cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
708
Views
0
Helpful
4
Replies

LNS PPP Multilink

Davidbb80
Level 1
Level 1

Hi all,

we've been terminating l2tp users on our LNS using a standart virtual-template and there are around 5K subscribers on it.

So, we'd like to implement a ppp multilink for bonding subscribers to increase their throughput. Currently there is no command like "ppp multilink" enabled under the virtual-template interface. As far as i know there is no way to enable "pp multilink" per user unless pre-authentication is used on the lac side, which is not possible in our case since lac belongs to the incumbent operator, so here are my questions:

1.) is there a way to enable ppp multilink per user on the lns side ?

2.) what happens if i globally enable the "ppp multilink" command under the global virtual template interfece considering 5K users are terminated by it.

3.) Will ppp multilink use per packet or per destination load sharing ?

Any answer will be appreciated.

Regards

1 Accepted Solution

Accepted Solutions

I was too fast in my reading, sorry ;-)

To implement PPP mulitlink or MLP, both the LAC and the LNS must be configured. Then when a users opens a second link and request for MLP support, it will be negotiated with the LAC and the LNS.

If the LAC doesn't support MLP, configuring it on the LNS will not change anything as I think you will never receive LCP request for MLP (LAC will not accept the user opens a 2nd link).

Regarding the algo itself, your understanding is correct. From RFC 1990:

"

The simplest possible algorithms of alternating packets between

channels on a space available basis (which might be called the Bank

Teller's algorithm) may have undesirable side effects due to

reordering of packets.

By means of a four-byte sequencing header, and simple synchronization

rules, one can split packets among parallel virtual circuits between

systems in such a way that packets do not become reordered, or at

least the likelihood of this is greatly reduced.

"

HTH

Laurent.

View solution in original post

4 Replies 4

Laurent Aubert
Cisco Employee
Cisco Employee

Hi,

The following feature should help you:

http://www.cisco.com/en/US/docs/ios/12_2t/12_2t11/feature/guide/ftppprad.html

You should make some performance tests to see how your LNS scales with this feature

With multilink, each packet is fragmented and one fragment is sent over each link.

HTH

Laurent.

Hi Laurent,

Thanks for the answer but as i mentioned previously in my mail, this feature doesn't match our case since pre-authentication is done on the lac side and lac is under the control of the incumbent operator.

As multilink is a part of the lcp negotiation, i don't think that such a feature could exist to be configured on the lns side.

However i really wonder what will happen if enable ppp multilink globally under the virtual template interface which is in charge of terminating thousand of subscribers that most of them don't need multilink.

On the other hand could you more detail please "each packet is fragmented " statement ? Do you mean that each packet is splitted into two equal packets and then sent over each link, looks like a different approach of per packet switching.

Regards

I was too fast in my reading, sorry ;-)

To implement PPP mulitlink or MLP, both the LAC and the LNS must be configured. Then when a users opens a second link and request for MLP support, it will be negotiated with the LAC and the LNS.

If the LAC doesn't support MLP, configuring it on the LNS will not change anything as I think you will never receive LCP request for MLP (LAC will not accept the user opens a 2nd link).

Regarding the algo itself, your understanding is correct. From RFC 1990:

"

The simplest possible algorithms of alternating packets between

channels on a space available basis (which might be called the Bank

Teller's algorithm) may have undesirable side effects due to

reordering of packets.

By means of a four-byte sequencing header, and simple synchronization

rules, one can split packets among parallel virtual circuits between

systems in such a way that packets do not become reordered, or at

least the likelihood of this is greatly reduced.

"

HTH

Laurent.

Hi Laurent,

Thank you very much for your answer, it was really satisfying. Appreciated and rated ;-)

Regards.