From owner-freebsd-isp@FreeBSD.ORG Sun Apr 6 14:26:36 2003 Return-Path: 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 0E54B37B401 for ; Sun, 6 Apr 2003 14:26:36 -0700 (PDT) Received: from netmint.com (netmint.com [207.106.21.130]) by mx1.FreeBSD.org (Postfix) with ESMTP id 5F73D43F93 for ; Sun, 6 Apr 2003 14:26:35 -0700 (PDT) (envelope-from support@netmint.com) Received: from netmint.com (localhost.netmint.com [127.0.0.1]) by netmint.com (8.12.6/8.12.5) with ESMTP id h36LQYp7022917 for ; Sun, 6 Apr 2003 17:26:34 -0400 (EDT) (envelope-from support@netmint.com) Received: from localhost (support@localhost) by netmint.com (8.12.6/8.12.3/Submit) with ESMTP id h36LQYQv022913 for ; Sun, 6 Apr 2003 17:26:34 -0400 (EDT) Date: Sun, 6 Apr 2003 17:26:34 -0400 (EDT) From: Support To: freebsd-isp@freebsd.org In-Reply-To: <3E90938D.2050307@mac.com> Message-ID: <20030406165741.K22348-100000@netmint.com> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Subject: Re: load testing and tuning a 4GB RAM server X-BeenThere: freebsd-isp@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Internet Services Providers List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sun, 06 Apr 2003 21:26:36 -0000 [ ... ] > Aren't you using a Dell PowerEdge and hot-swappable drives? I also > thought you mentioned you were using 15K drives for swap, which implies Yes, PowerEdge with PERC3-DI (128MB) for Raid 1 with 2x36GB 15KRPM drives and with PERC3-DC (128MB) for Raid 5 with 4x146GB 10KRPM drives. > SCSI...probably 80-pin SCA form-factor, right? The drives are hot-swappable in themselves, but I can't add a Raid array on the fly, nor can I add to the capacity of a Raid array. I am sure that even if the Raid card allowed me to do that, FreeBSD wouldn't support a new Raid array that just popped up without a reboot. I am not even sure if Windows can do that. Now you know the hardware, please correct me if I am wrong about this. > If you knew that the difference in KVA memory would be significant to > your usage, then you could evaluate whether saving a couple of swap > device slots is worth the loss of flexibility. Configuring a system > without any tolerance for change is a little like Procrustes being too > precise in measuring his own bed. :-) But if what I said about Raid is correct, how is having an extra slot in RAM for swap more flexible? :) Of course, if I _could_ add drives without rebooting, or better yet add Raid arrays, I would be a very happy camper, and no question about it, would up the NSWAPDEV. > I'm not sure you measure "established" the same way I do. Do you mean > you expect there to be 500 to 4000 active apache children all processing > transactions 24-7, or do you mean you expect to see anywhere up to 4000 > people using the site(s) at a time, clicking at whatever rate they'd use > the site(s) during normal transactions? Sorry, let me rephrase this a bit. I mean network connections to the server that are marked "ESTABLISHED" in netstat -an output, if you grep for the webserver. In other words, if my understanding is correct, these are connections currently active that the webserver is serving with information and have running PHP/Perl sessions or are otherwise downloading files. I could have 1500-2000 to as high as 4000 of those. > How much of your traffic is going to be over SSL? You might want to > look into getting a HI/FN crypto-accelerator card, particularly if you > have lots of small/short SSL sessions rather than few longer ones. Not sure as to the exact break-down, but estimate 5-15%. Is that large in the context of total connections? Perhaps. I will look into a crypto card. > You really want to run only one type of production database per machine; > you're risking VM thrashing otherwise. Even if the load on the second one is _much_ less? Please explain why this is so, if possible. > You might have something like a Dell/Adaptec PERC? RAID controller with > 128MB or so of I/O buffer memory which can also do the RAID-5 XOR > calculations? That will help, but even so RAID-5 write performance goes > from adequate to poor as the I/O load increases. Also, have you been > testing I/O while also hitting a database (or two) at the same time? Right. I suppose I _could_ get another 146GB drive, and reconfigure the 4-drive Raid 5 array to be a 2-drive Raid 1 and 3-drive Raid 5. Or, I could wait until enough users hit the server and as load creeps, invest into two new drives and add a Raid 1 array at that point. I think I will choose the latter approach. As far as my testing, it's been very raw and no, I haven't done any DB testing yet. I was mostly testing for stability under heavy heavy memory usage, i.e. a bahzillion lynx'es and -j500 makeworld. How would you do this? :) Thanks for your help, Andrew