Bandwidth reservation

[ohassairi]

Hi

we have a HQ and many remote sites connected by MPLS connections

one router is inserted in each site to get netflow statistics and QoS...this router is managed by us

In HQ we have many applications used by remote sites such as : web browsing, Antivirus updates, oracle DB, Mail, Domain controller...

some times AV or mail appliations consume all the WAN BW in remote sites: this will make oracle applications very slow.

so i need to garantee at least the 1/4 of the BW in remote sites for oracle application. if no oracle trafic is there, other applications can use it. but if there is  an oracle traffic so the QoS must be applied.

i found one example with this : class-map, policy-map, service-policy.....

however i am confused:

1-shall i use this in HQ side or remote side or both ?

2-for the service-policy command shall use the input or output option in HQ and in remotes

many thanks for your help

[Marwan ALshawi · Accepted solution]

Hi there

To answer your question there couple of stages you need to setup

First create some class maps with ACLs to match and identify the critical traffic such as the DB in each site

Then you need to shape each interface in the outbound direction in each site and this policy will be called parent policy

For example if the router interface is 100m while the actual bandwidth from the ISP is 10m then you need tonshape the interface to 10M the. You need to Apply a nested/chiled policy under the parent one to allocate the required bandwidth in the case of interface congestion as qos policy works in case ofninterface congestion only

Then this policy has to be applied in the outbound direction in each site with the relevant config

If you want to have end to end qos then you need to speak to your ISP to allocate some classes to be marked and allocated some bandwidth within the ISP cloud network if you think you need this level of end to end qos

Please refer to the bellow link to understand the terms and config methods mentioned above

https://supportforums.cisco.com/docs/DOC-8373

Hope this help

If helpful rate

[ohassairi]

thanks marwan

if i am specifing exactly the allocated bw (in kbps and not in %) can i bypass the shaping step ?

[Marwan ALshawi]

Yes

But even if you gonna use the absolute bandwidth not % I would recommend you to use the parent policy with shaping to add some software buffering capability to the output queuing

Hope this help

[ohassairi]

ok Marwan

i have 2 small notes

1-in your MQC document you configured class-default and you assigned to it fair-queue. is this necessary? what will be the difference if i don't configure a default class?

2-i think i must garantee in HQ a bandwith = sum of bw that are reserved in remotes. isn't it ?

thanks

[Joseph W. Doherty · Accepted solution]

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

How many is "many" in "many remote sites"?  Is it possible there would be remote-to-remote traffic?  Reason I ask these two questions, since you mention you're using MPLS, part of your QoS strategy might include MPLS QoS, if supported by your MPLS service provider.

For example, assuming the hub's connection to the MPLS cloud has more available bandwidth that a spoke's connection to MPLS cloud, one common point of congestion is cloud egress to the spoke.  One method to manage this congestion point is to artificially move it to the hub using a shaper for all the traffic from the hub to the spoke (as described by Marwahshawi).

The other approach would be to not shape, but have a MPLS QoS policy enabled at cloud egress for a spoke for which the hub only needs to correctly tag packets to utilize.  (NB: with MPLS QoS, although shaping shouldn't be needed, you'll still want a policy to manage possible congestion on the interface to the MPLS cloud.)

BTW, converse applies from spoke to hub (or other spoke).

[ohassairi]

thanks joseph, but in our case our ISP does not support MPLS QoS

for the many, we have just 4 remote sites and no traffic between spokes

[Marwan ALshawi]

In this case what you do create a class map per application per site using ACL on each class map that match source as the HQ network address and destination as the remote site (one per remote site )

Then give it the desired amount of bandwidth to allocated per site

You cN do it different ways but this is one way you can achieve it assuming your topology like hub and spoke and traffic mostly spoke to hub or hub to spoke

Hope this help

[ohassairi]

do i need to configure class-default? is this necessary? what will be the difference if i don't configure a default class?

[Marwan ALshawi]

No the class default is by default there if you want to enable fair queuing under it you need then to manually add it which is not must but i recommend you to do it if you can

Hope this help