[
https://issues.apache.org/jira/browse/IGNITE-4164?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15632037#comment-15632037
]
Alexey Kuznetsov commented on IGNITE-4164:
------------------------------------------
I think it will be very hard to implement.
But {{CacheAbstractJdbcStore}} already can process custom SQL queries in
parallel.
You could do it from user code like this:
{code}
cache.loadCache(null,
"someKeyClass", "select * from SomeTable where id < 10000",
"someKeyClass", "select * from SomeTable where id >= 10000 and id < 20000",
.....
"someKeyClass", "select * from SomeTable where id >= XXXXX and id < YYYY")
{code}
All listed queries will be executed in parallel.
You may build such list of queries as array and pass it to
{{cache.loadCache()}} because
{{cache.loadCache()}} required second argument as {{Object ...args}}
If this work for you, please close this issue as {{Won't Fix}}.
> Add support for parallel loading of caches using custom SQL load queries in
> org.apache.ignite.cache.store.jdbc.CacheAbstractJdbcStore#loadCache
> -----------------------------------------------------------------------------------------------------------------------------------------------
>
> Key: IGNITE-4164
> URL: https://issues.apache.org/jira/browse/IGNITE-4164
> Project: Ignite
> Issue Type: Improvement
> Components: SQL
> Affects Versions: 1.7
> Reporter: Anghel Botos
>
> Please add support for parallel loading of caches using custom SQL load
> queries in
> org.apache.ignite.cache.store.jdbc.CacheAbstractJdbcStore#loadCache. For the
> moment this is not possible, as the current implementation performs the load
> for each entity type in the cache using a single
> {{LoadCacheCustomQueryWorker}}, as opposed to the the approach that is used
> when no custom SQL queries are provided for the load, where for each entity
> type, the load is distributed across several threads based on some ranges.
> While it may not be possible to support parralel load with any custom SQL
> query (as this would mean that Ignite would have to somehow understand the
> meaning of that custom query), it would still be a significant improvement if
> it would be possible to have the parallel load when providing a custom
> {{WHERE}} clause for each entity type (instead of a full custom query).
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
