[
https://issues.apache.org/jira/browse/GRIFFIN-194?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16642315#comment-16642315
]
Nikolay Sokolov commented on GRIFFIN-194:
-----------------------------------------
With 216 databases (some have lots of tables, some have very few) time to load
all table lists is 2.5-3 seconds and payload size is 800 Kb. It feels like
network roundtrip is major contributor of that, I'll try to implement listing
for all databases and see whether there will be any improvement.
> [service] Hive API improvement
> ------------------------------
>
> Key: GRIFFIN-194
> URL: https://issues.apache.org/jira/browse/GRIFFIN-194
> Project: Griffin (Incubating)
> Issue Type: Sub-task
> Reporter: Nikolay Sokolov
> Priority: Minor
>
> Purpose is mainly to support GRIFFIN-195 with single request to get table
> list information, while avoiding transferring all table metadata and making
> lots of metastore requests.
> Hive API provides following relevant APIs right now:
> * listing DBs
> * getting all table names in DB
> * listing all table _objects_ in all _dbs_
> What's seems to be missing, is API call for all table names in all DBs (as
> middle ground between n+1 API requests and 1 API request with huge payload
> and n*m+1 metastore requests on backend).
> This api request should take no parameters, and return Map<String,
> List<String>> in response.
> Proposed API endpoint: TBD
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
