- changed status to resolved
Enhancement of SemMedDb query performance.
Issue #4
resolved
Queries of SemMedDb are somewhat sluggish for some concepts. How can this be improved?
-
MySQL has a number of performance enhancement measures that can be taken. These need to be investigated and some experimental benchmarking is in order.
-
Use of other database types (e.g. Neo4j) might be worth considering given the problem domain (of semantic networks of data).
Comments (2)
-
reporter -
- removed milestone
Removing milestone: SemMedDb MySQL database connected to Mark2Cure KB (automated comment)
- Log in to comment
Use of alternate databases (i.e. Neo4j) reviewed and rejected. MySQL query performance enhanced by optimized of a large memory server with large page caching of tables and large query buffer. Running GBK on a smaller less costly server might be desirable, but we know that, for now, sufficient performance is achieved by using a large enough machine. In the future, load balancing (read only) queries and perhaps, other parallel strategies, might allow for additional improvements.