+1
On 2023/12/07 22:37:55 Mike Carey wrote: > +1 > > We really need this! > > On 12/7/23 2:21 PM, Glenn Justo Galvizo wrote: > > Every time a query is issued to AsterixDB, the query must undergo > > compilation. If the same query is run repeatedly, this query must be > > recompiled each and every time. A query plan cache can help AsterixDB > > achieve a lower floor on the end-to-end time by storing the job > > specifications for previously compiled queries, ultimately skipping the AST > > rewriting and Algebricks compilation of a previously executed query. > > > > (APE copied from contributor Sushrut Borkor) > > > > This APE is about adding a query plan cache to AsterixDB. More > > specifically, this query plan cache acts as a hash table that skips 1) the > > AST rewriting, 2) the entire Algebricks plan translation to Algebricks > > optimization, and 3) the Hyracks job generation. The keys of this hash > > table are: > > • AST String. We cache this instead of the original query string > > before parsing because it is resilient to minor changes in the query, such > > as adding spaces or empty lines. > > • SessionConfig. For example, if the user runs a query, changes part > > of the session configuration (e.g. the preferred output format), and reruns > > the query, this prevents the second query from being served from the cache. > > • Config, to capture the effects of used SET statements. > > • Active Dataverse, e.g., as defined in a USE statement. > > • Result Set ID, which distinguishes among queries in multi-statement > > requests. > > > > While the values of each hash table entry are: > > • Hyracks Job Spec to be submitted to Hyracks. > > • Cached warnings. Since we skip compilation when serving queries from > > the cache, we cannot detect compile time warnings. To get around this, we > > cache warnings issued during rewriting and compilation, and then reissue > > them for cache hits. As a result, line numbers in warnings may be incorrect > > for queries answered using the cache. > > • Lock. Since running the same job from multiple threads does not > > work, we include a lock in the cache value. To use a cached job spec, a > > thread must acquire this lock, and then release it after the job has > > finished running. If the lock is held by another thread, we recompile the > > query instead of blocking. > > > > The proposed changes are the following: > > > > Interface: > > We introduce two new statements for controlling cache access: > > • “SET `compiler.querycache.bypass` "true";” forces the current query > > to ignore the cache. > > • “SET `compiler.querycache.clear` "true";” clears all cache entries. > > The current query may still insert into the cache. > > We also add a boolean HTTP API parameter bypass_cache which does the same > > thing as the first SET statement above. Finally, the parameter > > query.cache.capacity can be configured in the [cc] section of the cc.conf > > file to control the maximum cache size before replacement. > > > > Changes: > > • Compilation logic is changed in the source code since we skip > > rewriting and compilation for cache hits. > > • Hints are now included in the AST string to prevent incorrect cache > > lookups that would otherwise miss the hints. > > • A bug is fixed where the AST string of WINDOW expressions did not > > include FROM LAST or IGNORE NULLS. > > > > Seehttps://issues.apache.org/jira/projects/ASTERIXDB/issues/ASTERIXDB-3183 > > for the JIRA issue, as well > > ashttps://cwiki.apache.org/confluence/display/ASTERIXDB/APE+2%3A+Query+Plan+Cache > > for more details. > > > > Please vote on this APE. We will keep this open for 72 hours and pass with > > either 3 votes or a majority of positive votes.