Help with configuring QOS 1

[beatdown]

I work for a small company with a single T1 line to the internet. We connect over the web, through an IP-Sec VPN, to an application server at our remote office, using Microsoft Terminal Server (RDP Protocol). The problem is that the Terminal Server connections are often slow, due to other traffic on the T1. So we'd like to configure QOS to give priority to the RDP traffic.

We have a 2611 router with an IOS that supports QOS, but I have no idea how to configure this. I did a search on Google and the Cisco web page for QOS, and all the results I came up with were huge documents that seem to apply to configuring QOS for an ISP or an entire enterprise.

I hate to be lazy, but can anyone point me in the direction of a white paper that can tell me how to do a "simple" QOS config, such as what we need to do in order to give priority to RDP traffic. Or better yet, if anyone else has done this, can you post your config?

Thanks for your help!

 

[ADB100]

The problem here is the typical traffic flows and where you have and haven't control.

Typically your inbound traffic rate will be higher than your outbound and the only place you have control is your outbound. You can therefore police traffic out but not back in since this would have to be done by your ISP.

If all your traffic goes via the VPN then you have more of a chance since you will control the outbound at your remote office and the outbound at the central office. If you connect to the internet directly at the remote office (and not via the VPN) then there isn't much you can do without involving your ISP.

HTH

Andy

 

[beatdown]

Both offices connect to the internet through their own T1. The vast majority of inbound traffic at both offices comes over the VPN from the office at the other side of the VPN. So does this mean we'd be able to control traffic fairly well by doing some QOS configs on the routers at both offices?

Thanks,
Steve

 

[KiscoKid]

Here's a config I did not too long ago for RDP QoS. It's applied over an ADSL link but all you should need to do is apply the service policy (called PM2 under the ATM VC) to your serial interface instead:

class-map match-all RDPCM
match access-group 100
!
!
policy-map PM2
class RDPCM
priority 50
class class-default
fair-queue
random-detect
!
interface ATM0
no ip address
ip route-cache policy
no atm ilmi-keepalive
pvc 0/38
vbr-nrt 113 113 190
encapsulation aal5mux ppp dialer
dialer pool-member 1
service-policy output PM2
!
dsl operating-mode auto
!
!
access-list 100 remark Windows Terminal Server Traffic
access-list 100 permit tcp any any eq 3389
access-list 100 permit tcp any eq 3389 any


Any questions, drop em in here.

 

[beatdown]

Thanks, this exactly what I need.

Just one question - is the command "priority 50" what determines how much bandwidth is allocated to RDP? Does this mean 50% of available bandwidth?

Thanks again,
Steve

 

[KiscoKid]

In this example, priority 50 allocates 50k of bandwidth for RDP. The interface bandwidth is taken from the size of the PVC (in this example) which is 113k.

If you want to set 50% of available bandwidth, you would instead use the following:

policy-map PM2
class RDPCM
priority percent 50

Incidentally you may not want to assign RDP to the priority queue (it is typically reserved for voice and delay-sensitive traffic). If this is the case, you can instead do the following:

policy-map PM2
class RDPCM
bandwidth percent 50

The differences between bandwidth and priority is simply that priority traffic is always serviced before anything else. So be wary of using priority queuing in case you starve other traffic. There are other differences too inc that bandwidth reserves a minimum bandwidth during congestion whilst a priority reservation reserves a minimum AND maximum during congestion (i.e. with priority percent 50, the RDP sessions will get at least 50% and no more during congestion)