home *** CD-ROM | disk | FTP | other *** search
- Comments: Gated by NETNEWS@AUVM.AMERICAN.EDU
- Path: sparky!uunet!uvaarpa!darwin.sura.net!paladin.american.edu!auvm!UC.EDU!LINDA.NEWMAN
- X-Envelope-to: CDROMLAN@IDBSU.IDBSU.EDU
- MIME-version: 1.0
- Content-type: TEXT/PLAIN
- Content-transfer-encoding: 7BIT
- Posting-date: 25 Jan 1993 15:38:00 -0500 (EST)
- Importance: normal
- Priority: non-urgent
- X-Hop-count: 2
- A1-type: MAIL
- Message-ID: <01GTXXBVMKJY8WXFF5@pmdf.uc.edu>
- Newsgroups: bit.listserv.cdromlan
- Date: Mon, 25 Jan 1993 15:08:00 -0500
- Sender: "CDROMLAN@IDBSU - Use of CDROM Products in Lan Environments"
- <CDROMLAN@IDBSU.BITNET>
- From: Linda.Newman@UC.EDU
- Subject: Re: Proquest 4.0 Searchware
- Lines: 60
-
- We have conducted some concurrent user tests on our network (CBIS/Novell)
- of the 4.0 Proquest Searchware, and have experienced some problems.
- I'd like to hear if anyone else has done a deliberate concurrent user
- test, what your results were like, and what your network is like.
-
- Our tests have been with ABI/Inform, installed on a CBIS 386 Server with
- 8 mb of RAM. Our workstations vary. I have noticed an improvement in
- performance on a 386 with DOS 5.0 and about 590K of application RAM
- available, versus a 386/SX with DOS 4.0 (and no other memory manager
- like QEMM) and 509K available. However, the application still runs on
- that lower-powered machine. We haven't tried to run it on a networked
- XT or 286 yet.
-
- However, our real problem occurs with concurrent use. In our test,
- when we increased to 5 or 6 users, searches which would otherwise
- work with acceptable response time, no matter how ridiculously general
- (ex. 'quality circles and management') started taking 8 to 10 minutes.
- In at least two instances, several workstations received 'file error'
- messages generated by the ProQuest interface, and at that point, the
- CBIS file server had to be re-booted, because users of non-UMI products on our
- network started experiencing problems, and it appeared that something
- (cache memory?) on the CBIS server had become trashed, as a result
- of our user test with the UMI software. Oddly, the file error
- messages from the UMI software implied that the workstations had run
- out of room on the local C drive, which was where the temporary files
- were being routed, but 30 to 40 mb of space was available on all 6
- workstations being used for the test. It appeared that about 15 users
- had active connections to that CBIS server the second time it was trashed.
-
- In our test, we were cueing the testers so that all 5 or 6 hit the
- enter key at the same time, with the same search. When we use
- the previous version of the ProQuest software, having users do the same
- search speeds up results, whereas we see significant slow-downs when
- we get up to 6 users doing different searches, all general.
- Admittedly, our tests placed more demand on the product than one would
- expect with a typical 6 to 8 concurrent users, who would not really be
- entering their searches at the same time, but we are still concerned.
-
- We have not yet made this software available to the public, but will
- be doing so soon, and I'll be watching to see if we have to start
- re-booting our CBIS servers much more frequently when users are using
- UMI products.
-
- We have plans within the next month to upgrade all workstations which
- are 386/SX or better to DOS 5.0 or QEMM, but we don't have control
- over remote users on campus Novell LANS not in our Libraries. We also
- have plans to upgrade our two CBIS servers to 486's with 16mb of RAM.
- I will post a message to this list with the results, if, after making
- those upgrades, we do another concurrent use test of the ProQuest 4.0
- software, or if actual production use of the new software yields
- different results than our tests.
-
- I'd be very interested in hearing about other sites' experiences.
-
- +------------------------------------------------------------------+
- | Linda Newman | BITNET: LNEWMAN@UCBEH |
- | University of Cincinnati | INTERNET: LNEWMAN@UCBEH.SAN.UC.EDU |
- | Library Systems Office | FAX: (513)556-2161 |
- | Cincinnati, OH 45221-0033 | VOICE: (513)556-1441 |
- +------------------------------------------------------------------+
-