home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: bit.listserv.cwis-l
- Path: sparky!uunet!think.com!sdd.hp.com!ux1.cso.uiuc.edu!news.cso.uiuc.edu!uxa.cso.uiuc.edu!bau50671
- From: ullmer@uiuc.edu (Brygg Ullmer)
- Subject: Re: Central Management / Desktop Gopher (Universe)
- References: <CWIS-L%92111514443821@WUVMD.WUSTL.EDU>
- Message-ID: <BxsDoz.5s3@news.cso.uiuc.edu>
- Originator: bau50671@uxa.cso.uiuc.edu
- Sender: usenet@news.cso.uiuc.edu (Net Noise owner)
- Reply-To: ullmer@uiuc.edu
- Organization: University of Illinois at Urbana
- Date: Mon, 16 Nov 1992 02:02:58 GMT
- Lines: 50
-
- Steve Cavrak <SJC@UVMVM.BITNET> writes:
-
- >On Fri, 13 Nov 1992 04:40:12 EST Shyela Aberman said:
- > As someone pointed out, this gets more complicated because of its
- > nature. If the system is decentralized, management becomes that
- > much harder. What if I turn my machine off every night? The data
- > I have on it becomes unavailable to the rest of the universe. Thus,
- > for now, centralized information servers seem to still have a
- > definite place in CWIS's.
-
- >What we'll see, I think, is a "spreading" of information. Something that
- >probably closely mirrors the paper information systems we're used to.
-
- In many cases, distributed redundancy is one of the most
- powerful tools for dealing with distributed information. For
- instance, any given execution of "archie" is likely to
- return dozens of locations where the same resource
- or various versions of the same resource are available. In the
- case of Gopher and other distributed information systems,
- one of the key features necessary to seamlessly build upon this
- redundancy is the Universal Document Identifier. Hopefully,
- these identifiers will offer easy handles on resource versions
- so that redundant sources of information can be effectively
- coordinated.
-
- Concerning the point regarding machines turned off at night: with
- decentralized systems, data is often serviced on the
- local networks from a machine which *isn't* turned off. Mainframes,
- Unix workstations, and many other similar machines generally
- aren't turned off, either; however, your point is well taken,
- and I agree that the coordination of information hosted on
- personal systems is a point which will have to be dealt with.
-
- In terms of Gopher and distributed systems, I've operated for nearly
- half a year the Gopher service at the Univ. of South Carolina Dept.
- of Mathematics, upon which the vast majority of information is
- distributed across the net (access to many, many gigabytes is linked
- in, while fewer than 10 megabytes are stored locally). Downtime
- of linked machines has indeed been a significant problem, particularly
- in cases where the remote sites only store links to other remote sites
- (so while the actual source of information may be up, the Gopher service
- can't find its way to the site because the link file is not available).
- Intelligent caching would be a real plus here, and is a solution I
- hope to implement in the "Personal Gopher" (with personally configurable
- information spaces) I'm currently developing. Here again, though,
- redundancy is the key feature I hope to exploit for more constant
- uptime of resources.
-
- Brygg Ullmer (ullmer@uiuc.edu)
-
-