04-30-2008 07:35 PM - edited 03-03-2019 09:46 PM
Hi,
My OSPF neighbors reset them selves, I want to apply qos for HELLO and SLA traffic. I want to mark them but not sure how much bandwidth I need to allocate for it.
Can someone please advise or an approperaite estimate/config?
Regards
05-01-2008 12:45 AM
Hi,
you can create a policy marking local traffic - originated by the router - with a specific IP preference value. Note, by default OSPF packets will be marked with IP preference 6. On the interfaces you can then include a class for this traffic in your policy-map. It is hard to tell which amount of bandwidth you should allocate, as this depends on parameters like interface bandwidth and also on how much local traffic the router generates. An example:
ip local policy route-map IPP6
route-map IPP6 permit 10
match ip address LocalPrioTraffic
set ip precedence 6
ip access-list extended LocalPrioTraffic
permit ospf any any
permit tcp any any eq telnet
permit tcp any eq telnet any
permit udp any any eq snmp
!
! Match your local traffic here, like telnet, SNMP, IPSLA, etc.
!
class-map match-any NMS
match ip precedence 6
policy-map MyQoS
class NMS
bandwidth percent 10
class class-default
fair-queue
random-detect
interface Serial0
ip address 10.1.1.1 255.255.255.252
bandwidth 128
service-policy output MyQoS
For a detailed syntax description please have a look at
http://www.cisco.com/en/US/docs/ios/12_2/iproute/command/reference/1rfindp1.html#wpxref37880
Be aware the service-policy will also "catch" all transit telnet traffic, respectively all traffic marked with IP precedence 6 and you need to allocate enough resources for the combined traffic (local+transit).
Hope this helps! Please use the rating system.
Regards, Martin
05-01-2008 04:28 PM
thanks for the suggestion, do I need to set IP LOCAL-POLICY given that the ospf traffic are marked with IP precedence 6. what about the policy below? isn't it enough to treat the ospf traffic as desired.
clss-map OSPF-TRAFFIC
match ip precedence 6
policy-map MY-POLICY
class OSPF-TRAFFIC
bandwidth percent 10
Given that OSPF traffics are marked with IP precedence 6 by default, I can just match them and give it 10% of the bandwidth.
Regards
05-01-2008 01:10 AM
Hello,
a reference document could be RFC 1245 "OSPF Protocol Analysis" where OSPF bandwidth usage is studied.
You have to take in account that even in a stable network LSAs are refreshed evry 30 minutes.
The bandwidth usage depends from the number of IP routes in the OSPF database.
OSPF packets have usually IP Prec = 6 so you can create a class-map for them and defines some bandwidth for it in a CBWFQ scheduler.
You don't need to mark OSPF packets.
Received OSPF packets are placed in a queue for process switching.
Because the bandwidth command in CBWFQ does not imply a fixed reservation don't be afraid of giving too much bandwidth to OSPF traffic.
Some platforms like a C3640 and other small ones have a system queue for routing traffic but other platforms like c7500 and 12000 haven't got one and you need to define one.
hope to help
Giuseppe
05-01-2008 04:32 PM
Thanks for the suggestion, this is what I am leaning towards as well and I have asked "mheusing" to confirm.
05-01-2008 01:21 PM
By default only 75% of the bandwidth configured on an ineterface is used by traffic. Rest 25% is reserved for use by protocol related traffic which include the Hello's & LSA's u were talking about. So if u are using default interface configurations, there seems no need to use any qos features.
05-01-2008 04:35 PM
Can you please post us the link for reference as I am not quite sure what the default interface configuration is. Because by default an interface doesn't have a bandwidth statement but it it's configured I assume it changes the default behaviour and there are other similar configs...
05-01-2008 11:13 PM
Hello,
this is not true for all the router platforms.
Best Regards
Giuseppe
Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community: