cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
905
Views
12
Helpful
7
Replies

Bandwidth Management in Cos Policies

Kevin Melton
Level 2
Level 2

I consult for a company that has an

MPLS circuit between their Headquarters and their Disaster Recovery Site.  The MPLS circuit has

3 meg of bandwidth via two bundled T1 circuits.

We have a CoS Policy on the routers on each side which effectively governs what classes get what bandwidth.

We have two main applications that use this pipe.  The first application (SCADA) monitors our electrical generation capabilities and the other (netApp) is for replication traffic between HQ and the DR site.

The SCADA application pretty much runs all the time and collects data from collectors and delivers them to servers.  We have it configured to run in CoS Class 2 because for this company SCADA is our top priority.

Our replication app (NetApp) is configured to run in CoS Class 4.  We give it a huge amount of bandwidth in off hours when the replication is set to run.  It is scaled back during the day as replication is not set to occur then.

What i want to understand is this.  If we have a certain amount of bandwidth guaranteed in our CoS policy for Class 2 (we will say 400K for hypothesis sake), does the amount of available bandwidth for CoS Class 2 have the abiltiy to exceed 400K when there is not much traffic in the other classes?  Or is it the case that the max amount of bandwidth ever available to Class 2 is 400K?

Thanks

Kevin

7 Replies 7

Jon Marshall
Hall of Fame
Hall of Fame

k-melton wrote:

I consult for a company that has an

MPLS circuit between their Headquarters and their Disaster Recovery Site.  The MPLS circuit has

3 meg of bandwidth via two bundled T1 circuits.

We have a CoS Policy on the routers on each side which effectively governs what classes get what bandwidth.

We have two main applications that use this pipe.  The first application (SCADA) monitors our electrical generation capabilities and the other (netApp) is for replication traffic between HQ and the DR site.

The SCADA application pretty much runs all the time and collects data from collectors and delivers them to servers.  We have it configured to run in CoS Class 2 because for this company SCADA is our top priority.

Our replication app (NetApp) is configured to run in CoS Class 4.  We give it a huge amount of bandwidth in off hours when the replication is set to run.  It is scaled back during the day as replication is not set to occur then.

What i want to understand is this.  If we have a certain amount of bandwidth guaranteed in our CoS policy for Class 2 (we will say 400K for hypothesis sake), does the amount of available bandwidth for CoS Class 2 have the abiltiy to exceed 400K when there is not much traffic in the other classes?  Or is it the case that the max amount of bandwidth ever available to Class 2 is 400K?

Thanks

Kevin

Kevin

Is the MPLS network owned by the company or by a provider. If it is leased from a provider then it is likely that if you have paid for 400k then it is limited to that amount but you would need to check with them or check the contract.

Jon

We do lease it from AT&T.  But the two routers in question are not managed,i.

e., we control them and configure them and are responsible for the CoS policy configured on each side.

Thanks Jon

Kevin

k-melton wrote:

We do lease it from AT&T.  But the two routers in question are not managed,i.

e., we control them and configure them and are responsible for the CoS policy configured on each side.

Thanks Jon

Kevin

Kevin

It depends on your router config then but if you have used MQC with the bandwidth command then usually if there is spare bandwidth it can be used.

But the thing to note is that you can mark your packets to whatever you want but AT&T won't necessarily honour those markings unless you have agreed/paid AT&T to do so.

Jon

Jon

We are not using the MQC command in our configuration.  here is a copy of the policy that we use.  Do you think we are missing something?

policy-map QOS
class COS2_TRAFFIC
  set dscp af31
  shape average percent 24
  bandwidth percent 24
  random-detect dscp-based
class COS3_TRAFFIC
  set dscp af21
  shape average percent 20
  bandwidth percent 20
  random-detect dscp-based
class class-default
  fair-queue
  random-detect

Thanks

Kevin

Jon

We are not using the MQC command in our configuration.  here is a copy of the policy that we use.  Do you think we are missing something?

policy-map QOS
class COS2_TRAFFIC
  set dscp af31
  shape average percent 24
  bandwidth percent 24
  random-detect dscp-based
class COS3_TRAFFIC
  set dscp af21
  shape average percent 20
  bandwidth percent 20
  random-detect dscp-based
class class-default
  fair-queue
  random-detect

Thanks

Kevin

Kevin

1.Those commands are MCQ commands, whether made by admin, SDM or Auto QOS.. they are MQC

2. IF both routers are configured similar I dont see any "Policeing" going on that would put a hard cap of bandwidth for those apps. if more bandwidth is available it should be able to use it.

3. The routers may be yours but you are traversing AT&Ts network, and as stated before unless you have agreements to honor the DSCP values you are setting, AT&T is only providing you best effort accross thier network. (someone correct me if i am wrong, on an MPLS network for AT&T to honor QOS you would be setting MPLS EXP bits settings for AT&T to work with.)

-Chris

chris_younkers wrote:

Jon

We are not using the MQC command in our configuration.  here is a copy of the policy that we use.  Do you think we are missing something?

policy-map QOS
class COS2_TRAFFIC
  set dscp af31
  shape average percent 24
  bandwidth percent 24
  random-detect dscp-based
class COS3_TRAFFIC
  set dscp af21
  shape average percent 20
  bandwidth percent 20
  random-detect dscp-based
class class-default
  fair-queue
  random-detect

Thanks

Kevin

Kevin

1.Those commands are MCQ commands, whether made by admin, SDM or Auto QOS.. they are MQC

2. IF both routers are configured similar I dont see any "Policeing" going on that would put a hard cap of bandwidth for those apps. if more bandwidth is available it should be able to use it.

3. The routers may be yours but you are traversing AT&Ts network, and as stated before unless you have agreements to honor the DSCP values you are setting, AT&T is only providing you best effort accross thier network. (someone correct me if i am wrong, on an MPLS network for AT&T to honor QOS you would be setting MPLS EXP bits settings for AT&T to work with.)

-Chris

Chris

Don't know about AT&T but here in the UK certainly with BT you would have to purchase a certain amount of bandwdith and then they would map your CoS value to an MPLS EXP value when it traverses their network. Or at least that was how it worked when i last dealt with them.

Not sure about the hard cap statement though. If you shape the queues you are in fact limiting how much data can be sent ie. the different between policing and shaping is that they both limit the transmission of data but where policing drops packets shaping buffers them. So it does look there is a limit each class can use.

Jon

My next question based upon the fact that it does not look like the bandwidth adjusts between classes

is:  How can I get visibility into how much traffic is traversing each class?  Right now I have CoS 2 set to 12 percent bandwidth.  How can I tell if this class is full and maxed out, which would subsequently cause some of the timeouts in processes within the apps?  How would i be able to see what times of day the traffic is full and what times it is empty?

Thanks

Kevin

Getting Started

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: