Infinite CacheFlow in Software-Defined Networking Na Naga Katta Omid - - PowerPoint PPT Presentation

▶
infinite cacheflow in software defined networking
SMART_READER_LITE
LIVE PREVIEW

Infinite CacheFlow in Software-Defined Networking Na Naga Katta Omid - - PowerPoint PPT Presentation

Mar 22, 2023 107 likes •464 views

Infinite CacheFlow in Software-Defined Networking Na Naga Katta Omid Alipourfard, Jennifer Rexford, David Walker Princeton on University Recent News 2 SDN Promises Flexible Policies Controller Switch TCAM 3 SDN Promises Flexible Policies

slide-1
SLIDE 1

Infinite CacheFlow in Software-Defined Networking

Na Naga Katta

Omid Alipourfard, Jennifer Rexford, David Walker Princeton

  • n University
slide-2
SLIDE 2

Recent News

2

slide-3
SLIDE 3

SDN Promises Flexible Policies

3

Controller

Switch TCAM

slide-4
SLIDE 4

SDN Promises Flexible Policies

4

Controller

Switch TCAM

Lot of fine- grained rules

slide-5
SLIDE 5

SDN Promises Flexible Policies

5

Controller

Lot of fine- grained rules

slide-6
SLIDE 6

SDN Promises Flexible Policies

6

Controller

Lot of fine- grained rules

slide-7
SLIDE 7

SDN Promises Flexible Policies

7

Controller

Lot of fine- grained rules

slide-8
SLIDE 8

SDN Promises Flexible Policies

8

Controller

Limited rule space!

slide-9
SLIDE 9

SDN Promises Flexible Policies

9

Controller

Limited rule space! What now?

slide-10
SLIDE 10

State of the Art

10

Hardware Switch Software Switch Rule Capacity Low (~2K-4K) High Lookup Throughput High (>400Gbps) Low (~40Gbps) Port Density High Low Cost Expensive Relatively cheap

slide-11
SLIDE 11

TCAM as cache

11

CacheFlow TCAM Controller

S2 S1

Software Switches

slide-12
SLIDE 12

TCAM as cache

12

CacheFlow TCAM Controller

S2 S1

<5% rules cached

slide-13
SLIDE 13

TCAM as cache

13

CacheFlow TCAM Controller

S2 S1 low expected cache-misses

slide-14
SLIDE 14
  • High throughput + high rule space

TCAM as cache

14

CacheFlow TCAM Controller

S2 S1

slide-15
SLIDE 15
  • High throughput + high rule space

TCAM as cache

15

CacheFlow TCAM Controller

S2 S1 Flexible Deployment

slide-16
SLIDE 16
  • Abstraction of an “infinite” switch

Ø Correct: realizes the policy Ø Efficient: high throughput & large tables Ø Transparent: unmodified applications/switches

16

A Correct, Efficient and Transparent Caching System

slide-17
SLIDE 17
  • 1. Correct Caching
slide-18
SLIDE 18

Caching under constraints

18

Easy: Cache rules greedily

Rule Match Action Priority Traffic R1 110 Fwd 1 3 10 R2 100 Fwd 2 2 60 R3 101 Fwd 3 1 30

slide-19
SLIDE 19

Caching Ternary Rules

19

Rule Match Action Priority Traffic R1 11* Fwd 1 3 10 R2 1*0 Fwd 2 2 60 R3 10* Fwd 3 1 30

  • Greedy strategy breaks rule-table semantics
slide-20
SLIDE 20

Caching Ternary Rules

20

Rule Match Action Priority Traffic R1 11* Fwd 1 3 10 R2 1*0 Fwd 2 2 60 R3 10* Fwd 3 1 30

  • Greedy strategy breaks rule-table semantics

Rules Overlap!

slide-21
SLIDE 21

Caching Ternary Rules

21

Rule Match Action Priority Traffic R1 11* Fwd 1 3 10 R2 1*0 Fwd 2 2 60 R3 10* Fwd 3 1 30

  • Greedy strategy breaks rule-table semantics
  • Beware of switches that claim large rule tables

Rules Overlap!

slide-22
SLIDE 22

Dependency Graph

22

R3 R2 R1 R6 R5 R4 (*)

Rule Match Action Priority Traffic R1 0000 Fwd 1 6 10 R2 000* Fwd 2 5 20 R3 00** Fwd 3 4 90 R4 111* Fwd 4 3 5 R5 11** Fwd 5 2 10 R6 1*** Fwd 6 1 120

slide-23
SLIDE 23

Dependent-Set Caching

23

R3 R2 R1 R6 R5 R4 (*)

  • All descendants in DAG are dependents
  • Cache dependent rules for correctness
slide-24
SLIDE 24
  • 2. Efficient Caching
slide-25
SLIDE 25

Dependent-Set Overhead

25

R3 R2 R1 R6 R5 R4 (*)

Too Costly?

slide-26
SLIDE 26

Cover-Set

26

Rule Match Action R1 000 Fwd 1 R2 00* Fwd 2 R3 0** Fwd 3 R4 11* Fwd 4 R5 1*0 Fwd 5 R5^ 1*0 To_SW R6 10* Fwd 6 (*) *** To_SW

R6 R5 R4 (*) R5^ R3 R2 R1

Cover rule

slide-27
SLIDE 27

Dependency Splicing reduces rule cost!

27

Dependent-Set Cover-Set

Rule Space Cost

slide-28
SLIDE 28

Deep Dependency Chains – Clear Gain

28

  • ClassBench Generated ACL

20 40 60 80 100 0.5 1 2 5 10 25 50

% Cache-hit traffic % TCAM Cache Size (Log scale)

Cover-Set Algo Dependent-Set Algo

slide-29
SLIDE 29

Shallow Dependency Chains – Marginal Gain

29

20 40 60 80 100 1 2 5 10 25 50

% Cache-hit traffic % TCAM Cache Size (Log scale)

Dependent-Set Algo Cover-Set Algo

  • Stanford Backbone Routing table
slide-30
SLIDE 30
  • 3. Transparent Caching
slide-31
SLIDE 31

31

CacheFlow HW_Cache (TCAM) Controller

S4 S1 S2 S3

OpenFlow Datapath

  • 3. Transparent Design
slide-32
SLIDE 32

32

CacheFlow HW_Cache (TCAM) Controller

S4 S1 S2 S3

Virtual switch OpenFlow Datapath

  • 3. Transparent Design
slide-33
SLIDE 33

33

CacheFlow HW_Cache (TCAM) Controller

S4 S1 S2 S3

OpenFlow Datapath

Emulates counters, barriers, timeouts etc.

  • 3. Transparent Design

Virtual switch

slide-34
SLIDE 34
  • Rule caching for OpenFlow rules

Ø Dependency analysis for corr correctness ectness Ø Splicing dependency chains for effi fficiency Ø Tr Transparent design

Conclusion

34

slide-35
SLIDE 35

Infinite Ca$heFlow in SDN

Question

  • ns?