/
Towards an Elastic Distributed SDN Controller Towards an Elastic Distributed SDN Controller

Towards an Elastic Distributed SDN Controller - PowerPoint Presentation

kittie-lecroy
kittie-lecroy . @kittie-lecroy
Follow
543 views
Uploaded On 2015-10-06

Towards an Elastic Distributed SDN Controller - PPT Presentation

Advait Dixit Fang Hao Sarit Mukherjee TV Lakshman Ramana Kompella Physical Network Infrastructure SDN Control Plane Distributed Control Plane Single point of failure ID: 151970

controller distributed master switch distributed controller switch master slave openvswitch application control plane flow change migration controllers sdn phase host arrival load

Share:

Link:

Embed:

Download Presentation from below link

Download Presentation The PPT/PDF document "Towards an Elastic Distributed SDN Contr..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.


Presentation Transcript

Slide1

Towards an Elastic Distributed SDN Controller

Advait Dixit,

Fang

Hao

,

Sarit

Mukherjee,

T.V

.

Lakshman

,

Ramana

KompellaSlide2

Physical Network Infrastructure

SDN Control Plane

Distributed Control Plane

Single point of failure

Performance bottleneckSlide3

Spatial Partitioning

OverloadSlide4

Growing the Control PlaneSlide5

Shrinking the Control PlaneSlide6

Goals

Build a distributed control plane which

Load balances

Grows

ShrinksThis requiresLoad estimation at controllersSwitch migration protocolSlide7

Naïve Switch Migration

SLAVE

MASTER

MASTER

SLAVE

Role Change to MasterSlide8

Problem With Naïve Switch Migration

Packet-In

MASTER

MASTER

SLAVE

SLAVE

Packet-Out

Role Change to Master

Packet-Out from Slave is droppedSlide9

Migration Protocol Requirements

Safety: Exactly

1 controller processes every

message from the switch

Liveness: For each switch, at least 1 controller is active at all timesOpenflow compliantSlide10

Flow-Add

Flow-Delete

Flow-Removed

4-Phase Switch Migration Protocol

MASTER

SLAVE

Flow-Removed

EQUAL

Role Change to Master

SLAVE

MASTER

Role Change to Equal

Barrier Request

Barrier Reply

Phase 1: Change from Slave to Equal

Phase 2: Insert and remove dummy flow

Phase 3: Flush in-flight message

Phase 4: Change from Equal to MasterSlide11

A Mininet

Testbed

Problem: Cannot generate sufficient traffic for a large distributed control plane

v

eth

Pair

OpenvSwitch

OpenvSwitch

veth Pair

OpenvSwitch

veth

PairEmulation HostSlide12

A Multi-Host Mininet

Testbed

OpenvSwitch

OpenvSwitch

OpenvSwitch

OpenvSwitch

OpenvSwitch

OpenvSwitch

Emulation Host

OpenvSwitch

OpenvSwitch

OpenvSwitch

GRE Tunnel

`

GRE Tunnel

GRE Tunnel

Emulation Host

Emulation HostSlide13

EvaluationSlide14

EvaluationSlide15

Next Step: ElastiCon

Physical Network Infrastructure

Core Controller Module

Application 1

Application 2

Core Controller Module

Application 1

Application 2

Distributed Data Store (e.g.,

Hazelcast

)

Node 1

Load Measurements

Load Balance

Scale Up

Scale Down

Load Adaptation Decisions

Distributed SDN Control Plane

Node 2

Actions:

Migrate switch

Remove controller

Add controllerSlide16

Thank youSlide17

Distributed Control Plane in a Datacenter

Median flow arrival rate requires 1-5 controllers, peak requires 150*

Distributed

control plane should

grow and shrinkFlow arrival rates vary across switches and timeNeed a switch migration (handover) protocol

* Calculations based on Benson et al, IMC 2010Slide18

Distributed SDN Control Plane

Physical Network Infrastructure

Core Controller Module

Distributed Data Store (e.g.,

Hazelcast

)

Controller Node 1

Controller Node 2

Application 1

Application 2

Core Controller Module

Application 1

Application 2

Hazelcast

Client StubSlide19

Next Steps

Build a control loop and algorithms for dynamically:

Changing

switch-controller mapping based

Growing and shrinking distributed controllerThanks! Questions?Slide20

EvaluationSlide21

Need title for this slide

For a data center with 100K hosts:

Peak flow arrival rate = 300M* flows/sec

Median flow arrival rate = 1.5-10M* flows/sec

Impossible to predict flow arrival rates at a switchImplications:Distributed controller needs to grow and shrinkNeed a switch migration protocol

* From Benson et al, IMC 2010Slide22

New Problems in Distributed SDN Controllers

How to manage distributed state?

Where to place controllers?

How to write distributed controller applications?

How many controllers?Which switch connects to which controller?Slide23

New Problems in Distributed SDN Controllers

How to manage distributed state?

Where to place controllers?

How to write distributed controller applications?

How many controllers?Which switch connects to which controller?Slide24

Distributed SDN Controller

Distributed Global Network State

Physical Network Infrastructure

Application 1

Application 2

Application 3

Application 4Slide25

Problem Statement

How many servers?

How to determine switch-controller mapping?Slide26

Naïve Switch Migration//Change title?

SLAVE

MASTER

MASTER

SLAVE

Role Change to Master

SLAVE

SLAVE

SLAVE