nabeel92
asked on
cbr value on atm interface cisco 877 ??
Hi there,
I've a cisco 877 series router and i need to set either ubr, cbr or vbr value (which am not sure of what to set it to ? ) on the atm interface on cisco 877 to succesfully implement QOS policy... its an adsl 2 link with 24/1 speed (24 mbps down and 1 mbps up) ... how can i determine what should i set it to and what value ? somebody told me since its not a dedicated atm interface but just a normal cisco 877 with atm interface, it should be UBR .... but i need a specific logical answer ?
Similarly, I'd like to know what would be the tx-ring-limit value or rx-ring-limit value that can be applied on atm interface on cisco 877 and what's an optimal value ? thanks ....
I've a cisco 877 series router and i need to set either ubr, cbr or vbr value (which am not sure of what to set it to ? ) on the atm interface on cisco 877 to succesfully implement QOS policy... its an adsl 2 link with 24/1 speed (24 mbps down and 1 mbps up) ... how can i determine what should i set it to and what value ? somebody told me since its not a dedicated atm interface but just a normal cisco 877 with atm interface, it should be UBR .... but i need a specific logical answer ?
Similarly, I'd like to know what would be the tx-ring-limit value or rx-ring-limit value that can be applied on atm interface on cisco 877 and what's an optimal value ? thanks ....
ASKER
ok..ill try vbr-rt 1176 1176 at off peak hours today ... is your ADSL 2 router a cisco 877 series as well ? .. i had posted a QOS query earlier in this forum but was unable to get a response so i hope u can help me ... this is the first time ive applied QOS on 877 router with ADSL 2 connection ... am doing marking and classification both on the router, is there any huge disadvantage for that ? secondly, its an internet cafe for which primary goal is to make user web browsing experience faster and also prioritize voip traffic .... other requirements include limiting p2p traffic in hours of congestion and guarantee certain bandwidth for management traffic .... there are only 5 SIP phones using G.729 so i thought bandwidth percent 10 (which would give 76 kbps of bandwidth) should be enough since each call would take 55 kbps ....
given below is wht i've tried ... hope u can advise me if it's ok to implement and how can i improve this design ....
class-map match-any voip
match protocol rtp audio
match protocol sip
match protocol h323
match protocol mgcp
class-map match-any mgmt
match access-group name mgmt (ACL for telnet and VNC)
class-map match-any http
match protocol http
match protocol secure-http
match access-group name ack
class-map match-any p2p
match protocol gnutella file-transfer "*"
match protocol kazaa2 file-transfer "*"
match protocol fasttrack file-transfer "*"
match protocol bittorrent
match protocol edonkey file-transfer "*"
class-map match-any voip_signal
match access-group name voip_signal (ACL for voip signaling)
class-map match-any video
match protocol http url "*youtube.com"
match protocol rtsp
match protocol cuseeme
!
policy-map GS1
class video
bandwidth percent 10
set dscp af31
class mgmt
bandwidth percent 5
set dscp cs2
class http
bandwidth percent 30
set dscp af41
class voip
priority percent 10
set dscp ef
class voip_signal
bandwidth percent 5
set dscp cs3
class p2p
bandwidth percent 10
set dscp default
class class-default
fair-queue
!
interface ATM0
no ip address
no ip redirects
no ip unreachables
no ip proxy-arp
ip route-cache flow
no atm ilmi-keepalive
pvc 8/35
vbr-rt 1176 1176
tx-ring-limit 3
dialer pool-member 1
service-policy output GS1
protocol ppp dialer
!
dsl operating-mode auto
end
And also, i apply it on the dialer interface (which has the public I.P)
interface Dialer1
description -- ADSL Link --
bandwidth 1000
ip address x.x.x.x x.x.x.x
ip mtu 1440
ip flow ingress
ip flow egress
ip nat outside
no ip virtual-reassembly
encapsulation ppp
ip route-cache flow
dialer pool 1
dialer idle-timeout 0
dialer persistent
dialer-group 1
ppp authentication chap callin
ppp chap hostname xxxx
ppp chap password 0 xxxx
ppp multilink
service-policy output GS1
end
I have also noticed that after doing the above configuration, marking of traffic takes place at the dialer interface and queuing of traffic takes place at the atm interface .... is that the way it works ?
given below is wht i've tried ... hope u can advise me if it's ok to implement and how can i improve this design ....
class-map match-any voip
match protocol rtp audio
match protocol sip
match protocol h323
match protocol mgcp
class-map match-any mgmt
match access-group name mgmt (ACL for telnet and VNC)
class-map match-any http
match protocol http
match protocol secure-http
match access-group name ack
class-map match-any p2p
match protocol gnutella file-transfer "*"
match protocol kazaa2 file-transfer "*"
match protocol fasttrack file-transfer "*"
match protocol bittorrent
match protocol edonkey file-transfer "*"
class-map match-any voip_signal
match access-group name voip_signal (ACL for voip signaling)
class-map match-any video
match protocol http url "*youtube.com"
match protocol rtsp
match protocol cuseeme
!
policy-map GS1
class video
bandwidth percent 10
set dscp af31
class mgmt
bandwidth percent 5
set dscp cs2
class http
bandwidth percent 30
set dscp af41
class voip
priority percent 10
set dscp ef
class voip_signal
bandwidth percent 5
set dscp cs3
class p2p
bandwidth percent 10
set dscp default
class class-default
fair-queue
!
