Title: Inferring the Topology and Traffic Load of Parallel Programs in a VM environment
1Inferring the Topology and Traffic Load of
Parallel Programs in a VM environment
- Ashish Gupta
- Peter Dinda
- Department of Computer Science
- Northwestern University
2Overview
- Motivation behind parallel programs in a VM
environment - Goal To infer the communication behavior
- Offline implementation
- Evaluating with parallel benchmarks
- Online Monitoring in a VM environment
- Conclusions
3Virtuoso A VM based abstraction for a Grid
environment
4(No Transcript)
5Motivation
- A distributed computing environment based on
Virtual Machines - Raw machines connected to users network
- Our Focus Middleware support to hide the Grid
complexity
6(No Transcript)
7(No Transcript)
8Motivation
- A distributed computing environment based on
Virtual Machines - Raw machines connected to users network
- Our Focus Middleware support to hide the Grid
complexity - Our goal here Efficient execution of Parallel
applications in such an environment
9Parallel Application Behavior
Intelligent Placement and virtual networking of
parallel applications
Virtual Networks With VNET
VM Encapsulation
10VNET
- Abstraction A set of VMs on same Layer 2
network - Virtual Ethernet LAN
11Goal of this project
- Through low level packet traffic monitoring and
analysis - Inferring communication properties of parallel
applications - Topology
- Bandwidth requirements
- Other ?
12Goal of this project
?
Low Level Traffic Monitoring
13Approach
Design an offline framework
Evaluate with parallel benchmarks
If successful, design an online framework for VMs
14An offline topology inference framework
- Goal
- A test-bed for traffic monitoring and evaluating
topology inference methods
15The offline method
Synced Parallel Traffic Monitoring
Traffic Filtering and Matrix Generation
Matrix Analysis and Topology Characterization
16The offline method
Synced Parallel Traffic Monitoring
Traffic Filtering and Matrix Generation
Matrix Analysis and Topology Characterization
17The offline method
Synced Parallel Traffic Monitoring
Traffic Filtering and Matrix Generation
Matrix Analysis and Topology Characterization
18The offline method
Synced Parallel Traffic Monitoring
Traffic Filtering and Matrix Generation
Matrix Analysis and Topology Characterization
19The offline method
Synced Parallel Traffic Monitoring
Traffic Filtering and Matrix Generation
Matrix Analysis and Topology Characterization
PVMPOV Inference
20 Synced Parallel Traffic Monitoring
Traffic Filtering and Matrix Generation
Matrix Analysis and Topology Characterization
Infer.pl
21Parallel Benchmarks Evaluation
- Goal
- To test the practicality of low level traffic
based inference
22Parallel 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
23Application benchmarks
- NAS PVM benchmarks
- Popular benchmarks for parallel computing
- 5 benchmarks
- PVM-POV Distributed Ray Tracing
- Many others possible
- The inference not PVM specific
- Applicable to all communication .
- e.g. MPI, even non-parallel apps
24Patterns application
3-D Toroid
3-D Hypercube
2-D Mesh
Reduction Tree
All-to-All
25PVM NAS benchmarks
Parallel Integer Sort
26Traffic Matrix for PVM IS benchmark
27Traffic Matrix for PVM IS benchmark
Placement of host1 is crucial on the network
28An Online Topology Inference Framework VTTIF
- Goal
- To automatically detect, monitor and report the
global traffic matrix for a set of VMs running on
a overlay network
29Overall Design
- VNET
- Abstraction A set of VMs on same Layer 2
network - Virtual Ethernet LAN
30A VNET virtual layer
VNET Layer
Physical Layer
31Overall Design
- VNET
- Abstraction A set of VMs on same Layer 2
network - Extend VNET to include the required features
- Monitoring at Ethernet packet level
- The Challenge here
- Lacks manual control
- Detecting interesting parallel program
communication ?
32Detecting interesting phenomenon
Reactive Mechanisms
Proactive Mechanisms
Like a Burglar Alarm
Video Surveillance
33Physical Host
VM
VNET daemon
VNET overlay network
Traffic Analyzer
Rate based Change detection
Traffic Matrix Query Agent
To other VNET daemons
VM Network Scheduling Agent
34Traffic 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 (When do you push the traffic
matrix ?) - Operation is associative reduction trees for
scalability
The proxy daemon
35Evaluation
- Used 4 Virtual Machines over VNET
- NAS IS benchmark
36Conclusions
Possible to infer the topology with low level
traffic monitoring
37Current Work
- Capabilities for dynamic adaptation into VNET
- Spatial Inference ? Network Adaptation for
Improved Performance - Prelim Results Improved performance upto 40 in
execution time - Looking into benefits of Dynamic Adaptation
38For more information
- http//virtuoso.cs.northwestern.edu
- VNET is available for download
- PLAB web site
- plab.cs.northwestern.edu