RB wrote:
I, on the other hand, am still interested in how things work without
the pfSense box in-between.

I'm more interested in how a good network performance measurement looks. Backup software performance is extremely complex, there are numerous potential bottlenecks. Processor, RAM, disk I/O, tape or other media I/O, etc. Network throughput is a concern, but generally on Gb networks it will not be your first bottleneck. I'm not familiar with Retrospect specifically, and the operation of any backup software will vary greatly depending on your config.

It may be that it uses relatively short spikes of network throughput, and without a filter between maybe it can hit close to or over 900 Mb. With a VLAN router, the traffic is going in and out the same NIC, and on the same PCI bus. It's really pushing double the actual throughput you're getting. On a good day, with a PCI NIC or an onboard NIC on that class of machine, you *may* get 500 Mb in a VLAN routing scenario (you're going to max out the PCI bus at about that rate, maybe less).

This is why I'm much more interested in actual throughput than what your backup software is getting, because that's not very useful in determining if you really have a problem. Attempting to route and filter at gigabit wire speed on a desktop class machine is not going to get you the same speeds as when there's nothing between. If that's a problem, you either need to not split subnets, or use server class hardware.

Now there could be an actual performance problem, but the speed of backup software is useless in determining whether that's the case. Like I suggested, try iperf or netperf between VLAN's and see what you get. If you're only getting like 25 Mb or something, you have a major problem. If you're getting 300+ Mb, then you are probably going to have to live with what you're seeing, get faster hardware, or keep boxes that require very fast throughput between them on the same segment.


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to