interface ATM0
no ip address
no ip redirects
no ip unreachables
no ip proxy-arp
ip route-cache flow
no atm ilmi-keepalive
pvc 8/35
vbr-rt 1176 1176
tx-ring-limit 3
dialer pool-member 1
service-policy output GS1
protocol ppp dialer
!
dsl operating-mode auto
end
And also, i apply it on the dialer interface (which has the public I.P)
interface Dialer1
description -- ADSL Link --
bandwidth 1000
ip address x.x.x.x x.x.x.x
ip mtu 1440
ip flow ingress
ip flow egress
ip nat outside
no ip virtual-reassembly
encapsulation ppp
ip route-cache flow
dialer pool 1
dialer idle-timeout 0
dialer persistent
dialer-group 1
ppp authentication chap callin
ppp chap hostname xxxx
ppp chap password 0 xxxx
ppp multilink
service-policy output GS1
end
I have also noticed that after doing the above configuration, marking of traffic takes place at the dialer interface and queuing of traffic takes place at the atm interface .... is that the way it works ?
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
Ok..I will increase the voip bandwidth to 15 % ...
You mentioned
2. " As you have a dialer interface you shouldn't need to bother with the vbr-rt command. My situation is different as my IP is directly on an ATM sub interface" ... So Should I leave it the way as it is ?
3. You don't need to use an access-list for voice signalling, but you do want to put sip, h.323, and mgcp in the voip signalling class as opposed to the voip class...I thought when we say match protocol mgcp or match protocol sip, it actually matches the sip traffic stream and not the signaling (whereas for signaling u need specific ports .. what I did in my acl voip_signal was
21 permit tcp any any eq 1720
22 permit tcp any any range 11000 11999
23 permit udp any any eq 2427
24 permit tcp any any range 2000 2002
25 permit udp any any eq 5060
4. "I would suggest not giving p2p a bandwidth percentage, and to be honest you probably don't need this so long as any important traffic does have a bandwidth guarantee. You can then allocate more bandwidth to more important classes" ... So is there any way i can improve http or throttle P2P traffic in hours of congestion only (possibly by using incoming .... Would creating another policy in inbound direction would do the job for me ? )
Your comments are really appreciated ....
You mentioned
2. " As you have a dialer interface you shouldn't need to bother with the vbr-rt command. My situation is different as my IP is directly on an ATM sub interface" ... So Should I leave it the way as it is ?
3. You don't need to use an access-list for voice signalling, but you do want to put sip, h.323, and mgcp in the voip signalling class as opposed to the voip class...I thought when we say match protocol mgcp or match protocol sip, it actually matches the sip traffic stream and not the signaling (whereas for signaling u need specific ports .. what I did in my acl voip_signal was
21 permit tcp any any eq 1720
22 permit tcp any any range 11000 11999
23 permit udp any any eq 2427
24 permit tcp any any range 2000 2002
25 permit udp any any eq 5060
4. "I would suggest not giving p2p a bandwidth percentage, and to be honest you probably don't need this so long as any important traffic does have a bandwidth guarantee. You can then allocate more bandwidth to more important classes" ... So is there any way i can improve http or throttle P2P traffic in hours of congestion only (possibly by using incoming .... Would creating another policy in inbound direction would do the job for me ? )
Your comments are really appreciated ....
ASKER
Hellow ??
ASKER
would really appreciate if some expert can answer this query please ? thanks ...
Hi,
sorry for the delay in getting back to you.
2. Yes leave it as is.
3. No, if you match h.323 or sip, you are matching the voice signalling only, not the media stream (the actual audio packets). The media is matched using match protocol rtp no matter what signalling you are using. It is not necessary to create access lists to match specific ports.
4. You can throttle p2p traffic with policing, but that would permanently stop p2p traffic from using a specific amount of bandwidth. If your http traffic has a bandwith guarantee, this will prioritse it a bit over the p2p traffic.
sorry for the delay in getting back to you.
2. Yes leave it as is.
3. No, if you match h.323 or sip, you are matching the voice signalling only, not the media stream (the actual audio packets). The media is matched using match protocol rtp no matter what signalling you are using. It is not necessary to create access lists to match specific ports.
4. You can throttle p2p traffic with policing, but that would permanently stop p2p traffic from using a specific amount of bandwidth. If your http traffic has a bandwith guarantee, this will prioritse it a bit over the p2p traffic.
As you only have one VC, you want to allocate all the bandwith to that VC. I would recommend using VBR-rt if you are planning on using voice or video, which I am guessing is why you want to configure QoS.
With regards to the tx-ring-limit value, I would recommend leaving this at its default setting. You won't really get any benefit from changing this.
Below is an example config from my ADSL 2+ router:-
class-map match-all dscp_ef
match dscp ef
class-map match-all dscp_af41
match dscp af41
policy-map outside_qos
class dscp_ef
priority percent 30
class dscp_af41
bandwidth remaining percent 75
interface ATM0.1 point-to-point
ip address x.x.x.x 255.255.252.0
ip access-group acl_outside in
no ip redirects
no ip unreachables
no ip proxy-arp
ip mtu 1452
ip flow ingress
ip flow egress
ip nat outside
ip inspect INS_LOW out
ip ips sdm_ips_rule in
ip virtual-reassembly
atm route-bridged ip
pvc 0/101
vbr-rt 1176 1176
encapsulation aal5snap
service-policy output outside_qos
!
end
This will guarantee 30% of the bandwith for voice traffic, and of the remaining bandwidth (which will be 75% * 70%* link bandwith) prioritise AF41 traffic to use up to 50% of that bandwidth.