Autonomic Functions Coordination By The Example - PowerPoint Presentation

Slide1

Autonomic Functions Coordination By The Example

P. Peloso, L. Ciavaglia

ANIMA WG, IETF 96

th

Slide2

Objectives

2

| ANIMA | IETF 96

th

- Show

feasibility of

coordination between ASAs

- Identify required elements for a coordination function

- Derive requirements for ASA and ANI

Slide3

Explaining the use case

3

| ANIMA | IETF 96

th

Network composed

of

:

edge and core

routers

3

clients on the edges

2

types of traffic for each client (premium/best-effort)

6 LSP to cope with each traffic

LSP

1P

LSP

1BE

1

2

3

LSP

3P

LSP

3BE

LSP

2BE

LSP

2P

A

D

B

C

Slide4

Explaining the use case ASA sets

4

| ANIMA | IETF 96

th

2 ASAs

running on router A

Load

Balancing

ASA

Balances client traffic between network interfaces

Risk Aware

Routing ASA

Avoids paths at risk for premium traffic

1

2

3

A

D

B

C

ASA

LB

A

ASA

RA

A

Slide5

1

st

ASA – Traffic Engineering ASA

Deployable over edge router

When deployed

Monitors network interfaces loads

Modifies LSPs routes to balance load

5

| ANIMA| IETF 96

th

Slide6

2

nd

ASA – Risk Aware ASA

Deployable over edge router

When deployed

Monitors network interfaces risks of failure

Modifies Premium LSPs routes to reduce their failure exposition

6

| ANIMA| IETF 96

th

Slide7

ASA plugging to ANI and device

7

| ANIMA | IETF 96

th

Current ANIMA picture

Excerpt from Ref Model

ASA use GRASP signaling in-between them

ASA monitor the node and modify its state directly using either

NetConf

, OpenFlow, SNMP, CLI…

Control loop enforcement

Control loop enforcement

Node 1

Node

2

ANI

Autonomi

c

Function A

AF B

ASA

ASA

ASA

Control loop enforcement

GRASP Signaling

Slide8

Deployment of ASAs onto network equipment

8

| ANIMA| IETF 96

th

Deployment means:The process during which the ASA “gets in touch” with the device(s) it controls

During deployment:

ASA should establish session with the device(s) (credentials?)

ASA should collect from the device some setting info

(e.g.

nbr of interfaces and interfaces id)Use these collected info to compute its Instance Manifest

Compute the actual metricsCompute the actual parametersShare its Instance Manifest within the ANIEither a selective sharing or broadcast sharing

Slide9

1

st

ASA – Load Balancing ASA Instance Manifest

Deployed over router A:

Monitors network interfaces loads

Modifies LSPs routes to balance load

9

| ANIMA| IETF 96

th

<

InstanceManiofest

>

<

ASA_Class_ID

>

<Name>LoadBalancing</Name> <Provider>

AnimaCorp</Provider> <Version>1.0.0</Version>

</ASA_Class_ID>

<ASA_Instance_ID>3567456</ASA_Instance_ID>

<AcquiredInputs>

<InfoSpec> <name>

InterfaceLoad</name> <contentType

>Numeric</contentType>

<context> {

RtrA:IfcAB , RtrA:IfcAB} </context> </InfoSpec> </AcquiredInputs> <PossibleActions> <ActionSpec

> <name>LSPPath</name> <contentType>ERO</contentType> <context>{LSP1P

, LSP1BE, LSP2P, LSP2BE, LSP3P, LSP3BE} </context> </ActionSpec> </PossibleActions> </InstanceManifest>

LB

ERO@LSP

1PERO@LSP1BEERO@LSP2PERO@LSP

2BEERO@LSP3PERO@LSP3BELoad@IfcABLoad@IfcAC

Slide10

2

nd

ASA – Risk Aware ASA Instance Manifest

Deployed over router A:

Monitors network interfaces risks of failure

Modifies Premium LSPs routes to reduce their failure exposition

10

| ANIMA| IETF 96

th

<

InstanceManifest

>

<

ASA_Class_ID

>

<Name>RiskAware</Name> <Provider>

AnimaCorp</Provider> <Version>1.0.0</Version>

</ASA_Class_ID>

<ASA_Instance_ID>7167456</ASA_Instance_ID>

<AcquiredInputs>

<InfoSpec>

<name>InterfaceRisk</name>

<contentType>Probability

</

contentType>

<context> {RtrA:IfcAB , RtrA:IfcAB} </context> </InfoSpec> </AcquiredInputs>

<PossibleActions> <ActionSpec> <name>LSPPath</name> <contentType>ERO

</contentType> <context>{LSP1P, LSP2P, LSP3P} </context> </ActionSpec> </PossibleActions> </InstanceManifest>

RA

ERO@LSP1PERO@LSP2PERO@LSP3PRisk@IfcABRisk@Ifc

AC

Slide11

And now – ASA Execution

11

| ANIMA| IETF 96

th

ASA LB

A

ASA RA

A

A

CHAOS ?

