EXPLAIN SSDCACHE option shows HorizonDB local SSD cache hits and misses per query
HorizonDB separates compute from remote storage and uses a local SSD cache to lower read latency. Previously, the author inferred cache behavior from I/O timings, but the new SSDCACHE option in EXPLAIN now reports it directly in the execution plan. In a 56 GB sequential scan example, nearly all page reads were served by the SSD cache, with just one miss going to lower storage. The author notes that an SSD-cache hit still counts as a PostgreSQL shared read, so SSDCACHE complements BUFFERS by showing where those reads were satisfied. The article also describes a three-tier model of shared buffers, the SSD cache, and cold storage, which maps to OLTP workloads with different latency needs.
This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)
Log in to join the discussion and vote.
Log in