System requirements
This page documents the hardware and software requirements of the Caplin Platform.
Server hardware
Caplin publish benchmarks to aid capacity planning.
Liberator
For the full report, see Liberator 6.1 Benchmarks (PDF). The benchmark hardware and headline results are reproduced below.
Benchmarks on more modern hardware will be published following the release of Liberator 7.0. When choosing a server specification to host Liberator, CPU clock speed is more important than number of cores. |
Liberator is highly-configurable and can be tuned to suit different use cases. Refer to the full report for the configurations used in each benchmark.
Vendor |
DELL |
Model |
PowerEdge R620 |
Processors |
2 x Intel Xeon E5-2643 v1 3.3 GHz (quad-core) |
Memory |
32GB |
Operating System |
CentOS 6.3 (Final) |
Network Card |
Broadcom NetXtreme II BCM5720 |
Scenario | Users | Messages per second | Message latency |
---|---|---|---|
Low updates |
100,000 |
1 |
4ms |
Medium updates |
40,000 |
10 |
3ms |
High updates (without message batching)* |
12,000 |
50 |
6ms |
High updates (with message batching)* |
30,000 |
50 |
55ms |
Very high updates (without message batching)* |
4,000 |
100 |
9ms |
Very high updates (with message batching)* |
16,000 |
100 |
55ms |
-
When message batching is enabled, Liberator sends messages in batches rather than individually. This allows Liberator to handle many more clients, but at the expense of message latency. For an introduction to batching, see Liberator bursting (batching). For more information on the scenarios in the results above, see Liberator 6.1 Benchmarks (PDF).
Transformer
Transformer hosts modules of application logic. The most resource-intensive module of the standard suite of Transformer modules is Refiner, which provides real-time data transformation services.
Refiner module
For the full report, see Refiner 5.0 Benchmarks (PDF). The benchmark hardware and headline results are reproduced below.
The benchmarks for Refiner 5 apply equally to Refiner 6.
Vendor |
DELL |
Model |
PowerEdge R415 |
Processors |
2 x AMD Opteron 4180 2.6GHz (hexa-core) |
Memory |
16GB |
Operating System |
CentOS 5.5 (Kernel 2.6.18-194.el5 64bit) |
Benchmark headline-results:
-
Filtering and sorting on rapidly updating field values is not recommended with large underlying containers.
-
When filtering rapidly updating fields, performance deteriorates rapidly when Caplin Refiner has to process more than five million updates per second. The update rate is given by multiplying the number of unique filter requests by the number of underlying container objects that update per second.
-
When Caplin Refiner provides historic trade blotter data, two thousand users with one thousand blotter entries each produces two million unique objects, and requires about 16GB of random access memory. Although update latency remains low at these levels, available memory is the limiting factor.
-
For unique filter request rates of 8.3 per second, average response times on containers approaching the upper limit of 65,555 constituent objects remained below 100 milliseconds.
Operating system
Operating systems supported in production and development environments.
Production environment
Caplin Platform Version | Operating System |
---|---|
6.2 |
RHEL 6 (64-bit) RHEL 5 (64-bit) |
6.1 |
RHEL 5 (64-bit) |
Development environment
Caplin Platform Version | Operating System |
---|---|
6.2 |
RHEL 6 (64-bit) RHEL 5 (64-bit) Windows 7 (32-bit, 64-bit) Mac OS X 10.11 Mac OS X 10.10 |
6.1 |
RHEL 5 (64-bit) Windows 7 (32-bit, 64-bit) Mac OS X 10.11 Mac OS X 10.10 |
Java virtual machine
Current releases of Caplin products that require a Java virtual machine (JVM) are compiled on Oracle Java 7 and tested on Oracle Java 8.
For information on the JVM versions supported by a previous release of a Caplin product, see the release notes for that release.
J2EE web container
Database server
The Persistence Service, a module of the Caplin Platform component Transformer, requires a JDBC-compatible database server that supports text-based authentication.