From owner-freebsd-isp Tue Mar 25 5: 0:28 2003 Delivered-To: freebsd-isp@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 0CC9B37B401 for ; Tue, 25 Mar 2003 05:00:25 -0800 (PST) Received: from web1.nexusinternetsolutions.net (web1.nexusinternetsolutions.net [206.47.131.12]) by mx1.FreeBSD.org (Postfix) with SMTP id 24A6343F93 for ; Tue, 25 Mar 2003 05:00:24 -0800 (PST) (envelope-from dave@hawk-systems.com) Received: (qmail 66025 invoked from network); 25 Mar 2003 13:00:21 -0000 Received: from unknown (HELO ws1) (24.157.103.51) by web1.nexusinternetsolutions.net with SMTP; 25 Mar 2003 13:00:21 -0000 From: "Dave [Hawk-Systems]" To: Subject: RE: Maximum recommended user limits on mail server Date: Tue, 25 Mar 2003 08:00:20 -0500 Message-ID: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit X-Priority: 3 (Normal) X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook IMO, Build 9.0.6604 (9.0.2911.0) In-Reply-To: <3E7FED30.3070709@mac.com> X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1106 Importance: Normal X-Spam-Status: No, hits=-12.2 required=5.0 tests=IN_REP_TO,MSGID_GOOD_EXCHANGE,QUOTED_EMAIL_TEXT autolearn=ham version=2.50 X-Spam-Level: X-Spam-Checker-Version: SpamAssassin 2.50 (1.173-2003-02-20-exp) Sender: owner-freebsd-isp@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.org >-----Original Message----- >From: owner-freebsd-isp@FreeBSD.ORG >[mailto:owner-freebsd-isp@FreeBSD.ORG]On Behalf Of Chuck Swiger >> I won't argue, however, that downtime on a properly configured Sun >> would be a fraction of a properly configured i386 box (I'm not too >> familiar with sun, but isn't there a model with hot-swap everything, >> including processor modules?). > >Most of Sun's lineup above the entry-class servers support hot-swap >everything, yes: things like an E4500. The 3-digit Sun boxes like the >E250, 280, 450, 480, etc are hot-swap disk only, not CPU or memory. It is sometimes easier to justify two intel based servers for redundancy at half or 2/3 the cost of 1 sun box... then have one of the intel servers down all day while you piddle with it... doesn't affect downtime for the sites/services hosted by the pair(or cluster) of servers. IMO this is even more relevant when you are talking sites/applications that require clusters. Sun gets pricey fast, and unless your accounting department still writes blank cheques... Would rather pay for decent cisco routing/switching equipment, and have a whack of cheaper redundant intel based servers. >> My current storage solution (FreeBSD 4-STABLE on a Celeron 600, 512MB, >> RAID5 [amr 466, 16MB, LVD, 40Mbit/s, SCA]) has seen less than 5 hours of >> downtime in the last 30 months. My SMTP/AV host has seen less than 2 >> hours of downtime in the last 18 months. Nearly all of that downtime >> was planned, and occurred in the wee hours of the morning. > >Sounds good to me. It's not that I believe that decent Intel hardware >is particularly unreliable, but faults tend to be more serious (if only >due to less redundancy and ECC thru various datapaths than in Sun's >hardware). OpenFirmware's a plus, too...not that I need to advocate OF >around people using a platform with FICL. There are a lot of arguments FOR sun architecture. Same arguments I use for Cisco vs other routing/switching/access products. In the end it comes down to preference. Like arguing that Macs are better than PCs, an argument that will not be settled since both camp is convinced they are correct and can provide the proof to back up their claims. >> Personally, I can't see needing a Sun for quite some time. I know that >> my current solution would handle at least 20k accounts without much >> issue at all. The only concern I currently have, is that the hardware >> is coming up on 3 years old and should probably be replaced sooner than >> later. > >That may be the biggest difference right there, although the presence of >SCSI adds significant longevity to the lifespan of an x86 server. >Still, getting replacement parts (ie, hard drives with identical >cylinder layouts to your original drives) after 3 or 5 years becomes a >concern that a Sun box wouldn't have. Again, you can afford to replace the Intel box 2, maybe 3 times if you want to, then re-deploy the other boxes to smaller sites/services. You will still come under budget compared to a sun based solution, without much if any change in performance from the end user's perspective. >That's one reason to pay the 3:1 or so markup for Sun-branded versus OEM >drives. Another is that you're getting drives that test well-- for >example, below-average spindle motor current required to maintain speed >(indicating a slightly better bearing). Same effect as distributing >resistors by quality into groups +/- 5%, +/- 10%, and +/- 20% tolerance. Again, easier to get 10 spare drives at $100ea approved vs 3 spare drives at $300 ea. Looks like more bang for the buch, and generally you get the same end results over time... again, IMO. Am beginning to wonder if this thread has strayed a bit though. Dave To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-isp" in the body of the message