Monitoring GENI Networks

33 

Loading....

Loading....

Loading....

Loading....

Loading....

Full text

(1)

Monitoring GENI Networks

Ji G iffi

d Z

i F i

Jim Griffioen and Zongming Fei

Laboratory for Advanced Networking

U i

it f K t k

University of Kentucky

Lexington, KY

(2)

Project Participants

Project Part c pants

University of Kentcuky

 James Griffioen (PI)  Zongming Fei

 Hussamuddin Nasir (Lead Programmer)  Xiongqi Wu (Research Assistant)

 Xiongqi Wu (Research Assistant)  Jeremy Reed (Research Assistant)  Charles Carpenter (Programmer)  Lowell Pike (Network Administrator)

W d M l (T h i l S )

 Woody Marvel (Technical Support)

AT&T Labs Research

 Jacobus van der Merwe  Emmanuil Mavrogiorgis

Internet2

 Eric Boyd

University of Kentucky

 Brian Cashman

(3)

Talk Outline

alk Outl ne

Motivations for Monitoring GENI Networks

Motivations for Monitoring GENI Networks

Architecture for a ProtoGENI Monitoring

and Instrumentation Tool

and Instrumentation Tool

Using the Instrumentation Tool

Future Work -- Monitoring Virtual Routers in

a ProtoGENI network

(4)

I t d ti d M ti ti

Introduction and Motivations

(5)

GENI

A national-scale suite of infrastructure for long-running,g g realistic experiments in Network Science and Engineering.

Deeply programmable

Virtualized

Sensor Network

Programmable, federated, with end-to-end virtualized slices

Federated International Infrastructure

University of Kentucky Joint Techs 2011 January 31, 2011

(6)

ProtoGENI

ProtoGENI

ProtoGENI components

p

A nationwide high-speed

backbone on Internet2’s wave

infrastructure

Slicable PCs with NetFPGA

cards embedded within the

cards embedded within the

backbone

A set of subnets including a

A set of subnets, including a

variety of wireless networks,

residential broadband, and

bl d l t

University of Kentucky Joint Techs 2011 January 31, 2011

(7)

Motivations for Monitoring

GENI N

k

GENI Networks

o

Much of the work on testbeds has focused on

o

Much of the work on testbeds has focused on

creating, setting up, and running an experiment

o

Create application-specific virtual networks

o

Time scale: minutes

o

The real challenge can often be monitoring and

analyzing the behavior of an experiment; it can be

analyzing the behavior of an experiment; it can be

a very involved, time consuming, manual process

that is repeated many times.

o

It involves setting up and launching a monitoring

environment.

(8)

Project Goals/Objectives

Project Goals/Object ves

Th lti t bj ti i t

The ultimate objective is to

 Automate the task of deploying an instrumentation and

measurement infrastructure within a GENI slice.

m m f

 Provide users with a convenient and simple-to-use

interface to the measurement infrastructure.

P li i

t

t

d thi bj ti

Preliminary steps toward this objective

 Implement an instrumentation and measurement tools to

work in the ProtoGENI environment (INSTOOLS) work in the ProtoGENI environment (INSTOOLS)

(9)

Monitoring Challenges

o

Multiple uses

o

Debug

V if th fi ti

o Verify the configuration

o Verify the experiment is behaving correctly

o

Learn/Discover

o Understand what happens and why

o Understand what happens and why

o Measure performance

o

Requires

o

deciding what information should be monitored about the g

network, OS, and applications

o

selecting the level of detail for monitoring

o

monitoring desired information

filt i i f m ti t b t d

o

filtering information to be reported

o

collecting/transporting recorded data

o

making information available/viewable to users.

(10)

Monitoring Challenges

o

Scaling to large experiments.

Monitoring load on components

Monitoring load on components

Collect/transport instrumentation data

Without interfering with the experiment Collecting data from many components Collecting data from many components

o

Logging for history view and archival

purposes.

p p

o

Security and authorization to access

monitoring capabilities.

o

Sandboxing the monitoring system.

o

Accounting for monitoring activity.

(11)

A hit t

Architectrure

(12)

INSTOOLS Architecture

(Functional Components)

1. Setup: deploy and initialize topology-specific software and

services

2. Capture: capture measurement data

3.. Collection:o ct on move data to processing/storage environmentsmo ata to proc ss ng/storag n ronm nts 4. Storage: store data on a temporary, short term, long term,

and archival basis

5 Processing: filter convert aggregate summarize etc 5. Processing: filter, convert, aggregate, summarize, etc.,

data

6. Presentation: present data to users in meaningful ways 7 A P t ti t t d d t

7. Access Protection: protect resources and data

8. Measurement Control: Dynamically control the above

components

(13)

INSTOOLS Architecture

(Functional Components)

1. Setup: deploy and initialize topology-specific software and

services

2. Capture: capture measurement data

3.. Collection:o ct on move data to processing/storage environmentsmo ata to proc ss ng/storag n ronm nts 4. Storage: store data on a temporary, short term, long term,

and archival basis

5 Processing: filter convert aggregate summarize etc 5. Processing: filter, convert, aggregate, summarize, etc.,

data

6. Presentation: present data to users in meaningful ways 7 A P t ti t t d d t

7. Access Protection: protect resources and data

8. Measurement Control: Dynamically control the above

components

University of Kentucky Joint Techs 2011

Conventional network management solutions exist

(14)

