Date: Mon, 12 Mar 2001 07:53:31 -0500 From: "Brian" <wrath@shianet.org> To: <freebsd-isp@freebsd.org> Subject: Re: co-location model Message-ID: <003b01c0aaf3$7124b5e0$0101a8c0@fear.wrath.net> References: <Pine.BSF.4.21.0103112027080.10411-100000@workhorse.iMach.com>
next in thread | previous in thread | raw e-mail | index | archive | help
I'm with you 100%. I see my school buying overpriced Sun pieces of shit just because they don't know any better. They've got an Ultra10 workstation that is used for nonmajor computer science students. My desktop has more horsepower, and it only cost me $900 compared to their $6000. And my machine doesn't puke out when it gets tired. I think the real issue is space among everything else. If you reliable run multiple instances of an operating system on one larger machine instead of multiple machines with a single operating system, it'll save space. Imagine an ISP offering introductory cobalt raq's to their customers. You've got hundreds of these little machines filling up your racks, and most of them aren't even being used at 10%. With a "mainframe" you can have hundreds of "cobalt raqs" running in the space of the hundreds of cobalt raqs, many times over. However, I'd still stick with a bunch of cobalt raqs. When one of them dies, you yank it out of the rack and slap a new one in. Oh yeah, and I don't use any Cobalt machines. I'm just giving an example as to most colocator's suggested option for new customers. I'll be sticking with my P6DBE's and katmai's for a good time to come. brian@wrath.com ----- Original Message ----- From: "Forrest W. Christian" <forrestc@imach.com> To: "Jeff Gray" <jwgray@netbox.com> Cc: <freebsd-isp@FreeBSD.ORG> Sent: Sunday, March 11, 2001 10:59 PM Subject: Re: co-location model > Let me address these points, one by one: > > > real reliability, > > If what you mean by this is what basically boils down to server uptime and > availability, I think you will run into some difficulty in this group > selling the fact that a mainframe is essentially more reliable than a > well-designed intel platform running FreeBSD. > > I think that most of us never have downtime other than that caused by > planning (hardware upgrades) or environmental (extended power failures, > floods, etc.) causes. I sure can't remember the last time a server > rebooted without me forcing it to. > > Every mainframe installation I've been around is definately subject to the > same failures as I've described before. > > > real fault tolerance > > I think I probably need to have this better defined. I know of ways to > do hot failover, high availibility raid, etc. etc. etc. under FreeBSD. is > there anything specific that the mainframe provides which isn't what most > of us think as the classic definition of fault tolerance. > > > real hardware efficiency [jail does this] > > real security by a well designed mainframe OS management system, > > I've lumped these two together for one reason: > > Under freebsd, jail provides you a level of security. Yes, I realize you > can get out of a jail. I think we could argue whether most Mainframe > Installations isolate their users more or less than jail does. > > > real scalability of user resources like storage space. > > An example: > ------------------------------------------------------------------------- > 230-Welcome to ftp.cdrom.com, a service of Digital River, Inc. > 230-There are currently 528 users out of 3000 possible. > 230- > 230-This machine is a Xeon/500 with 4GB of memory & 1/2 terabyte of RAID > 5. > 230-The operating system is FreeBSD. Should you wish to get your own copy > of > 230-FreeBSD, please visit http://www.freebsd.org for more information. > ------------------------------------------------------------------------- > > Define scalability. The above is scaled up there rather high. > > 4GB is the most we can go memory-wise right now. This is a restriction > due to 32 bit addressing in most programs. FreeBSD could very easily > access up to 8TB of memory on a 64bit machine, assuming that the hardware > was up to the task. > > The maximum FILESYSTEM limit under freebsd is about 1TB. The filesystem > is actually capable of up to 8 or 16TB but there are some internals which > cause this to not be availble without some modifications. > > > -Is this a reasonable long term model for ISPs and or server farms? > > I think that this MIGHT be marketable to the right people. I know that > there are a lot of management types out there which still love to hear > the mainframe story line. I'm not sure how this could be marketable to > say, the members of this list. > > > -Does anyone offer this today at the scale of rack size bites of > > physical space? > > Not aware of any, but who says they don't exist.. > > - Forrest W. Christian (forrestc@imach.com) AC7DE > ---------------------------------------------------------------------- > iMach, Ltd., P.O. Box 5749, Helena, MT 59604 http://www.imach.com > Solutions for your high-tech problems. (406)-442-6648 > ---------------------------------------------------------------------- > > > To Unsubscribe: send mail to majordomo@FreeBSD.org > with "unsubscribe freebsd-isp" in the body of the message > To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-isp" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?003b01c0aaf3$7124b5e0$0101a8c0>