[ 
https://issues.apache.org/jira/browse/HBASE-30022?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Vladimir Rodionov updated HBASE-30022:
--------------------------------------
    Description: 
Description

Refactor HFileReaderImpl to use CacheAccessService as the entry point for block 
cache access instead of directly interacting with BlockCache.

This change updates the read path to delegate cache operations through 
CacheAccessService, including:
        •       block lookup on read
        •       cache population on read miss
        •       cache-related metrics updates (if applicable)

Scope
        •       Replace direct BlockCache calls in HFileReaderImpl with 
CacheAccessService
        •       Adapt existing call parameters to use request/write context 
objects
        •       Ensure compatibility with current caching behavior

Notes
        •       No behavior change intended
        •       Existing cache implementations (LruBlockCache, BucketCache, 
CombinedBlockCache) remain unchanged
        •       CacheAccessService implementation continues to delegate to 
BlockCache in this phase
        •       This is part of migrating read path code toward a pluggable 
cache architecture

  was:
Refactor existing CombinedBlockCache into a topology abstraction:
        •       TieredExclusiveTopology
        •       TieredInclusiveTopology

Move responsibilities:
        •       L1/L2 routing
        •       promotion/demotion logic
        •       metric aggregation

Ensure behavior remains identical to current implementation.


> Refactor HFileReaderImpl to use CacheAccessService for block cache access
> -------------------------------------------------------------------------
>
>                 Key: HBASE-30022
>                 URL: https://issues.apache.org/jira/browse/HBASE-30022
>             Project: HBase
>          Issue Type: New Feature
>          Components: BlockCache, Performance
>            Reporter: Vladimir Rodionov
>            Assignee: Vladimir Rodionov
>            Priority: Major
>
> Description
> Refactor HFileReaderImpl to use CacheAccessService as the entry point for 
> block cache access instead of directly interacting with BlockCache.
> This change updates the read path to delegate cache operations through 
> CacheAccessService, including:
>       •       block lookup on read
>       •       cache population on read miss
>       •       cache-related metrics updates (if applicable)
> Scope
>       •       Replace direct BlockCache calls in HFileReaderImpl with 
> CacheAccessService
>       •       Adapt existing call parameters to use request/write context 
> objects
>       •       Ensure compatibility with current caching behavior
> Notes
>       •       No behavior change intended
>       •       Existing cache implementations (LruBlockCache, BucketCache, 
> CombinedBlockCache) remain unchanged
>       •       CacheAccessService implementation continues to delegate to 
> BlockCache in this phase
>       •       This is part of migrating read path code toward a pluggable 
> cache architecture



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

Reply via email to