INSTOOLS Architecture:

(

d

)

(Automated Setup)

Slice 1

MC

Slice 1

Topology

MC

Physical Links Physical Links, VLANs, or GRE Tunnels Physical Links, VLANs, GRE Tunnels, or normal IP path

Sli

2

MC

MCs are automatically

dd d t

ll t

University of Kentucky Joint Techs 2011

Slice 2

Topology

added to collect

measurement data

(15)

Implemenation Approach

Implemenat on Approach

(16)

Implemenation Approach

Implemenat on Approach

(17)

U i th I t

t ti T l

Using the Instrumentation Tool

(18)

ProtoGENI System

Maps Interface

Scripts Interface

ProtoGENI

University of Kentucky Joint Techs 2011

Aggregate A Aggregate B Aggregate C

(19)

User creates a “slice” using one of

the ProtoGENI APIs.

Maps Interface

Scripts Interface

ProtoGENI

User’s Slice

University of Kentucky Joint Techs 2011

Aggregate A Aggregate B Aggregate C

(20)

INSTOOLS Software

User instruments the slice

MC

Software

SNMP, Netflow, OS Utilities, etc

Maps Interface

Scripts Interface

ProtoGENI

User’s Slice

University of Kentucky Joint Techs 2011

Aggregate A Aggregate B Aggregate C

(21)

INSTOOLS Software

INSTOOLS adds MC to slice

MC

Software

SNMP, Netflow, OS Utilities, etc

Maps Interface

Scripts Interface

ProtoGENI

MC

User’s Slice

University of Kentucky Joint Techs 2011

Aggregate A Aggregate B Aggregate C

(22)

INSTOOLS Software

INSTOOLS adds monitoring software to slice

resources

MC

Software

SNMP, Netflow, OS Utilities, etc

Maps Interface

Scripts Interface

ProtoGENI

MC

User’s Slice

University of Kentucky Joint Techs 2011

Aggregate A Aggregate B Aggregate C

(23)

INSTOOLS Software

User views measurement data on the MC.

MC

Software

SNMP, Netflow, OS Utilities, etc Web Server  Access to MC

Maps Interface

Scripts Interface

Access to MC

ProtoGENI

MC

User’s Slice

University of Kentucky Joint Techs 2011

Aggregate A Aggregate B Aggregate C

(24)

Portal for Monitoring

Portal for Mon tor ng

(25)

Portal for Monitoring

Portal for Mon tor ng

(26)

Traffic Graphs of a Experimental Node

(27)

F t

W k

Future Work

(28)

Future Work

M i

i Vi

l R

Monitoring Virtual Routers

Deploy “virtualizable” commercial routers

Deploy virtualizable commercial routers

(Juniper m7i) in the ProtoGENI backbone

that support commercial OS/software

that support commercial OS/software.

Add software support to these virtual

routers that will enable per slice

routers that will enable per-slice

monitoring and measurement.

D

l l d f

h ll

Develop tools and interfaces that will

allow slice users to use the measurement

University of Kentucky

infrastructure in simple and easy ways.

January 31, 2011 Joint Techs 2011

(29)

ProtoGENI ShadowNet Sites

University of Kentucky Joint Techs 2011 January 31, 2011

(30)

ProtoGENI Backbone Node Architecture

Internet 2

Gigabit Ethernet Switch Gigabit Ethernet Switch

Sliver 1 N etFPGA N etFPGA N etFPGA Sliver n N N N General Purpose Slivers Non-sliced PC

University of Kentucky Joint Techs 2011 January 31, 2011

Non-sliced PC

(31)

ProtoGENI Backbone Node Architecture

Internet 2

Gigabit Ethernet Switch Gigabit Ethernet Switch

Logical Router 1 Logical Router 2 Logical Router n Juniper M7i Router

Sliver 1 N etFPGA N etFPGA N etFPGA Sliver n Juniper Component Manager Virtual Server rfSONAR 1 rfSONAR n N N N ShadowBox Controller g pe r pe r General Purpose Slivers Measurement Slivers

Non-sliced PC GENI ShadowNet Router

University of Kentucky Joint Techs 2011 January 31, 2011

Non-sliced PC

(32)

Availability of Operational/Utilization Data

Infrastructure data:

We have the ability to

collect physical router data for network

p y

operations:

Router/Link up/down status

Li k tili ti (#bit #f

#

k t /

d)

Link utilization (#bits, #frames, #packets / second)

CPU utilization

Interface (link) error counters

Interface (link) error counters

Interface packet loss/drop counters

Slice-specific data

Slice specific data

Slice owner controls access to this data

May be made available to network operations

(33)

Thank You!

Q

i

?

Questions?

University of Kentucky Joint Techs 2011

This material is based upon work supported in part by the National Science Foundation under grant number CNS-0834243. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of GPO Technologies, Corp, the GENI Project Office, or the National Science Foundation.

Figure

Updating...

References

Updating...

Related subjects :