[ 
https://issues.apache.org/jira/browse/YARN-3448?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14485446#comment-14485446
 ] 

Zhijie Shen commented on YARN-3448:
-----------------------------------

bq.  In fact, all rolling dbs from now unto ttl may be active.

Yeah, actually this is the point I'd like make. For example, if ttl = 10h and 
rolling period = 1h, we will have 10 active rolling dbs. Though 2 - 10 dbs are 
not current, but they can't be deleted because they contain the data that is 
still alive. Only rolling dbs from 11 and son will be deleted. While ttl = 10h, 
we change rolling period = 10h, and I will only have 1 active 10 rolling db, 
and its size should be equivalent to prior 10 1h rolling dbs. Therefore, my 
point is that if rolling period smaller than ttl, we still need to keep all the 
data alive, it's not necessary to separate them into multiple dbs rather than 
keeping them together in the current db.

One benefit I can think of about multiple-rolling-db approach (as well as 
different dbs for different data type) is to increase concurrency. However, I 
didn't see we have multiple threads to write different dbs concurrently.

> Add Rolling Time To Lives Level DB Plugin Capabilities
> ------------------------------------------------------
>
>                 Key: YARN-3448
>                 URL: https://issues.apache.org/jira/browse/YARN-3448
>             Project: Hadoop YARN
>          Issue Type: Improvement
>            Reporter: Jonathan Eagles
>            Assignee: Jonathan Eagles
>         Attachments: YARN-3448.1.patch, YARN-3448.2.patch, YARN-3448.3.patch
>
>
> For large applications, the majority of the time in LeveldbTimelineStore is 
> spent deleting old entities record at a time. An exclusive write lock is held 
> during the entire deletion phase which in practice can be hours. If we are to 
> relax some of the consistency constraints, other performance enhancing 
> techniques can be employed to maximize the throughput and minimize locking 
> time.
> Split the 5 sections of the leveldb database (domain, owner, start time, 
> entity, index) into 5 separate databases. This allows each database to 
> maximize the read cache effectiveness based on the unique usage patterns of 
> each database. With 5 separate databases each lookup is much faster. This can 
> also help with I/O to have the entity and index databases on separate disks.
> Rolling DBs for entity and index DBs. 99.9% of the data are in these two 
> sections 4:1 ration (index to entity) at least for tez. We replace DB record 
> removal with file system removal if we create a rolling set of databases that 
> age out and can be efficiently removed. To do this we must place a constraint 
> to always place an entity's events into it's correct rolling db instance 
> based on start time. This allows us to stitching the data back together while 
> reading and artificial paging.
> Relax the synchronous writes constraints. If we are willing to accept losing 
> some records that we not flushed in the operating system during a crash, we 
> can use async writes that can be much faster.
> Prefer Sequential writes. sequential writes can be several times faster than 
> random writes. Spend some small effort arranging the writes in such a way 
> that will trend towards sequential write performance over random write 
> performance.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Reply via email to