cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
7661
Views
5
Helpful
9
Replies

What is the importance of "Fragmentation"in Multilink

Lijesh NC
Level 1
Level 1

Hi All,

Would like to know more about  fragmentation in  MLPP.In which all cases "No PPP Multilink fragmentation" is configurd what differance it make in multilink.Is it enabled by default.

Kinldy help in understanding this better.

interface Multilink21
description ***** Test  *****
ip vrf forwarding Test_Mul
ip address 10.1.100.100 255.255.255.252
no ip directed-broadcast
no cdp enable
ppp multilink
no ppp multilink fragmentation ========= ???!!!!!
multilink-group 21

Thanks in advance.

Lijesh

2 Accepted Solutions

Accepted Solutions

Lei Tian
Cisco Employee
Cisco Employee

Hi Lijesh,

MLPPP fragmentation is enable by default.

The Multilink PPP feature provides load balancing functionality over  multiple WAN links, while providing multivendor interoperability, packet  fragmentation and proper sequencing, and load calculation on both  inbound and outbound traffic. The Cisco implementation of MLP supports  the fragmentation and packet sequencing specifications in RFC 1990.  Additionally, you can change the default endpoint discriminator value  that is supplied as part of user authentication. Refer to RFC 1990 for  more information about the endpoint discriminator.

MLP allows packets to be fragmented and the fragments to be sent at the  same time over multiple point-to-point links to the same remote address.  The multiple links come up in response to a defined dialer load  threshold. The load can be calculated on inbound traffic, outbound  traffic, or on either, as needed for the traffic between the specific  sites. MLP provides bandwidth on demand and reduces transmission latency  across WAN links.

Enabling fragmentation reduces the delay latency among bundle links, but  adds some load to the CPU. Disabling fragmentation may result in better  throughput.

If your data traffic is consistently of a similar size, we recommend  disabling fragmentation. In this case, the benefits of fragmentation may  be outweighed by the added load on the CPU.

http://www.cisco.com/en/US/partner/docs/ios/dial/configuration/guide/dia_media-ind_multi_ppp_ps6441_TSD_Products_Configuration_Guide_Chapter.html

HTH,

Lei Tian

View solution in original post

Hi Lijesh,

The no ppp multilink fragmentation command was changed to ppp multilink fragment disable. The no ppp multilink fragmentation command was recognized and accepted through Cisco IOS Release 12.2.

' delay latency among bundle links' refers to when you have mix large size  and small size delay sensitive packets pass the mlppp bundle; with fragmentation and LFI, small size packet doesn't need to wait for large size packet be processed first, it will be transmitted between fragments of the  large packet hence reduce the serialization delay.

HTH,

Lei Tian

View solution in original post

9 Replies 9

Lei Tian
Cisco Employee
Cisco Employee

Hi Lijesh,

MLPPP fragmentation is enable by default.

The Multilink PPP feature provides load balancing functionality over  multiple WAN links, while providing multivendor interoperability, packet  fragmentation and proper sequencing, and load calculation on both  inbound and outbound traffic. The Cisco implementation of MLP supports  the fragmentation and packet sequencing specifications in RFC 1990.  Additionally, you can change the default endpoint discriminator value  that is supplied as part of user authentication. Refer to RFC 1990 for  more information about the endpoint discriminator.

MLP allows packets to be fragmented and the fragments to be sent at the  same time over multiple point-to-point links to the same remote address.  The multiple links come up in response to a defined dialer load  threshold. The load can be calculated on inbound traffic, outbound  traffic, or on either, as needed for the traffic between the specific  sites. MLP provides bandwidth on demand and reduces transmission latency  across WAN links.

Enabling fragmentation reduces the delay latency among bundle links, but  adds some load to the CPU. Disabling fragmentation may result in better  throughput.

If your data traffic is consistently of a similar size, we recommend  disabling fragmentation. In this case, the benefits of fragmentation may  be outweighed by the added load on the CPU.

http://www.cisco.com/en/US/partner/docs/ios/dial/configuration/guide/dia_media-ind_multi_ppp_ps6441_TSD_Products_Configuration_Guide_Chapter.html

HTH,

Lei Tian

Hi Lei ,

Thanks for the detailed explanation.Its partially clear now.I tried to execute the same in lab just to varify how its happening but failed.I couldnot find the option called "no ppp multilink fragmentation" under multilink below is the out put. ,Would be helpfull if you can guide some more on this .Also i have a query on "" delay latency among bundle links"" please help in understanding this better  and sorry if this is a very basic quiestion.


