LITMUS is a comprehensive framework, that will serve the academicians, researchers, DMS developers and the organizations who employ DMSs for the efficient consumption of their data, by allowing a choke-point driven performance comparison and analysis of various Data Management Solutions (Graph and RDF-based), with respect to different third-party real and synthetic datasets and queries.
The following DMS systems are supported currently.
- Orient
- Sparksee
- Neo4J
- TinkerGraph
- Jena
- RDF3X
- Opensource Virtuoso
- 4store
- Execution Time
- Cycles : The number of cycles it takes to execute a task (loading a database for a DMS, or running a particular query for a DMS).
- Instructions : The number of instructions which are executed in the task.
- Cache references : The total number of cache references which are made during the task.
- Cache misses : The total number of cache misses that happen during the task.
- Bus cycles : The number of bus cycles happening during the execution of the task.
- L1 data cache loads : The total number of data cache loads for L1 cache that happen during the execution of the task.
- L1 data cache load misses : The total number of L1 data cache load misses that happen during the execution of the task.
- L1 data cache stores : The L1 data cache stores that happen during the execution of the task
- dTLB loads : The data translation lookaside buffer loads that happen during the execution of the task.
- dTLB load misses : The data Translation lookaside buffer loads misses that happen during the execution of the task.
- dTLB prefetch misses : The data Translation lookaside buffer prefetch misses that happen during the execution of the task.
- LLC loads : The Last level Cache loads that happen during the execution of the task.
- LLC load misses : The last level cache load misses that happen during the execution of the task.
- LLC stores : The last level cache stores that happen during the execution of the task.
- LLC Prefetches : The last level cache prefetches that happen during the execution of the task.
- Branches : The total number of branches that are encountered during the execution of the task.
- Branch misses : The total number of branches that are missed in the execution of the task.
- Context switches : The total context switches that happen when the task is executed.
- CPU migrations : The CPU migrations that happen when the task is executed.
- Page faults : The page faults that happen when the task is executed.
A docker container for the Litmus-Benchmark-Suite.
Install Docker Community Edition (CE). The instructions for the installation process can be found here.
The Instructions.md file has all the instructions to setup and run the Litmus Benchmark Suite.
docker build -t litmus:local /path/of/the/repo
docker run --privileged litmus:local
We are using the --privileged flag, because we need root access to clear the caches when we run the benchmarking process.