wedgingt@udenva.UUCP (Will Edgington/Ejeo) (09/03/85)
Sorry if you've seen this before; we had problems with our news feed
that probably didn't let this get out the first time.
Here at the University of Denver (as you may have found out in
another recent posting of mine, net.wanted & net.unix <797@udenva.UUCP>)
we have just acquired networking hardware for most of our computers.
We have a VAX 11/750 running BSD 4.2; another running Ultrix-32 v1.0
(both soon to be BSD 4.3); a PDP 11/70 running BSD 2.9 which, according
to current plans, will not be networked; a VAX 11/780 running VMS
4.1; and 4 (soon to be 5) VAX 11/750s running VMS 4.1, clustered to
each other and the 780. All these are in one machine room; we also
have a Pyramid 90x running OSx (BSD 4.2 and System V in tandem) which
is a couple of blocks away. All of the VAXs are about to be networked,
either directly or through the VMS cluster. We have DEUNA boards for
the ethernet connections; we will almost certainly run TCP/IP (see my
other posting for more info).
The two Unix VAXs and possibly the PDP are about to inherit a tape
drive from the VMS cluster. The first VAX (4.2) and the PDP currently
share a dual ported Cypher that is leaving (none of us has ever liked
or even felt indifferent to it). The second VAX (Ultrix) has a TU80
that we like but is only 1600 bpi. The drive we're inheriting is a
System Industries tri-density (800, 1600, and 6250 bpi; autoloading,
which neither of the others are). The current plans are to dual port
the SI drive to the two VAXs and give the TU80 to the PDP. I would
rather see the SI ported between the PDP and, say, the first VAX.
Then the second VAX could use the SI over the network, resulting in
all three machines being able to use the SI drive instead of only two
and two tape drives available to the VAXs instead of only one.
First, how much will dumping an 808 MB filesystem (we have ra81s on
the VAXs) over the network affect :
A. The receiving machine (the one the SI tape drive is actually
connected to, in this case) ? In what ways ?
B. The network as a whole, i.e., what percent of the possible
traffic on the network will be taken up by the dump ? How
much will other traffic be restricted ? Will VMS machines
talking to each other over the same network be affected ?
C. The time it takes to make the dump ?
Note that the machine that is dumping will be in single user mode and
that we normally do full dumps only quarterly (level 1 dumps weekly
and level 2 daily). Would any of the above be affected if the
receiving machine (with the tape drive) was also in single user mode
(and don't say "Obviously part A will be" :-) ? If so, how much ?
Second, does anyone know for sure which driver we should be using
with the SI (I know there's not much to go on, but it's all I have
myself) ? I've heard rumors a ht driver will work (and I'll test it
by booting on /genvmunix and looking at what it says is out there).
If the ht driver works, will it allow software selection of the bpi ?
I know this is possible because VMS does it; 'Software Select' is
on of the 'buttons' on the front of the drive (this is also another
clue as to exactly what model the tape drive is; 10 brownie points to
whoever can tell me :-). Also, how well will it stream ? All the
time ? Never ? Only without disk processing ? Both the 750s have
8 MB of memory, if that affects anything (even the PDP has 1.5 MB !).
If the ht driver is only so-so, does anyone have a better one ? I've
heard that SI had to install some code themselves to get VMS to work
with it; is this needed for Unix also ?
Third, you thought I was going to ask another question, didn't you ?
(-:-) (since you can take that two ways ! :-)
Thanks for any help you can give me !!!
--
Will Edgington | Phone: (303) 871-2081 (work), 722-5738 (home)
Computing Services Staff | USnail: BA 469, 2020 S. Race, Denver CO 80210
University of Denver | Home: 2035 S. Josephine #312, Denver CO 80210
Electronic Address (UUCP only): {hplabs, seismo}!hao!udenva!wedgingt
or {boulder, cires, denelcor, ucbvax!nbires, cisden}!udenva!wedgingt