughput........... 1,005.87 MB/min


Wednesday Diff.
03/05/2003
Totals For................... DOMINO1 ( 10.1.1.11 )
Total Session(s)............. 4
Total Directories............ 2,331
Total File(s)................ 2,637
Total Skip(s)................ 0
Total Size (Disk)............ 89,095.26 MB
Total Size (Media)........... 89,109.56 MB
Elapsed Time................. 1h 10m 18s
Average Throughput........... 1,267.55 MB/min


Thursday Diff.
03/06/2003
Totals For................... DOMINO1 ( 10.1.1.11 )
Total Session(s)............. 4
Total Directories............ 2,335
Total File(s)................ 2,682
Total Skip(s)................ 0
Total Size (Disk)............ 89,922.12 MB
Total Size (Media)........... 89,936.50 MB
Elapsed Time................. 1h 10m 14s
Average Throughput........... 1,280.50 MB/min



Saturday Full
03/08/2003
Totals For................... DOMINO1 ( 10.1.1.11 )
Total Session(s)............. 4
Total Directories............ 2,344
Total File(s)................ 34,238
Total Skip(s)................ 0
Total Size (Disk)............ 96,466.70 MB
Total Size (Media)........... 96,560.81 MB
Elapsed Time................. 14h 16m 20s
Average Throughput........... 112.73 MB/min

 

[jonesjp]

I am fighting a similar problem on my Dell PE2550. My host server backs up at over 1000 MB per minute which was fixed by replacing the SCSI driver that Windows 2000 SP3 had overwritten with the latest from Dell. SCSI drivers ar a real issue
I still have all the remote servers backing up at around 50 MB per minute. I will let you know if I get a fix for that.
Jim

 

[NYCamc]

Thanks for your help! I had to rebuild the server to restore my bandwidth. I was very careful and documented any and all changes I applied to the server while I ran test backups checking my bandwidth. I have read on another post that SCSI drivers are probably the cause of the slowdown. I will be much more cautious about updating any SCSI or NIC drivers on this server.

 

[PinkFloydGal]

Hi All,
I have the same type issue. Brightstor works one day and then throughput crashes to 50MB/Min if the connection does not time out. I have run network analyzers, cut all services on the servers to be backed up ie. SQL, Oracle, NAV Etc, I have connected 2 test servers directly into the same switch bypassing the patch, put the cards to 1/2 duplex, xferred the data via a copy paste to make sure of the data integrity, pinged 1024 packets both directions and all sorts of other tests recommended by CA which of course was no help! I get EC 10060 and other TCP disconnection errors which I have troubleshot from CAs papers and nothing works. Please please please, any help greatly appreciated since I have not been able to run a successful back for over a month.

 

[PinkFloydGal]

P.S.
I have everything updated per CA and I have the latest SCSI drivers applied after any OS updates. Nothing seems to have changed from the point of working to not working. Any suggestions?

 

[rebelnorth]

I had a similar problem... just stopped working. Had to make sure that each machine being backed up had the nic set at half duplex. Just happened out of the blue. If your on a hub and not a switch like us. You need to set all your cards to half duplex.
Ian

 

[PinkFloydGal]

Thanks Rebelnorth! I have also tried this. It did not work however today our Seagate Viper 2000 autoloader ate one of my tapes and I have to get an RMA. I could not get a response from Seagate as to whether this may cause our specific problems so I am on hold till the new drive arrives thnaks for the response! Good Luck

 

[dominika]

In response to the initial problem which stated local backups are just as slow, this rules out any network problems. This may sound stupid, but try shutting down the services & renaming the log file. We had a similiar problem when the log file was about 16MB. We now do this on a regular basis and have not experienced any problems since.

 

[Selke]

I'm having similar issues. Dominika, are you referring to the BrightStor.log file? That file is around 22MB on our server here.

 

[Nightwatch]

Hi guys:

Refer to this thread, maby it will help

Server Rack > Back-up Systems > Computer Associates: ARCserve back-up Forum
Slow Backups
thread478-446991

 

[jsavely]

I had the same issue. Fought with it for weeks.
Replaced the adaptec scsi driver with an older version.
(ver. 4.2 Dated 2-21-2001)
My throughput went from 30mb/min up to 850mb/min on my faster clients and 350mb/min on my slowest client machine.

Adaptec 29160 scsi adapter.
The driver I used says it is an Adaptec AIC-7899- Ultra 160 scsi under device manager.

 

[impaler1]

ARCserve does not determine throughput as it does not move data; Data is being moved at the Mocrosoft API level. Grab about 1000MB of data from DOMINO1 and do a simple network file copy to the backup server and time how long it takes and calculate the throughput. Most likely these speeds are similar to that of the backup. Things to check are NIC card drivers, and duplex settings. If anything is set to auto set it to 100full. Also check AV settings as real time scanning can severely hinder backup speeds.