Hello All,
I want to implement Qos in our company and i want your help in reviewing, correcting and suggesting a new way to implement or configure.
my company network is hub and spoke network the branches routers connect to the hub via two link one primary and the second is a backup link,
the primary link is mpls link and the secound is 3G link.
we have two type of traffic, traffic that originated from the HUB and traffic originated
from the spoke.
--Traffic originated from the GPX
1- AS400 TELNET traffic (the most valuable traffic) Value: 5 20% of our link Port 23
2- AS400 Printing traffic Value: 4 40% of the remaining bandwidth Port 9001 5001 9600
3- AS400 Query Traffic Value:2 40% of the remaining bandwidth Port 8471
4- Mail Traffic Value:2 40% of the remaining bandwidth Port 1352
5- Internet and other traffics. Value:1 20% of the remaining bandwidth
--Traffic originated from the Branches
1- AS400 tenet Traffic Value:5 30% of our link Port 23
2- Mail Traffic Value:2 60% of the remaining bandwidth Port 1352
3- Internet and other traffic Value:1 40% of the remaining bandwidth
the Value mean the IP precedence value that I will assign to each one of these traffics.
We will use NBAR "Network based application recognition" well-know PDLM to classify the AS400 TELNET traffics and LOUTS NOTES traffics, and we will define new Custom PDLM to classify the AS400 Printing and Query traffics.
We will use a hierarchy POLICY-MAP (Parent child policy) in our branches routers because all the branches router connect via sub interfaces or BVI interfaces and those is not supported with normal POLICY-MAP.
I will provide two configuration example, one for the branches routers (BNS) and one for the Hub (GPX) router. the configuration will be divided into five steps.
a. Classification.
b. Marking.
c. Congestion Management.
d. Applying service policy.
e. Verifying the configuration.
Configuration Example For The GPX Routers:
Classification:
1- we will define two new NBAR PDLM one for the AS400 Query traffic and the other for the AS400 Printing Traffics.
GPX(config)#ip nbar custom AS400_QRY TCP 8471
GPX(config)#ip nbar custom AS400_PRT TCP 9001 5001 9600
2- We will enable NBAR in every interface on the routers.
GPX(config)#interface fastEthernet 0/1
GPX(config-if)#ip nbar protocol-discovery
3- we will classify the packets using CLASS-MAP and NBAR.
GPX(config)#class-map AS400-QRY
GPX(config-cmap)#match protocol AS400_QRY
GPX(config)#class-map AS400-PRT
GPX(config-cmap)#match protocol AS400_PRT
GPX(config)#class-map AS400-TELNET
GPX(config-cmap)#match protocol TELNET
GPX(config)#class-map LOUTS-NOTES
GPX(config-cmap)#match protocol NOTES
Marking
1- we will mark the packets with IP precedence value by using policy map this policy map will be applied on the inbound direction under the local interface.
GPX(config)#policy-map MARK-TRAFFIC
GPX(config-pmap)#class AS400-TELNET
GPX(config-pmap-c)#set ip precedence 5
GPX(config-pmap)#class AS400-PRT
GPX(config-pmap-c)#set ip precedence 4
GPX(config-pmap)#class AS400-QRY
GPX(config-pmap-c)#set ip precedence 2
GPX(config-pmap)#class LOUTS-NOTES
GPX(config-pmap-c)#set ip precedence 2
GPX(config-pmap)#class class-default
GPX(config-pmap-c)#set ip precedence 1
2- we will make a CLASS-MAP to match the the different ip precedence value we assigned to our traffics.( classifying what we previously marked )
GPX(config)#class-map IP-PRECEDENCE-1
GPX(config-cmap)#match ip precedence 1
GPX(config)#class-map IP-PRECEDENCE-2
GPX(config-cmap)#match ip precedence 2
GPX(config)#class-map IP-PRECEDENCE-4
GPX(config-cmap)#match ip precedence 4
GPX(config)#class-map IP-PRECEDENCE-5
GPX(config-cmap)#match ip precedence 5
Congestion Management
1- we will define policy map and will use the marking we did before to assigns different treatment for each one of our traffics.
