Backup Exec 2010 - Specific task decreasing network performance

[PANASAM]

Hi guys,

We have an issue with our backup exec 2010 R3. There is a task that runs every day at 6:00 p.m., but it seems that this task saturates the core switch which has static routes to remote offices and outputo to Internet. Suddenly, all links to offices and also Internet go down, the core switch sends me 15-20ms ping responses and when the task is finished, everything is fixed.

Please any advice, I already tried using wireshark to see what´s going on but I don´t see anything weird (broadcast storm or other issues), even the backup exec´s server doesn´t seem to overflood the switch with frames. 

I would like to know if there is a way to control the bandwidth used for this backup task, I´m newbie using backup exec.

thanks...

[teiva-boy]

WIthin BackupExec, there is no way to throttle bandwidth.  You would only be able to do it at the host or switch level.

[alexander_H]

You should try to think about backup strategy.

Move your jobs out of working hours, consider client-side dedublication, also you may want to install local media consoles in order to decrease the traffic.

Provide us with more information about your setup so we could assist you.

[PANASAM]

Hi guys,

Ok attached you could see a sample of what we have.

You could see that both servers (backup exec´s server and remote agent server) are in the same switch, and end users are located in another switch.

This task is running out of working hours, but I would like to find a solution for this...maybe there is something I haven´t see it....

thanks for your time....

[Ken_Putnam]

With that setup, backup data should never hit the Fiber Link.  it should go

RAWS <-> switch <-> media server

double check the setup on that switch

[PANASAM]

I should mention that Remote Agent Server has 2 IP address configured in the same network card, both address are in the same network ( this is a needed for another application). But BE server is configured to seek for just one address...

Do you think this could be the root of the problem ?

[teiva-boy]

Are you trying to backup over slower WAN links?

[Ken_Putnam]

According to his diagram, both the media server and the remote server connect to the same switch

[PANASAM]

Hi,

It is a local backup, both servers are on the same VLAN, same subnet, same site.

[teiva-boy]

Sounds more like a network issue that more traffic capturing needs to be done to see what is escapaing out to other vlans and such.

[Lesta_G]

Perhaps setting QoS on the switch and making BE traffic the lowest priority is a workaround...

BUT as they are on the same subnet..............the core switch should no be affected. As Ken has suggested, the issue is likely to be with the first switch

[Sergio_Maroni]

I`m think you need configure first switch.

[PANASAM]

Ok.

I´m still trying to figure out what could be the problem. I already checked configuration in each switch, all of them has the same VLAN, their IP default gateways and ip addresses.

But, there is something new I saw in wireshark. All packets that I captured show a flow conversation between the BE and remote server using protocol MSNMS (port 1863).

Does BE use this protocol ? I sniffed packets for the other backup tasks and didn´t see this protocol.

[PANASAM]

Guys,

Can checksum errors decrease the performance of a network ?

I saw a lot of packets telling me that there is a checksum error 0x0000 and the value should be another one. This errors come from the remote agent server...

[Lesta_G]

Careful using wirshark

Most network cards these days  to the checksum offload on the card. wireshark cannot see this going on and produces an error (lots!)

To confirm this is a false positive, turn off tcp offload on the card(s) reboot PC/serve and test again

http://www.wireshark.org/docs/wsug_html_chunked/ChAdvChecksums.html