Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84; site mtxinu.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!allegra!mit-eddie!godot!harvard!seismo!hao!hplabs!pesnta!amd!dual!unisoft!mtxinu!ed
From: ed@mtxinu.UUCP (Ed Gould)
Newsgroups: net.periphs
Subject: Re: 6250 tape drive experience ??
Message-ID: <273@mtxinu.UUCP>
Date: Fri, 18-Jan-85 19:43:01 EST
Article-I.D.: mtxinu.273
Posted: Fri Jan 18 19:43:01 1985
Date-Received: Mon, 21-Jan-85 04:37:29 EST
References: <177@hsi.UUCP>
Organization: mt Xinu, Berkeley, CA
Lines: 25

> We're looking for a modestly priced ($10K - $20K) 6250 bpi tape system.
>  ...
> My questions are: has anyone had any experience with either of these ??
> Any other recommendations ??  Do you really think that the Cipher
> with 128 Kbytes of cache will ever stream on a 785 ??  (We don't want to
> make any kernel changes to have the drive stream.)  Thanks.

I don't have any experience with the Cipher 6250 drive, but I've
used their 800/1600 CacheTape (M891).  It has a 64K cache, and it
works quite well.  It really does stream much of the time;
the software won't see the difference when it doesn't.

Among the things that Cipher did right in their CacheTape is EOT
handling.  The problem is that you don't want data stored in the
cache when the EOT marker goes by - the system should know about
end of tape at the right time.  What Cipher did is monitor how much
tape is left on the supply reel, and when it gets below some
(small) threshold, they start reducing the effective size of the
cache until nothing is cached at all just before EOT.  Of course,
the drive runs start-stop for the last several feet, but who
cares?

-- 
Ed Gould		    mt Xinu, 739 Allston Way, Berkeley, CA  94710  USA
{ucbvax,decvax}!mtxinu!ed   +1 415 644 0146