author: ioannis sourdis , sri harsha katamaneni publisher: ieee asap,2011 presenter: jia-wei yo

19
1 Author: Ioannis Sourdis , Sri Harsha Katamaneni Publisher: IEEE ASAP,2011 Presenter: Jia-Wei Yo Date: 2011/11/16 Longest prefix Match and Updates in Range Tries

Upload: airlia

Post on 23-Feb-2016

59 views

Category:

Documents


0 download

DESCRIPTION

Longest prefix Match and Updates in Range Tries. Author: Ioannis Sourdis , Sri Harsha Katamaneni Publisher: IEEE ASAP,2011 Presenter: Jia-Wei Yo Date: 2011/11/16. Introduction. B asic Range Trie does not support LPM, while updating the data structure in hardware is not simple . - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Author: Ioannis Sourdis , Sri Harsha Katamaneni Publisher: IEEE ASAP,2011 Presenter: Jia-Wei Yo

1

Author:Ioannis Sourdis , Sri Harsha KatamaneniPublisher:IEEE ASAP,2011Presenter: Jia-Wei YoDate:2011/11/16

Longest prefix Match and Updates in Range Tries

Page 2: Author: Ioannis Sourdis , Sri Harsha Katamaneni Publisher: IEEE ASAP,2011 Presenter: Jia-Wei Yo

2

Introduction Basic Range Trie does not support LPM, while updating

the data structure in hardware is not simple. describe a technique for storing and updating prefixes in a

Range Trie in order to support LPM; present the new Range Trie hardware design, prototyped in

a Virtex4-60 FPGA, and synthesized for 90-nm ASICs; measure the memory overhead of supporting LPM in a

Range Trie using real Internet routing tables; extract traces of Internet router updates for a period of 12

months and evaluate the performance of our proposal.

Page 3: Author: Ioannis Sourdis , Sri Harsha Katamaneni Publisher: IEEE ASAP,2011 Presenter: Jia-Wei Yo

3

*Range Trie(1/4)

Page 4: Author: Ioannis Sourdis , Sri Harsha Katamaneni Publisher: IEEE ASAP,2011 Presenter: Jia-Wei Yo

4

*Range Trie(2/4)

 

Page 5: Author: Ioannis Sourdis , Sri Harsha Katamaneni Publisher: IEEE ASAP,2011 Presenter: Jia-Wei Yo

5

*Range Trie(3/4)

Update complexity

(a) O(n)(b) O(m)(c) O()

Page 6: Author: Ioannis Sourdis , Sri Harsha Katamaneni Publisher: IEEE ASAP,2011 Presenter: Jia-Wei Yo

6

*Range Trie(4/4)

Page 7: Author: Ioannis Sourdis , Sri Harsha Katamaneni Publisher: IEEE ASAP,2011 Presenter: Jia-Wei Yo

7

Range Tries With LPM & incremental Update(1/7)

Range trie structure maintained three field: the compared address part values, some control bits for the comparisons a pointer to the next level

For LPM support we have added the lengths the action pointers of the prefixes stored in the node the action of the corresponding prefix stored at each

parent

Page 8: Author: Ioannis Sourdis , Sri Harsha Katamaneni Publisher: IEEE ASAP,2011 Presenter: Jia-Wei Yo

8

Range Tries With LPM & incremental Update(2/7)

lb_up : greater equal than b2hb_up : lower than d2

Page 9: Author: Ioannis Sourdis , Sri Harsha Katamaneni Publisher: IEEE ASAP,2011 Presenter: Jia-Wei Yo

9

Range Tries With LPM & incremental Update(3/7)

Check whether the newly inserted prefix is longer than the existing one

Only then the new prefix in inserted in the corresponding entry, by updating its prefix length and action pointer

The prefix is deleted from the entries stored and replaced with the next longer prefix

Page 10: Author: Ioannis Sourdis , Sri Harsha Katamaneni Publisher: IEEE ASAP,2011 Presenter: Jia-Wei Yo

10

Range Tries With LPM & incremental Update(4/7)

Inserting a new prefix bound in a node can be complex since it require to support Range trie optimization employed to reduce the number of address bits per comparison.

In hardware would require complex logic and hence slow circuit.

Split trie in two part : fixed part , updatable part.

=>bounds can be inserted or deleted only at the updatable part of the tree

Page 11: Author: Ioannis Sourdis , Sri Harsha Katamaneni Publisher: IEEE ASAP,2011 Presenter: Jia-Wei Yo

11

Range Tries With LPM & incremental Update(5/7)

To reduce the complexity of inserting and deleting bounds in the updatable-levels, we apply in the corresponding nodes only Address Alignment.

=>store address of length This reduces the number of bits per

comparison in the updatable part and allows all stored addresses to have the same width and hence to simplify updates.

Page 12: Author: Ioannis Sourdis , Sri Harsha Katamaneni Publisher: IEEE ASAP,2011 Presenter: Jia-Wei Yo

12

Range Tries With LPM & incremental Update(6/7)

Range trie ( software ) implementations allow node-split or –merges during inserting and deletion of prefix bounds.

=>become balanced tree

=>requires to feed-back and update the parent of the splitted/merged node=>stall pipeline for O() cycle per updaten : number of leafs m : branching factor in tree

Page 13: Author: Ioannis Sourdis , Sri Harsha Katamaneni Publisher: IEEE ASAP,2011 Presenter: Jia-Wei Yo

13

Range Tries With LPM & incremental Update(7/7)

Range Trie implementation has a limited number of spare tree-levels to support the updatable tree part.=>stall incoming address lookup requests for a few hundreds of cycles=>tree-rebuild : infrequent

Page 14: Author: Ioannis Sourdis , Sri Harsha Katamaneni Publisher: IEEE ASAP,2011 Presenter: Jia-Wei Yo

14

Hardware Design(1/2)

Page 15: Author: Ioannis Sourdis , Sri Harsha Katamaneni Publisher: IEEE ASAP,2011 Presenter: Jia-Wei Yo

15

Hardware Design(2/2)

=> Support one lookup per cycle and update four cycles

Page 16: Author: Ioannis Sourdis , Sri Harsha Katamaneni Publisher: IEEE ASAP,2011 Presenter: Jia-Wei Yo

16

Experimental Result(1/4)

Page 17: Author: Ioannis Sourdis , Sri Harsha Katamaneni Publisher: IEEE ASAP,2011 Presenter: Jia-Wei Yo

17

Experimental Result(2/4)

Page 18: Author: Ioannis Sourdis , Sri Harsha Katamaneni Publisher: IEEE ASAP,2011 Presenter: Jia-Wei Yo

18

Experimental Result(3/4)

Page 19: Author: Ioannis Sourdis , Sri Harsha Katamaneni Publisher: IEEE ASAP,2011 Presenter: Jia-Wei Yo

19

Experimental Result(4/4)