You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@ignite.apache.org by "Andrey Mashenkov (Jira)" <ji...@apache.org> on 2021/04/15 15:02:00 UTC
[jira] [Created] (IGNITE-14557) Imrove row layout.
Andrey Mashenkov created IGNITE-14557:
-----------------------------------------
Summary: Imrove row layout.
Key: IGNITE-14557
URL: https://issues.apache.org/jira/browse/IGNITE-14557
Project: Ignite
Issue Type: Improvement
Reporter: Andrey Mashenkov
h3. Motivation.
When one try to read a column value from a row, the very first check will be a null-check.
As Null-Table resides right after an Offset-Table, the we need 2 jumps to for the null-check.
h3. Description.
Assuming, Null-Table reserves a bit for each columns even if the columns is not Nullable,
Null-table has constant size (within same version of schema) and we no need extra bytes to persist it's length into the tuple.
* Null-checks will not require extra read for Offset-Table size and extra jump.
* Offset-Table will not need extra read/jump to reach as Null-Table size is constant.
Let's just swap these tables in layout and fix docs README.md and IEP.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)