You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@ignite.apache.org by "Vladimir Ozerov (JIRA)" <ji...@apache.org> on 2018/11/16 18:08:00 UTC
[jira] [Created] (IGNITE-10311) SQL: Create system view with query
co-location plans
Vladimir Ozerov created IGNITE-10311:
----------------------------------------
Summary: SQL: Create system view with query co-location plans
Key: IGNITE-10311
URL: https://issues.apache.org/jira/browse/IGNITE-10311
Project: Ignite
Issue Type: Task
Components: sql
Reporter: Vladimir Ozerov
Fix For: 2.8
It is very important to understand model of distributed execution for certain queries for performance tuning. Let's create a syhstem view which will iterate over cached two-step queries and print their plans in table form.
Approximate structure (very raw):
{code}
CREATE VIEW sql_query_plan (
plan_id, // Unique plan ID (node ID + unique local ID)
sql, // Plain text
sql_hash. // May be more convenient that query_id
flags // Same query with different flags may result in different plans
)
CREATE VIEW sql_query_plan_fragments (
plan_id, // Same plan ID
order, // Together with plan_id it forms PK
action, // What is this - "reduce", "skip reduce", "map", etc. (may be we will need multiple columns)
sql, // SQL of the fragment (if applicable)
)
{code}
Next user may list cached plans, select interesting, and query plan fragments view.
We need to analyze other vendors to better understand what to show in views.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)