hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Matteo Bertozzi (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-13260) Bootstrap Tables for fun and profit
Date Fri, 17 Apr 2015 08:33:59 GMT

    [ https://issues.apache.org/jira/browse/HBASE-13260?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14499474#comment-14499474
] 

Matteo Bertozzi commented on HBASE-13260:
-----------------------------------------

{quote}For assignment, the store might become the bottleneck, but in case of fast procs, there
will not be flushes for most of the time (since procs will get deleted from memstore). So
WAL is the bottleneck. Between the WAL format in WALProcStore and the FSHLog, I did not check
which one has more overhead vs write speed. But both will have to go thru wal IO.{quote}
aside compactions, I don't think the perf problem me and stack raised is around adding procs
to the WAL, but it is on replay. With the current proc wal, we can throw the wal away at any
point if we know that all the procs are deleted or all of them are in the latest wal. with
the region wal as far as I know we have to keep all the wals around until we have a flush.
to with the region the wal replay will be much more heavier. but maybe we can do some tricks
there too. e.g. if the memstore is empty after deletes roll the wal and archive the others

> Bootstrap Tables for fun and profit 
> ------------------------------------
>
>                 Key: HBASE-13260
>                 URL: https://issues.apache.org/jira/browse/HBASE-13260
>             Project: HBase
>          Issue Type: Bug
>            Reporter: Enis Soztutar
>            Assignee: Enis Soztutar
>             Fix For: 2.0.0, 1.1.0
>
>         Attachments: hbase-13260_prototype.patch
>
>
> Over at the ProcV2 discussions(HBASE-12439) and elsewhere I was mentioning an idea where
we may want to use regular old regions to store/persist some data needed for HBase master
to operate. 
> We regularly use system tables for storing system data. acl, meta, namespace, quota are
some examples. We also store the table state in meta now. Some data is persisted in zk only
(replication peers and replication state, etc). We are moving away from zk as a permanent
storage. As any self-respecting database does, we should store almost all of our data in HBase
itself. 
> However, we have an "availability" dependency between different kinds of data. For example
all system tables need meta to be assigned first. All master operations need ns table to be
assigned, etc. 
> For at least two types of data, (1) procedure v2 states, (2) RS groups in HBASE-6721
we cannot depend on meta being assigned since "assignment" itself will depend on accessing
this data. The solution in (1) is to implement a custom WAL format, and custom recover lease
and WAL recovery. The solution in (2) is to have the table to store this data, but also cache
it in zk for bootrapping initial assignments. 
> For solving both of the above (and possible future use cases if any), I propose we add
a "boostrap table" concept, which is: 
>  - A set of predefined tables hosted in a separate dir in HDFS. 
>  - A table is only 1 region, not splittable 
>  - Not assigned through regular assignment 
>  - Hosted only on 1 server (typically master)
>  - Has a dedicated WAL. 
>  - A service does WAL recovery + fencing for these tables. 
> This has the benefit of using a region to keep the data, but frees us to re-implement
caching and we can use the same WAL / Memstore / Recovery mechanisms that are battle-tested.

>  



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

Mime
View raw message