MRNet: Multicast/Reduction Network
Efficient Communication for Large Scale Systems
MRNet API
MRNet Improves Tool Performance by Distributing Data Processing and Analyses
User Interface
Analysis and Control
Analysis and Control
MRNet
Communication Processes
 Overlay network for efficient data
multicasts and reductions.
User Interface
Tool
Front-End
MRNet Overview
 Tree topology for distribution of
data analysis and management.
Daemon0
Daemon1
Daemon2
Daemon3
Process0
Process1
Process2
Process3
 Efficient multicasts for scalable
control.
 Concurrent data channels with
different aggregation schemes.
 Open-source licensing.
Daemon1
Daemon2
Daemon3
Processn-1
Processn
Process0
Process1
Process2
Process3
…
Daemonn-1
Daemonn
Processn-1
Processn
 Streams are virtual channels that bind
the end-points in a communicator for
group communication.
 Streams are created with filters that
perform specified data aggregations on
the data flow.
 Tool-specific filters can be loaded into
MRNet for custom aggregations.
MRNet-enabled Distributed Tool
Sample Performance
Improving the Performance of Paradyn’s Data Analysis
Service Rate/Arrival Rate
 High-throughput data transfers.
Daemon0
 End-points, retrieved from instantiated
networks, are grouped into
communicators.
MRNet Filters
 Flexible process-tree topologies.
 Built-in filters for elementary
aggregations.
Daemonn
Un-enhanced Distributed Tool
MRNet Features
 Dynamic loading of user-defined
data aggregations.
…
Daemonn-1
Tool
Back-Ends
 Customizable components for use
in diverse tool applications.
 C++ based API for integration into tool
front-end and daemons.
Microbenchmark Results
• ASCI Blue Pacific:
•256x4 PowerPC
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
No performance
degradation
with MRNet!
• 512 back-ends
•~64 byte packet sizes
0
100
200
300
400
500
Number of Back-Ends
32 metrics, flat tree
16 metrics, flat tree
8 metrics, flat tree
1 metric, flat tree
32 metrics, 8-way fanout
600
• 8-way vs. Flat Tree
•Startup Latency:
•13.9 Speedup
•Round-Trip Latency:
•13.4 Speedup
• Reduction Throughput
•8-way: 55.23 ops/sec.
•Flat tree: 4.03 ops/sec
(256 back-ends).
 Native MRNet Filters
 MIN, MAX, SUM, AVG
 Concatenation
 Custom Paradyn Filters
 Clock synchronization.
 Equivalence classes for
application data.
 Time-based data synchronization.
Status
 Release v1.1:
 November 2004
 Tested Platforms:
 Linux, Aix, Solaris
 Additional Information:
 www.paradyn.org/mrnet/