Title page
Preface
Copyright and disclaimer
What this document contains
Who should read this document
Related documents
Feedback
Acknowledgments
Overview and headline results
About the benchmark tests
Headline figures
Caplin's benchmark tools
Test results for headline figures
Low update rates for large numbers of end users
Medium update rates for large numbers of end users
High update rates for large numbers of end users
Very high and extreme update rates
Test details
How to benchmark using Caplin's tools
Package contents
Architecture
Test DataSource application (Benchsrc)
Test RTTP client application (Benchrttp)
Benchmark requirements
Installation and Setup
Installing Benchtools
Benchsrc setup
Benchrttp setup
Liberator installation
Liberator setup
How Caplin's benchmark tests were conducted
Test method
Test configurations
Test software
Test hardware
The network
Results of additional tests
Effect of session threads on performance
Low backend data set
Medium backend dataset
Effect of DataSource threads on performance
High backend dataset
High backend dataset - adding session threads
Medium backend dataset
Effect of burst configuration (batching) on performance
Low backend dataset
Effect of message size on performance
Overall size of a message
Number of fields in a message
Effect of failover on performance
Latency during failover scenario
Liberator 4.4 versus 4.2
Low backend data set
Medium backend data set
Frequently asked questions
What burst configuration should we use?
How many threads should we configure?
How do message sizes affect performance?
How much bandwidth will our Liberator use?
How many subscriptions can Liberator handle?
How much disk space will our Liberator need?
Why are there no CPU usage measurements for some tests?
Glossary of terms and acronyms

© Caplin Systems Ltd. 2007