Skip Links

Network World

Dennis Hartmann
Back to Cisco Subnet

Cisco Unified Communications

by Dennis Hartmann
Previous Article Next Article

Cisco QoS: Link Fragmentation and Interleaving

By Dennis Hartmann on Wed, 03/04/09 - 2:54pm.
Newsletter Signup

In this blog, we will discuss link fragmentation and interleaving (LFI). The goal of LFI is to lower serialization delay on slow speed links. Serialization delay is the amount of time it takes a router to serialize a packet onto a WAN link. The goal of LFI is to ensure serialization delay never takes more than 10ms. Serialization delay with the default Ethernet maximum transmission unit (MTU) of 1518 bytes is only a consideration if the WAN link is 768kbps or slower. The Cisco router displays the MTU size as 1500 bytes, but the Cisco router does not take the layer 2 Ethernet framing into consideration (18 bytes/octets). Serialization delay considerations would change drastically if Ethernet jumbo frames were enabled on the interface. Ethernet jumbo frames are technically 9KB (9,216 bytes), but most hardware limits Ethernet jumbo frames to 9000 bytes. Ethernet jumbo frames are not recommended over slow speed interfaces and seldom considered when referring to LFI technology.

AutoQoS will turn LFI on by default for any interfaces running at or below 768kbps. I personally try to avoid turning on LFI because it is a very processor intensive function. Deployments that have low propagation delays (New York to Washington D.C) normally do not require LFI technology if the interface is running at 512kbps or above. Cisco’s goal of 10ms is generic and can be avoided if a one-way end to end delay budget of under 150ms can be achieved while avoiding LFI technology.

The following information is the serialization delay of an interface when transmitting full 1500 byte MTU size frames (Ethernet default):

768kbps = 15ms
512kbps = 23ms
256kbps = 46ms
128kbps = 93ms
64kbps = 187ms
56kbps = 214ms

LFI technology will fragment the large 1500byte packet into smaller packet sizes that will result in a lower serialization delay. Voice over IP media (RTP) is interleaved between the resulting data fragments to ensure each 20ms voice sample does not incur significant serialization delay.

LFI is not supported over HDLC encapsulations on WAN interfaces. LFI is supported over multilink PPP (ML-PPP) interface or frame relay interfaces (FRF.12). HDLC point-to-point links will need to be converted to ML-PPP in order to perform LFI technology. The following example configuration could be used for ML-PPP LFI:

interface Serial0/3/0
no ip address
encapsulation ppp
ppp multilink
ppp multilink group 100

interface Multilink100
ip address 10.1.1.1 255.255.255.252
ppp multilink
ppp multilink group 1
ppp multilink fragment delay 10
ppp multilink interleave
bandwidth 512
service-policy output llq-policy

The physical Serial 0/3/0 interface must not have an IP address associated with it. The IP address configuration is moved to the logical multilink interface. The PPP multilink group command is used to map the physical serial interface to the logical multilink interface. PPP multilink is normally used to combine multiple physical interfaces or B channels in ISDN into one logical entity. We are using PPP multilink on a single interface to take advantage of the fragmentation and interleaving feature that was developed for ML-PPP. The fragment delay 10 command instructs the interface to use 10ms as a serialization delay as a target based on the interface MTU of 1500 bytes and the bandwidth of 512kbps. The bandwidth 512 command must be specified because the router assumes a bandwidth of 1544kbps by default. The LLQ policy is attached to the multilink interface, not the physical serial interface.

FRF.12 is a frame relay forum standard for doing link fragmentation and interleaving over frame relay interfaces. The following configuration example is the recommended configuration for a 512kbps frame relay circuit (assume CIR matches port speed):

interface serial 0/3/0
encapsulation frame-relay
!
interface serial 0/0.200 point-to-point
frame-relay interface-dlci 200
class frf
!
map-class frame-relay frf
frame-relay fragment 640
service-policy output llq-policy

Frame Relay LFI policies are more complex than ML-PPP LFI policies because the byte size of the fragmentation to be used must be specified in the map class policy that is mapped to the frame sub-interface with the class command. The recommended fragment sizes for frame relay port speed are as follows:

768kbps = 1000 bytes
512kbps = 640 bytes
256kbps = 320 bytes
128kbps = 160 bytes
64kbps = 80 bytes
56kbps = 70 bytes

I normally do not recommend LFI technologies unless they are absolutely necessary. This blog wraps up the QoS blog series. Please post any questions you may have related to QoS.

REFERENCES
Global Knowledge – Cisco Quality of Service class
http://www.globalknowledge.com/training/course.asp?pageid=9&courseid=757...

Cisco QoS SRND
http://www.cisco.com/univercd/cc/td/doc/solution/esm/qossrnd.pdf

Tags

Daer Dennis Hartmann, Could

0

Daer Dennis Hartmann,
Could you pls let me know, can I enable LFI on interface BVI (bride virtual interface)?
Thank you very much.
Br,
Phisitchai

Significance of the term "Link"

0

Hi,

Why this feature is called "Link" fragmentation and not Packet fragmentation or Frame Fragmentation. What does the term "Link" signify?

In case of MLPPP, fragmentation is at the level of MLPPP and still it is called "Link" Fragmentation. Why?

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

Post new comment

The content of this field is kept private and will not be shown publicly.
  • You can use BBCode tags in the text.
  • Lines and paragraphs break automatically.
  • Allowed HTML tags:



More information about formatting options

CAPTCHA
This question is for testing whether you are a human visitor and to prevent automated spam submissions.
Welcome, visitor. Register Log in
About Cisco Unified Communications

Dennis Hartmann, CCIE No. 15651, is a consultant with www.highpoint.com and author of . Dennis is also a lead instructor at Global Knowledge. Dennis has various certifications, including the Cisco CCVP, CCSI, CCNP, CCIP, and the Microsoft MCSE.  Dennis has various specializations including unified communications, data center, routing & switching, service provider (MPLS and optical).  Dennis has worked for various Fortune 500 companies, including AT&T, Sprint, Merrill Lynch, KPMG, and Cabletron Systems. He lives with his wife and children in Hopewell Junction, New York.

Global Knowledge

  • RSS
  • Contact Requires Login
Archives
January 2010
December 2009
November 2009
October 2009
September 2009
August 2009
July 2009
June 2009
May 2009
April 2009
March 2009
February 2009
January 2009
December 2008
November 2008
October 2008
September 2008
August 2008
July 2008
Categories
Careers
Cisco
ESW PoE 802.3af 802.3at ePoE
ISR
ISR ISRG2
PVDM3 ISR
VML
bat