Looking at the log of vrs-cache genegraph running against a variation-normalizer service running with 5 replicas in our cluster, it looks like vrs-cache is spending ~5.8ms per variant. This includes both the cluster-scoped http request to do the normalization, and the localhost http request to put it in the rocksdb cache.
Looking at the log of
vrs-cachegenegraph running against avariation-normalizerservice running with 5 replicas in our cluster, it looks like vrs-cache is spending ~5.8ms per variant. This includes both the cluster-scoped http request to do the normalization, and the localhost http request to put it in the rocksdb cache.