Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!uunet!world!bzs From: bzs@world.std.com (Barry Shein) Newsgroups: comp.unix.large Subject: Re: Hello Message-ID: Date: 11 Sep 90 02:41:47 GMT References: <6f7y02Ubc6wm01@amdahl.uts.amdahl.com> Sender: bzs@world.std.com (Barry Shein) Organization: The World Lines: 30 In-Reply-To: barn@convex.com's message of 10 Sep 90 13:02:45 GMT The difference is that when things get big you eventually hit a paradigm shift in the management of those resources. You can't manage a big system by simply doing what you do for small systems, only more of it. Consider when you get to the point that you can't perform a daily backup in less than 24 hours. Obviously something would have to change, more of the same won't cut it. If you think that's ludicrous, there are terabyte Unix systems out there. That's 10^12 bytes. If 10% needed to be backed up every day that would be 10^11 bytes. At about 2x10^8/tape we get 500 full tapes per day. If there were one tape drive that would leave about 3 minutes per tape to get it done in exactly 24 hours (and then start again.) 5 tape drives running full blast simultaneously, a more realistic 15 minutes per tape, and it still takes 10 hours to do a lousy daily incremental. And you have 500 tapes per day to manage! Clearly things do not scale linearly as systems get large, completely different management and technology strategies must be employed. It is those different strategies that I would hope this group were interested in. Where are the discontinuities? How does one manage them? -- -Barry Shein Software Tool & Die | {xylogics,uunet}!world!bzs | bzs@world.std.com Purveyors to the Trade | Voice: 617-739-0202 | Login: 617-739-WRLD