[
https://issues.apache.org/jira/browse/PHOENIX-4666?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16437881#comment-16437881
]
Maryann Xue commented on PHOENIX-4666:
--------------------------------------
Thank you for explaining 2! And now it's clear to me. It is actually an
optimization implemented in PHOENIX-852. It doesn't not apply to all join
queries, so I guess in the test case you have there this optimization is
triggered. So I'm thinking two options here:
1) Call {{HashCacheClient#evaluateKeyExpression()}} to get the key ranges if
cache is already available on the server side, in which case
{{CachedSubqueryResultIterator}} would still be needed but we do not add cache
one more time. We can have a client-side cache for such key-range values as
well. And if this is the first client building the cache for the first time, we
get these values from calling {{addHashCache()}} and then cache them on the
client side.
2) A more radical but easier approach is to disable this "child-parent (FK-PK)
join optimization" when using persistent cache. This makes some practical
sense: if we can make a big performance gain from avoiding rebuilding the hash
cache, it could indicate that the cache itself might be of some considerable
side, and thus the key ranges generated from a relatively large amount of
values might not be that useful to narrow down the scan anyway.
For now, I actually prefer the second approach, since we can focus on the main
part of this issue and move forward faster.
For 5: Just for simplicity. Feel like we can have less getters and "get" calls
here.
> Add a subquery cache that persists beyond the life of a query
> -------------------------------------------------------------
>
> Key: PHOENIX-4666
> URL: https://issues.apache.org/jira/browse/PHOENIX-4666
> Project: Phoenix
> Issue Type: Improvement
> Reporter: Marcell Ortutay
> Assignee: Marcell Ortutay
> Priority: Major
>
> The user list thread for additional context is here:
> [https://lists.apache.org/thread.html/e62a6f5d79bdf7cd238ea79aed8886816d21224d12b0f1fe9b6bb075@%3Cuser.phoenix.apache.org%3E]
> ----
> A Phoenix query may contain expensive subqueries, and moreover those
> expensive subqueries may be used across multiple different queries. While
> whole result caching is possible at the application level, it is not possible
> to cache subresults in the application. This can cause bad performance for
> queries in which the subquery is the most expensive part of the query, and
> the application is powerless to do anything at the query level. It would be
> good if Phoenix provided a way to cache subquery results, as it would provide
> a significant performance gain.
> An illustrative example:
> SELECT * FROM table1 JOIN (SELECT id_1 FROM large_table WHERE x = 10)
> expensive_result ON table1.id_1 = expensive_result.id_2 AND table1.id_1 =
> \{id}
> In this case, the subquery "expensive_result" is expensive to compute, but it
> doesn't change between queries. The rest of the query does because of the
> \{id} parameter. This means the application can't cache it, but it would be
> good if there was a way to cache expensive_result.
> Note that there is currently a coprocessor based "server cache", but the data
> in this "cache" is not persisted across queries. It is deleted after a TTL
> expires (30sec by default), or when the query completes.
> This is issue is fairly high priority for us at 23andMe and we'd be happy to
> provide a patch with some guidance from Phoenix maintainers. We are currently
> putting together a design document for a solution, and we'll post it to this
> Jira ticket for review in a few days.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