R1(config-if)#no ppp multilink fragment ?
  delay    Specify the maximum delay for each fragment
  disable  Suppress multilink fragmentation when possible
  maximum  Specify the maximum number of fragments per transmitted packet

R1(config-if)#

Thanks in advance.

Lijesh

Hi Lijesh,

The no ppp multilink fragmentation command was changed to ppp multilink fragment disable. The no ppp multilink fragmentation command was recognized and accepted through Cisco IOS Release 12.2.

' delay latency among bundle links' refers to when you have mix large size  and small size delay sensitive packets pass the mlppp bundle; with fragmentation and LFI, small size packet doesn't need to wait for large size packet be processed first, it will be transmitted between fragments of the  large packet hence reduce the serialization delay.

HTH,

Lei Tian

Hi Lijesh/ Lei Tian,

I  was going through this very informative post and just thought of adding one more point regarding this "fragmentation"
command.

The decision to issue "ppp multilink fragment disable" should also include the router platform that is being used. For those platforms where the fragmentation and reassembly is done is hardware like the SPA's in C12k or 7600, disabling the fragmentation should not make any improvements in the CPU usage.  If the customer is using T1/E1, T3/E3 or Sonet SPA's the HDLC controller on the SPA does the fragmentation and reassembly based on the fragment size configured on the SPA.

So the bottom line is " fragment disable" might improve the performance for multilink in software, it will not  give the desired result when the multilink is done in hardware.

Thanks

Shabeer

gnijs
Level 4
Level 4

I have a small additional question: correct me if i am wrong:

if i enable fragmentation, a packet is split across all links, for example a 1000 byte packet and 4 T1/E1 links, the packet is split in 250 byte packets which are sent on all 4 lines simultaniously and re-assembled on the other end again.

now, if i disable fragmentation, what kind of load-balancing is now active on the multilink ppp ? If i send the same 1000 byte packet, the router selects one of the 4 links and transmits the whole packet on it ? based on what ? source/destination ip address/mac address ???

regards,

Geert

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 i enable fragmentation, a packet is split across all links, for example a 1000 byte packet and 4 T1/E1 links, the packet is split in 250 byte packets which are sent on all 4 lines simultaniously and re-assembled on the other end again.

If I remember correctly, you actually also define a fragmentation size.  Your single 1000 byte packet might pass thru unfragmented or it might be pass thru as fragments, say, of 75 bytes each.  Basically you want to set fragment size, so small priority packets don't have to wait for too large fragments.

now, if i disable fragmentation, what kind of load-balancing is now active on the multilink ppp ? If i send the same 1000 byte packet, the router selects one of the 4 links and transmits the whole packet on it ? based on what ? source/destination ip address/mac address ???

I believe the whole packet is just sent on the 1st available link.  (Packets can arrive out of sequence, but MLPPP allows and corrects for that on the receiver.)

Hello JosephDoherty,

thanks for the information, i have a multilink in my network where in sometimes one of the link will be goin down, and because of it, network will be suffering drops and needs to shut the interface and bring it out of the multilink and when the link goes up then I need to add it back to the bundle. will the command "ppp multilink fragment disable" can help me in this regards, what are the supportive command I need to add just in case.

Regards
Thanveer
"Everybody is genius. But if you judge a fish by its ability to climb a tree, it will live its whole life believing that it is a stupid."

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

In theory, a link going down or up in a MLPPP bundle should be handled by MLPPP.  You shouldn't need to manually remove the port from MLPPP.

What platform(s) and IOSs are being used on the two sides of the MLPPP bundle?  What kind of MLPPP links?

Dear JosephDoherty,

Sorry for the late reply, Actually I was busy practing one of my exams, so the reason i could not reply you and I am once again sorry that I can give you the information as I hav already resigned from the company.

Thanks for your intrest, If you have something to comment after this, you are welcome.

Regards
Thanveer
"Everybody is genius. But if you judge a fish by its ability to climb a tree, it will live its whole life believing that it is a stupid."

Review Cisco Networking products for a $25 gift card