Return-Path: X-Original-To: apmail-hbase-dev-archive@www.apache.org Delivered-To: apmail-hbase-dev-archive@www.apache.org Received: from mail.apache.org (hermes.apache.org [140.211.11.3]) by minotaur.apache.org (Postfix) with SMTP id 3DC717F24 for ; Wed, 26 Oct 2011 22:05:15 +0000 (UTC) Received: (qmail 88421 invoked by uid 500); 26 Oct 2011 22:05:14 -0000 Delivered-To: apmail-hbase-dev-archive@hbase.apache.org Received: (qmail 88353 invoked by uid 500); 26 Oct 2011 22:05:14 -0000 Mailing-List: contact dev-help@hbase.apache.org; run by ezmlm Precedence: bulk List-Help: List-Unsubscribe: List-Post: List-Id: Reply-To: dev@hbase.apache.org Delivered-To: mailing list dev@hbase.apache.org Received: (qmail 88345 invoked by uid 99); 26 Oct 2011 22:05:14 -0000 Received: from nike.apache.org (HELO nike.apache.org) (192.87.106.230) by apache.org (qpsmtpd/0.29) with ESMTP; Wed, 26 Oct 2011 22:05:14 +0000 X-ASF-Spam-Status: No, hits=2.2 required=5.0 tests=HTML_MESSAGE,RCVD_IN_DNSWL_LOW,SPF_NEUTRAL X-Spam-Check-By: apache.org Received-SPF: neutral (nike.apache.org: local policy) Received: from [74.125.82.169] (HELO mail-wy0-f169.google.com) (74.125.82.169) by apache.org (qpsmtpd/0.29) with ESMTP; Wed, 26 Oct 2011 22:05:08 +0000 Received: by wyg34 with SMTP id 34so2893248wyg.14 for ; Wed, 26 Oct 2011 15:04:47 -0700 (PDT) Received: by 10.216.47.11 with SMTP id s11mr3161941web.24.1319666687147; Wed, 26 Oct 2011 15:04:47 -0700 (PDT) MIME-Version: 1.0 Received: by 10.216.8.208 with HTTP; Wed, 26 Oct 2011 15:04:26 -0700 (PDT) In-Reply-To: References: From: Ted Dunning Date: Wed, 26 Oct 2011 15:04:26 -0700 Message-ID: Subject: Re: Random I/O performance To: dev@hbase.apache.org Content-Type: multipart/alternative; boundary=001485f1bdc41bfe3104b03ad801 X-Virus-Checked: Checked by ClamAV on apache.org --001485f1bdc41bfe3104b03ad801 Content-Type: text/plain; charset=ISO-8859-1 But you can change the key definition to get semi-sorted behavior. This would give you some parallelism while still grouping recent keys into the same pages. Doing an on-the-fly key transition is pretty hairy, but doable. Accessing older data just requires knowing the cutoff date and doing one encoding or the other. On Wed, Oct 26, 2011 at 2:50 PM, Vladimir Rodionov wrote: > >>So a different schema would get cache into the mix? > > You can/t change schema while system is in production > --001485f1bdc41bfe3104b03ad801--