02-28-2014 01:58 AM - edited 03-04-2019 10:28 PM
Hi all.
Another last minute project has come up. I have 4 sites that are all going to connect via a MPLS service where each site will have a single feed into the WAN router from the service provider network. Each site will have different bandwidths eg 10mbps, 20mbps 40 Mbps and the last at 50mbps. How do I ensure that the higher speed sites don't saturate the slower links? Do I have to use the EXP bit in the MPLS frames? How do I have to write the COS based on source and destination subsets?
Eventually there will be a requirement to encrypt the WAN link, possibly look at GETVPN but I need to setup this in a lab environment.
Thanks
Switched
Sent from Cisco Technical Support iPad App
02-28-2014 02:53 AM
Disclaimer
The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.
Liability Disclaimer
In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.
Posting
A "MPLS service"? You should contact them for documentation of their QoS support and what exactly the "hand-off" will be. Often MPLS services do not "hand-off" native MPLS to the customer.
The "problem" with MPLS services, they often support multipoint communication. This can lead to multiple sites sending to one site, so you really want provider egress QoS support.
If you can logically run you topology as hub and spoke, you can shape from hub to each spoke and you can shape spokes to hub to insure their aggregate bandwidth won't exceed hub's bandwidth.
02-28-2014 03:21 AM
Hi Joesph thanks for replying. The service is actually a VPLS service were we are free to run any protocol or l2 or L3 service across. We have an existing MPLS network in place that really is two point to point links In a hub and spoke we are migrating off for other reasons I won't go into. I want to maintain using MPLS as we use multiple VRFs at each site.
We will logically run the new service as hub and spoke, the hub being where our internet gateway is and the spokes are the remote sites.
The provider has 4 cos classes we can map our traffic into depending on the traffic type.
What Im trying to find is examples of how best to configure the shaping at each site to ensure that we are not congesting.
Sent from Cisco Technical Support iPad App
02-28-2014 04:24 AM
Hello
To avoid such egress blocking you can shape on each particular hub and spoke connection to the same values.
Hub
policy-map SP1_Parent
class class-default
shape average 5120000
policy-map SP2_Parent
class class-default
shape average 20480000
policy-map SP3_Parent
class class-default
shape average 10240000
interface FastEthernet0/0.11
Description Spoke 1
service-policy output SP1_Parent
interface FastEthernet0/0.12
Description Spoke 2
service-policy output SP2_Parent
interface FastEthernet0/0.13
Description Spoke 3
service-policy output SP3_Parent
Spoke1
policy-map HUB_Parent
class class-default
shape average 5120000
interface FastEthernet0/0
Description HUB
service-policy output HUB_Parent
If you need then to spilt that Shaping values between various QOS policys you could apply HQoS
class qos1
match xxx
class qos2
match xxx
class EF
match xxx
policy-map HUB_Child
class qos1
bandwidth remaining percent 30
class qos2
bandwidth remaining percent 40
class EF
priority percent 20
class class-default
fair-queue
random-detect dscp-based
policy-map HUB_Parent
class class-default
shape average 5120000
service-policy HUB_Child
res
Paul
Please don't forget to rate any posts that have been helpful.
Thanks.
02-28-2014 05:35 AM
Disclaimer
The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.
Liability Disclaimer
In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.
Posting
If your MPLS service will support egress QoS to your spokes, which you say they do, you shouldn't need to shape on the hub.
On the hub, you configure QoS to mark to work with the MPLS service and you queue as desired on your egress to the MPLS service. (Often your queuing policy corresponds to the MPLS vendor policy, but it can be "better".) Spokes are configured likewise.
02-28-2014 03:30 PM
The data sheet report a single connection per site with any to any connectivity. They mention it is possible for congestion to happen when a site tries to transmit more data than the access circuit can handle. The provider includes QOS as a standard feature to aid optimal performance during access circuit congestion. Their recommendation for us the customer to match the outgoing traffic flow to access circuit bandwidth at sites for best performance.
With that being said, is it still best recommendation to configure our sits as hub and spoke?
Sent from Cisco Technical Support iPad App
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: