From owner-freebsd-isp Mon Jan 5 22:03:05 1998 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id WAA05063 for isp-outgoing; Mon, 5 Jan 1998 22:03:05 -0800 (PST) (envelope-from owner-freebsd-isp) Received: from zen.triax.com (zen.triax.com [206.58.96.13]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id WAA05044 for ; Mon, 5 Jan 1998 22:02:51 -0800 (PST) (envelope-from mfred@zen.triax.com) Received: from localhost (mfred@localhost) by zen.triax.com (8.8.7/8.8.7) with SMTP id WAA23826; Mon, 5 Jan 1998 22:02:34 -0800 (PST) Date: Mon, 5 Jan 1998 22:02:33 -0800 (PST) From: MegaFred To: =?iso-8859-1?Q?=22Luis_E=2E_Mu=F1oz=22?= cc: freebsd mailing list Subject: Re: ISP Conversion In-Reply-To: <3.0.5.32.19980104222917.007b9810@pop.cantv.net> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-isp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > I've seen many people using sendpage... Haven't tried it myself, but > certainly will. Anyone know if it has TAP support? > (1) User request a given article from your news server > (2) news server reads the file from the NFS server > (3) news server replies to the client Well, I definately wouldn't save the new traffic on a centralized NFS server, just maybe the web pages, the user home directories, etc. Is this still a flawed idea? > You also have all your eggs on a single basket; if one of your 4Gig > disks die, all of your services will go down though they might have > survived (you still have perfectly good servers). What about data mirroring? I'm not sure of all the raid levels, what they mean and what they do, nor even what the terminology is, but what about 12 hot-swappable 4-gig drives, 6 of which are live, the other 6 mirroring the previous ones? That way, if one of the drives goes bad, the OS instantly starts using the 'replicated' drive, allowing you time to pull the bad one out, replace it, and put a new one in for a live drive. Is there a term for this? Is this common practice? Is it feasable, cost-effective, or am I better off just going distributed? My hopes in this was to avoid re-partitioning drive when the need for more space arrives. > It *might* be reliable enough. Computers are too complex for this. > For instance, you only need your HD for booting the OS. After this, > the HD becomes a critical point. If it fails, your router goes down. True, but for all intents and purposes, would it work smoothly to route IP over frame-relay, serving nothing more then a terminal server of 16-32 ports? How about using it for nothing more then a gateway for a lan? All i need it for is network throughput... Reason I'm even considering this is we have so many 486 motherboards, CPU's, cases, et. al, that if all I needed were CSU cards to have an instant router, we'd probably save ALOT of money in the long run making our own. > If you're under a cost constraint, try to get a few used Cisco 2501. > They're *much* more reliable than servers for this job (in my opinion I was looking into these, and could not find anything other then there "enterprise" solutions (the 32k-50k high-end routers). Anyone know of a URL to get a catalogue of cisco's line of routers? TIA Joe