Skip to content

Caching

LAPIS uses Spring Boot’s caching capabilities to cache aggregated, insertion and mutation queries. We decided to cache SILO actions that are relatively expensive to compute, but don’t return much data (measured in size of the response body). Details and FASTA actions are expensive, but return potentially large amounts of data and are therefore not cached.

Internally, we use soft references to allow the JVM garbage collector to remove entries from the cache if the memory is needed. However, per default the cache is not guaranteed to have a fixed memory size, increasing with each stored entry. The maintainer must ensure that enough memory is available to store the cache entries, or provide a limit to the heap size of the JVM (eg. -Xmx4g, or -XX:+UseContainerSupport when running LAPIS in docker )

You can turn off caching by providing the command line argument --spring.cache.type=none to the execution.

LAPIS offers an endpoint to check the current cache size (in number of cached entries) at /actuator/metrics/cache.size:

https://s1.int.genspectrum.org/open/actuator/metrics/cache.size