Received: from malur.postgresql.org ([217.196.149.56]) by arkaria.postgresql.org with esmtps (TLS1.3:ECDHE_RSA_AES_256_GCM_SHA384:256) (Exim 4.92) (envelope-from ) id 1jMztH-0001u3-Pp for pgsql-sql@arkaria.postgresql.org; Fri, 10 Apr 2020 20:04:39 +0000 Received: from localhost ([127.0.0.1] helo=malur.postgresql.org) by malur.postgresql.org with esmtp (Exim 4.92) (envelope-from ) id 1jMztE-0006o1-60 for pgsql-sql@arkaria.postgresql.org; Fri, 10 Apr 2020 20:04:36 +0000 Received: from magus.postgresql.org ([2a02:c0:301:0:ffff::29]) by malur.postgresql.org with esmtps (TLS1.3:ECDHE_RSA_AES_256_GCM_SHA384:256) (Exim 4.92) (envelope-from ) id 1jMztD-0006ng-Vm for pgsql-sql@lists.postgresql.org; Fri, 10 Apr 2020 20:04:35 +0000 Received: from momjian.us ([72.94.173.45]) by magus.postgresql.org with esmtps (TLS1.3:ECDHE_RSA_AES_256_GCM_SHA384:256) (Exim 4.92) (envelope-from ) id 1jMzt6-000524-SV for pgsql-sql@lists.postgresql.org; Fri, 10 Apr 2020 20:04:35 +0000 Received: from bruce by momjian.us with local (Exim 4.92) (envelope-from ) id 1jMzsy-0007WP-Fn; Fri, 10 Apr 2020 16:04:20 -0400 Date: Fri, 10 Apr 2020 16:04:20 -0400 From: Bruce Momjian To: Lu Dillon Cc: Iuri Sampaio , Ed Behn , MichaelDBA , Steve Midgley , Erik Brandsberg , "pgsql-sql@lists.postgresql.org" Subject: Re: =?utf-8?B?5Zue5aSN?= =?utf-8?Q?=3A?= [External] Re: Separate volumes Message-ID: <20200410200420.GA24987@momjian.us> References: <5b60ae44-33e0-d9ea-d6e9-c6db9b4c7ef1@sqlexec.com> <399454FC-D175-421C-AC3A-31B1082E7881@gmail.com> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Disposition: inline Content-Transfer-Encoding: 8bit In-Reply-To: User-Agent: Mutt/1.10.1 (2018-07-13) List-Id: List-Help: List-Subscribe: List-Post: List-Owner: List-Archive: Precedence: bulk On Tue, Apr 7, 2020 at 04:31:51PM +0000, Lu Dillon wrote: > Hi All, > > This is a very intersting question. I believe this is not just a best practice > to PG. We can apply to all RDBMS. In my opinion, I agree with the others: with > SSD, you don’t separate tables and indexs to different disks. I think the IOPS > is enough. If you still have a problem of IOPS, you can try NVME device or U2 > device. If you are mixing magnetic and SSDs for the same database, having indexes on SSDs can really help, compared to table files on SSDs, where the benefit is more limited. Also, having current data on SSDs and archive data on magnetic is also useful, and you usually use time-based partitioning for such cases. -- Bruce Momjian https://momjian.us EnterpriseDB https://enterprisedb.com + As you are, so once was I. As I am, so you will be. + + Ancient Roman grave inscription +