Streaming TK50

Johnny Billquist bqt at softjar.se
Thu Mar 26 17:17:56 CDT 2009


Ethan Dicks <ethan.dicks at gmail.com> wrote:

> On Tue, Mar 24, 2009 at 3:30 PM, Paul Koning <Paul_Koning at dell.com> wrote:
>>>>>>> "Ethan" == Ethan Dicks <ethan.dicks at gmail.com> writes:
>>  Ethan> ...and I don't know if a uVAX-I can keep a TK50
>>  Ethan> busy, ...
>>
>> I should hope so, given that a modest PDP-11 could do so easily.
> 
> I wouldn't be so certain - my experience is that the the way the TK50
> works, you can store *up to* 95MB per tape *if* you can keep the silos
> full.  A MicroVAX-I is a slooooow beasty.  There's no problem with the
> Qbus itself, but my concern would be that system overhead, interrupt
> latency, RQDX performance, etc., could be bad enough that the OS just
> couldn't fill tape buffers fast enough.  Of course, trying to back up
> 30 MB onto a 95MB tape doesn't require perfect performance.

No. You're mixing two different things.
The speed of the drive is very much dependant on keeping the silos full. 
Otherwise it will go into start-stop mode, in which the speed gets 
horrenously slow. And I wouldn't be surprused finding out that a uVAX I 
can't keep a TK50 streaming. It's a slow beast. A "modern" PDP-11 is 
much better.

However, streaming or not should not affect capacity as such. Block 
sizes on the other hand do. Doing small blocks drastically reduces 
capacity, since you have inter-block gaps on the tape.

> Thinking about it, perhaps it's worth the effort to use Standalone
> Backup - that should minimize any OS effects on backup speeds.  Once
> the tape is written, ISTR the TK50 will back the tape up as necessary
> to extract what's on it.  It's writing that you _must_ optimize.

Well, from a performance point of view, both reads and writes should be 
optimized. Capactiry-wise shouldn't matter. Tape will start-stop and 
rewind as needed for both reads and writes.

I'm looking through TK50 manuals, but can't find anything about 
inter-block gaps. The TU81 (for example) can have two different sizes of 
IBGs, which will affect capacity, and which can be related to tape 
speed. As I said, I can't find anything like this for the TK50, but if 
there is, it might cause what you are thinking of (except that the IBG 
is small by default when in start-stop mode on the TU81, and you have to 
explicity ask for a larger IBG).

Other comments about TK50s? Well, they need a lot of cleaning, and you 
have to disassemble them when the pick-up skips the arm, which is pretty 
common. But anyone who have been playing with these drives probably 
knows this already. :-)
Otherwise they keep working pretty reliably.

	Johnny

-- 
Johnny Billquist                  || "I'm on a bus
                                   ||  on a psychedelic trip
email: bqt at softjar.se             ||  Reading murder books
pdp is alive!                     ||  tryin' to stay hip" - B. Idol


More information about the cctalk mailing list