Return-Path: X-Original-To: apmail-hbase-issues-archive@www.apache.org Delivered-To: apmail-hbase-issues-archive@www.apache.org Received: from mail.apache.org (hermes.apache.org [140.211.11.3]) by minotaur.apache.org (Postfix) with SMTP id 571F210076 for ; Tue, 27 May 2014 18:12:02 +0000 (UTC) Received: (qmail 35050 invoked by uid 500); 27 May 2014 18:12:02 -0000 Delivered-To: apmail-hbase-issues-archive@hbase.apache.org Received: (qmail 35012 invoked by uid 500); 27 May 2014 18:12:02 -0000 Mailing-List: contact issues-help@hbase.apache.org; run by ezmlm Precedence: bulk List-Help: List-Unsubscribe: List-Post: List-Id: Delivered-To: mailing list issues@hbase.apache.org Received: (qmail 35003 invoked by uid 99); 27 May 2014 18:12:02 -0000 Received: from arcas.apache.org (HELO arcas.apache.org) (140.211.11.28) by apache.org (qpsmtpd/0.29) with ESMTP; Tue, 27 May 2014 18:12:02 +0000 Date: Tue, 27 May 2014 18:12:02 +0000 (UTC) From: "stack (JIRA)" To: issues@hbase.apache.org Message-ID: In-Reply-To: References: Subject: [jira] [Commented] (HBASE-11238) Add info about SlabCache and BucketCache to Ref Guide MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 7bit X-JIRA-FingerPrint: 30527f35849b9dde25b450d4833f0394 [ https://issues.apache.org/jira/browse/HBASE-11238?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14010019#comment-14010019 ] stack commented on HBASE-11238: ------------------------------- I buy your argument that this better belongs in the refguide than in javadoc. The features depends so much on code reading, having the doc in javadoc makes it easier linking the code internals to the explanation. The doc is still under development and configuration of this feature is likely to change in near future too. If we add doc to refguide, we need to change the package-info so it points into refguide instead so we have doc in one place only. In following, it should be fetching from BucketCache or SlabCache will be slower than from native on heap LruBlockCache.... "Fetching + will always be slower when fetching from BucketCache, but latencies tend to be less + erratic over time, because there is less garbage collection." Also experience seems to have it that BucketCache makes less garbage than SlabCache so should be even less erratic (than SlabCache or LruBlockCache). No hard numbers, just reports by a few of the lads. Change '(CombinedBlockCache)' to '(DoubleBlockCache vs CombinedBlockCache)' You lost the link to Nick's 'BlockCache 101'? This is not right "SlabCache is enabled by default." (and you contradict yourself by writing later 'To enable it...") There are three block cache options. The original implemenation which is in the java heap called LruBlockCache. Then there are SlabCache and BucketCache, both of which are mainly intended for keeping blockcache data offheap (but they can also keep data onheap and in files, at least in the BucketCache case). We should say what hbase.offheapcache.percentage is a multiplicand for (the JVM setting). oh.... you say it later so there is some repeat going on here? Otherwise, looks great. > Add info about SlabCache and BucketCache to Ref Guide > ----------------------------------------------------- > > Key: HBASE-11238 > URL: https://issues.apache.org/jira/browse/HBASE-11238 > Project: HBase > Issue Type: Bug > Components: documentation > Affects Versions: 0.98.2 > Reporter: Misty Stanley-Jones > Assignee: Misty Stanley-Jones > Attachments: HBASE-11238.patch > > > Upstream issues: HBASE-11171 and HBASE-11098. Could back port some of what is in these issues, the package-info.java class for instance. -- This message was sent by Atlassian JIRA (v6.2#6252)