Slide12

Conflict Identification

12

| ANIMA| IETF 96

th

Role:

Identify potential conflicts

Inputs:

Instance Manifests of All ASAs in the ANI

Outputs:

Groups of ASAs (or Autonomic Functions) that may be conflicting

Situated:Either as a part of the ANIOr sitting on top of ANI

(Can be distributed)

Slide13

Situated Conflict Identification entity

13

| ANIMA | IETF 96

th

Control loop enforcement

Control loop enforcement

Node 1

Node

2

Node

3

Node N

Autonomi

c

Function A

Autonomic Function B

ASA

ASA

ASA

ASA

ASA

ASA

ANI

Conflict

Identification

Slide14

Conflict identification

Aggregate the graphs provided by each ASA Instance Manifest

Identify

loops:

Loops formed

by different ASAs

Loops

sharing edges

Return list of ASAs implied in conflicting

loops

<

PotentialConflict>

<ConflictID

>1</

ConflictID>

<ConflictingASAs

> {7167456 , 3567456} </

ConflictingASAs></PotentialConflict>14

| ANIMA| IETF 96th

RA

LB

Slide15

And now – ASA Execution

15

| ANIMA| IETF 96

th

ASA LB

A

ASA RA

A

A

CHAOS ?

Slide16

Between ASA Coordination

16

| ANIMA| IETF 96

th

Role:

Address potential conflict between ASAs

Inputs:

Potential Conflict description from Conflict Identification

Outputs:

Control commands to ASAs

Situated:Either as a part of the ANIOr sitting on top of ANI

(Can be distributed)

Slide17

Situated Coordination entities

17

| ANIMA | IETF 96

th

Control loop enforcement

Control loop enforcement

Node 1

Node

2

Node

3

Node N

Autonomi

c

Function A

Autonomic Function B

ASA

ASA

ASA

ASA

ASA

ASA

ANI

Conflict

Identification

Between ASA

Coordination

Slide18

Between ASA coordination

18

| ANIMA| IETF 96

th

Receives a Potential Conflict description

Identify applicable algorithm:

Depending on available algorithms

Depending on control capacities disclosed by ASAs Instance Manifest (If applicable, otherwise supposed no more than start/stop)

Instantiate an algorithm process in charge of the ASA group

If applicable set algorithm parameters

Depending on ASA priorities (from Intents)

Depending on ASAs features additionally disclosed by Instance Manifest

Run the algorithm process, which sends control commands to ASA

(e.g. a random token determines which ASA can execute)

Slide19

Time separation method

19

| ANIMA| IETF 96

th

time

Load Balancing ASA

Risk Aware ASA

Allow the least impacting ASA to converge in-between two iterations

of the most impacting one

Slide20

Conclusion

20

| ANIMA| IETF 96

th

Conflict resolution based on common coordination components is feasible

Applicable

to ASA complying to

set of requirements (manifests)

Showing the step-by-step process

Showing the information to be conveyedShowing possible formats (XML based, but can be TLV based)

Providing a basic method to achieve the process (multiple algorithms possible)

Slide21

Conclusion Requirements to ASA

Control

when/how ASA

runs

Know what

ASA “does”

to the network

ASAs must follow a

defined

process

Instance

Manifest

Start/Stop

ASA

Life-cycle

21

| ANIMA| IETF 96th

Slide22

Conclusion ASA life-cycle

Installed

Instantiated

Operational

Instantiating

Un-Instantiating

Receives a

Set-up

cmd

Receives

revoke Mandate

Receives a

Mandate

Undeployed

Undeployed

Advertizes

its

Manifest

Receives a

Set-down

cmd

Advertizes

blank Manifest

22

| ANIMA | IETF 96

th

Slide23

Augmenting ANI with

ASA Life Cycle Management

23

| ANIMA | IETF 96

th

Control loop enforcement

Control loop enforcement

Node 1

Node

2

Node

3

Node N

Autonomi

c

Function A

Autonomic Function B

ASA

ASA

ASA

ASA

ASA

ASA

ANI

Conflict

Identification

Between ASA

Coordination

Life Cycle Management