[comp.os.vms] UDABURSTRATE system parameter

SYSTEM@STAR.ST-AND.AC.UK.UUCP (10/03/87)

Hello all,
        If this is a well discussed topic I am sorry. I have been told that
if I change the value of UDABURSTRATE from it default (-1 on my microVMS
4.5 system) to a value of 4 disk throughput will improve. Since the machine
in question supports a group of Astronomers thrashing 2-D images on and
off disk if this process could be speeded up it would all go faster.
        Is there anybody out there who can tell me if this change of parameter
is good, bad ro a waste of a reboot?
        Many thanks in advance. If I collect interesting answers I will
collate them for the net.

Roger Stapleton
St.Andrews University Observatory, St.Andrews, Fife, KY16 9LZ  Scotland

JANET   SYSTEM @ UK.AC.ST-AND.STAR
EARN    SYSTEM @ STAR.ST-AND.AC.UK
ARPA    SYSTEM % STAR.ST-AND.AC.UK @ UCL-CS.ARPA  (if UCL-CS lets it in)
        SYSTEM % STAR.ST-AND.AC.UK @ UKACRL.BITNET  (maybe)
UUCP    ...MCVAX!UKC!UK.AC.ST-AND.STAR!SYSTEM       ( !perhaps)

Insert usual disclaimer where required.

dp@JASPER.Palladian.COM (Jeffrey Del Papa) (10/03/87)

    From:     SYSTEM@STAR.ST-AND.AC.UK

    Hello all,
	    If this is a well discussed topic I am sorry. I have been told that
    if I change the value of UDABURSTRATE from it default (-1 on my microVMS
    4.5 system) to a value of 4 disk throughput will improve. Since the machine
    in question supports a group of Astronomers thrashing 2-D images on and
    off disk if this process could be speeded up it would all go faster.
	    Is there anybody out there who can tell me if this change of parameter
    is good, bad ro a waste of a reboot?
	    Many thanks in advance. If I collect interesting answers I will
    collate them for the net.

    Roger Stapleton
    St.Andrews University Observatory, St.Andrews, Fife, KY16 9LZ  Scotland

    JANET   SYSTEM @ UK.AC.ST-AND.STAR
    EARN    SYSTEM @ STAR.ST-AND.AC.UK
    ARPA    SYSTEM % STAR.ST-AND.AC.UK @ UCL-CS.ARPA  (if UCL-CS lets it in)
	    SYSTEM % STAR.ST-AND.AC.UK @ UKACRL.BITNET  (maybe)
    UUCP    ...MCVAX!UKC!UK.AC.ST-AND.STAR!SYSTEM       ( !perhaps)

    Insert usual disclaimer where required.

As I remember this only applys to the UDA-50 Unibus controller, since microvaxen use
a different controller (q bus and all) I don't believe it really applys (but I
wouldn't be suprized if I was wrong) In any case you should be wary of tweaking this
parameter if you have a lot of other DMA hardware on your bus, especially the sort
with very shallow buffering (many older or locally cobbled dr-11w based interfaces,
this parameter controls how long the disk holds onto the bus before letting any other
activity)

If your disks are DEC RD 5x series, your bottleneck is not bus speed, but rather seek
times. Dec used PC class disks (high 20's - mid 30's average seek times) rather than
the more expensive timesharing grade (15-20ms seek). A third party controller and
real disks will make your system responsive (tho field service may try to disown you)

<dp>