You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@ignite.apache.org by "Pavel Tupitsyn (Jira)" <ji...@apache.org> on 2022/06/15 18:02:00 UTC
[jira] [Updated] (IGNITE-17179) Thin 3.0: SQL metadata caching
[ https://issues.apache.org/jira/browse/IGNITE-17179?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Pavel Tupitsyn updated IGNITE-17179:
------------------------------------
Description:
We send query metadata to the client with every request (IGNITE-17052). It is a waste of resources when the same query is executed frequently.
* Cache metadata on server, generate unique ID
* Return the ID to the client
* Client retrieves cached metadata from server when not known, caches it locally
However, some queries can be one-off and unique. The cache can grow too big over the lifetime of the application.
* Use the cache only for queries that have executed N times (detect this on the server)
* Limit cache size, use LRU eviction (client and server)
Do not use query text as cache key - column types can change. Use the entire metadata content to check cache entry equality.
was:
We send query metadata to the client with every request (IGNITE-17052). It is a waste of resources when the same query is executed frequently.
* Cache metadata on server, generate unique ID
* Return the ID to the client
* Client retrieves cached metadata from server when not known, caches it locally
> Thin 3.0: SQL metadata caching
> ------------------------------
>
> Key: IGNITE-17179
> URL: https://issues.apache.org/jira/browse/IGNITE-17179
> Project: Ignite
> Issue Type: Improvement
> Components: thin client
> Affects Versions: 3.0.0-alpha5
> Reporter: Pavel Tupitsyn
> Assignee: Pavel Tupitsyn
> Priority: Minor
> Labels: ignite-3
> Fix For: 3.0.0-alpha6
>
>
> We send query metadata to the client with every request (IGNITE-17052). It is a waste of resources when the same query is executed frequently.
> * Cache metadata on server, generate unique ID
> * Return the ID to the client
> * Client retrieves cached metadata from server when not known, caches it locally
> However, some queries can be one-off and unique. The cache can grow too big over the lifetime of the application.
> * Use the cache only for queries that have executed N times (detect this on the server)
> * Limit cache size, use LRU eviction (client and server)
> Do not use query text as cache key - column types can change. Use the entire metadata content to check cache entry equality.
--
This message was sent by Atlassian Jira
(v8.20.7#820007)