GPX(config)#policy-map LLQ-IBS
GPX(config-pmap)#class IP-PRECEDENCE-5
GPX(config-pmap-c)#priority percent 20
GPX(config-pmap-c)#class IP-PRECEDENCE-4
GPX(config-pmap-c)#bandwidth remaining percent 40
GPX(config-pmap-c)#class IP-PRECEDENCE-2
GPX(config-pmap-c)#bandwidth remaining percent 40
GPX(config-pmap-c)#class IP-PRECEDENCE-1
GPX(config-pmap-c)#bandwidth remaining percent 20
Applying service policy:
1- we will lock every interface on the router with the correct bandwidth that we have because if we left it with the default value we will get wrong calculation of the bandwidth percentage that we assigned to our traffics.
GPX(config-if)#interface fastEthernet 0/1
GPX(config-if)#bandwidth 4000
we will repeat this configuration with all the interface.
2-we will apply the policy map that we use for marking the traffic in the local interface that face our servers.
GPX(config-if)#interface fastEthernet 0/1 (for example)
GPX(config-if)#service-policy input MARK-TRAFFIC
3-we will apply the policy map that we use for congestion management in the outbound wan interface that face our branches.
GPX(config)#interface fastEthernet 0/0
GPX(config-if)#service-policy output LLQ-IBS
Verifying the configuration:
we can verify the configuration by using the # show policy-map interface X/X
Configuration Example For The Branches Routers:
Classification:
1-We will enable NBAR in every interface on the routers.
BNS(config)#interface fastEthernet 0/1
BNS(config-if)#ip nbar protocol-discovery
2-we will classify the packets using CLASS-MAP and NBAR.
BNS(config)#class-map AS400-TELNET
BNS(config-cmap)#match protocol TELNET
BNS(config)#class-map LOUTS-NOTES
BNS(config-cmap)#match protocol NOTES
3- we will mark the packets with IP precedence value by using policy map this policy map will be applied on the inbound direction under the local interface.
BNS(config)#policy-map MARK-TRAFFIC
BNS(config-pmap)#class AS400-TELNET
BNS(config-pmap-c)#set ip precedence 5
BNS(config-pmap)#class LOUTS-NOTES
BNS(config-pmap-c)#set ip precedence 2
BNS(config-pmap)#class class-default
BNS(config-pmap-c)#set ip precedence 1
4- we will make a CLASS-MAP to match the the different ip precedence value we assigned to our traffics.( classifying what we perviously marked )
GPX(config)#class-map IP-PRECEDENCE-1
GPX(config-cmap)#match ip precedence 1
GPX(config)#class-map IP-PRECEDENCE-2
GPX(config-cmap)#match ip precedence 2
GPX(config)#class-map IP-PRECEDENCE-5
GPX(config-cmap)#match ip precedence 5
Congestion Management
like we said before congestion management in the branches will be done by a hierarchy policy map because direct implementation of LLQ not supported in the sub interfaces.
1- we will first create the normal congestion management policy map like the one we made on the GPX
BNS(config)#policy-map CHILD-LLQ-IBS
BNS(config-pmap)#class IP-PRECEDENCE-5
BNS(config-pmap-c)#priority percent 30
BNS(config-pmap-c)#class IP-PRECEDENCE-2
BNS(config-pmap-c)#bandwidth remaining percent 60
BNS(config-pmap-c)#class IP-PRECEDENCE-1
BNS(config-pmap-c)#bandwidth remaining percent 40
2- then will create a parent policy map to match all traffic and shape it by 100% ( you can say that we didn't do any thing ) and we will apply the child policy-map under it.
BNS(config)#policy-map PARENT-LLQ-IBS
BNS(config-pmap)#class class-default
BNS(config-pmap-c)#shape average percent 100
BNS(config-pmap-c)#service-policy CHILD-LLQ-IBS
Applying service policy:
1- we will lock the wan interfaces with the correct bandwidth that we have because if we left it with the default value we will get wrong calculation of the bandwidth percentage that we assigned to our traffics.
BNS(config-if)#interface Serial 0/0/0.1
BNS(config-if)#bandwidth 1000
2-we will apply the policy map that we use for marking the traffic in the local interface that face our local network.
BNS(config-if)#interface fastEthernet 0/0
BNS(config-if)#service-policy input MARK-TRAFFIC
3-we will apply the policy map that we use for congestion management in the outbound wan interface that face our branches. in this case we will apply the Parent policy not the child policy
BNS(config)#interface Serial 0/0/0.1
BNS(config-if)#service-policy output PARENT-LLQ-IBS
Verifying the configuration:
we can verify the configuration by using the # show policy-map interface X/X
• Note: the interfaces used here and the bandwidth value used in this example not reflecting the actual interfaces or bandwidth on our network routers, this is for the example only.
I hope I don't mistype anything.
Thanks
