Inferring the Topology and Traffic Load of Parallel Programs in a VM

environment

Ashish Gupta

Resource Virtualization

Winter Quarter Project

Motivation

• A distributed computing environment based on Virtual Machines

• Goal: Efficient execution of Parallel applications in such an environment

ParallelApplication Behavior

Intelligent Placement and virtual networking

of parallel applications

VM EncapsulationVirtual Networks

With VNET

Goal of this project

Low Level Traffic Monitoring

?

An online topology inference framework for a VM environment

Approach

Design an offline framework

Evaluate with parallel benchmarks

If successful, design an online framework for VMs

An offline topology inference framework

Goal:

A test-bed for traffic monitoring and evaluating topology inference

methods

The offline method

Synced Parallel Traffic Monitoring

Traffic Filtering and Matrix Generation

Matrix Analysis and Topology Characterization

The offline method

Synced Parallel Traffic Monitoring

Traffic Filtering and Matrix Generation

Matrix Analysis and Topology Characterization

  h1 h2 h3 h4

h1   7.7 7.6 7.8

h2 13.1   6.6 6.5

h3 13.5 6.4   6.6

h4 13.2 6.5 6.5  

*numbers indicate MB of data transferred.

The offline method

Synced Parallel Traffic Monitoring

Traffic Filtering and Matrix Generation

Matrix Analysis and Topology Characterization

Parallel Benchmarks Evaluation

Goal:

To test the practicality of low level traffic based inference

Parallel Benchmarks used

• Synthetic benchmarks: Patterns– N-dimensional mesh-neighbor– N-dimensional toroid-neighbor– N-dimensional hypercubes– Tree reduction – All-to-All

• Scheduling mechanism to generate deadlock free and efficient schemes

1 2 3

Application benchmarks

• NAS PVM benchmarks– Popular benchmarks for parallel computing– 5 benchmarks

• PVM-POV : Distributed Ray Tracing

• Many others…

Patterns application

PVM NAS benchmarks

Parallel Integer Sort

  h1 h2 h3 h4 h5 h6 h7 h8

h1   19.0 19.6 19.2 19.6 18.8 13.7 19.3

h2 22.6   10.7 10.8 10.7 10.9 9.7 10.5

h3 22.2 8.78   11.2 10.4 10.1 10.5 10.5

h4 22.4 8.9 9.5   11.1 10.8 10.6 10.2

h5 22.3 10.0 9.51 9.72   11.7 10.9 11.9

h6 24.0 8.9 10.7 9.9 10.8   12.2 12.1

h7 23.2 10.0 9.7 9.5 10.3 10.2   12.0

h8 24.9 11.2 11.0 11.8 11.5 11.2 10.7  *numbers indicate MB of data transferred.

An Online Topology Inference Framework

Goal:

To automatically detect, monitor and report the global traffic matrix for a set of VMs running on a overlay network

Overall Design

• Extend VNET to include the required features– Allows a set of VMs to be on same Layer 2

domain– Monitoring at ethernet packet level

• Challenge– Lacks manual control– Detecting interesting parallel program

communication ?

Detecting interesting phenomenon

Reactive Mechanisms Proactive Mechanisms

•Certain address properties

•Based on Traffic rate

•Etc.

Provide support for queries by external agent

Rate based monitoring

Non-uniform discrete event sampling

What is the Traffic Matrix for the last n seconds ?

Traffic Analyzer

Rate based Change detection

Traffic MatrixQuery Agent

VM Network Scheduling Agent

VNET daemon

VM

VNET overlay network

To other VNET daemons

Physical Host

Traffic Matrix Aggregation

• Each VNET daemon keeps track of local traffic matrix– Need to aggregate this information for a global view– When the rate falls, the local daemons push the traffic

matrix

The proxy daemon

Evaluation

• Used 4 Virtual Machines over VNET

• NAS IS benchmark

Conclusions

Possible to infer the topology with

low level traffic monitoring

A Traffic Inference Framework for Virtual Machines

Ready to move on to future steps