[Info-vax] INIT/HIGHWATER

[Michael Moroney]

koehler at eisner.nospam.encompasserve.org (Bob Koehler) writes:

>In article <iil1g9$rob$1 at pcls6.std.com>, moroney at world.std.spaamtrap.com (Michael Moroney) writes:
>> helbig at astro.multiCLOTHESvax.de (Phillip Helbig---undress to reply) writes:
>> 
>>>How much of a performance hit is /HIGHWATER?
>> 
>> It just sets a bit in the drive metadata.

>   In response to which, there is processing.  Not nearly as much as
>   /ERASE, but we have seen it, and turned if off where appropriate
>   to the data we were storing.

I probably should have mentioned the effects of /HIGHWATER, but the 
original question seemed to ask about the speed of the $ INIT command
itself.

Highwater usually stays out of the way, most applications have no need
to try to read beyond the logical end of data and thus the highwater,
but when one does the read is intercepted to return "read" data of
all zeroes.

Writes are a little different.  Again, very few applications attempt to
write beyond the end of a file other than the obvious case of adding data
directly to the previous end.  However, if one does, with a gap between
the previous end and the new data, that data must be written with zeroes
before the new data is added.  This can cause a slowdown with large files
when a large number of blocks are zeroed.  Also in this category is simply
extending the logical end of file to a larger value without writing any
data.  As John mentioned, a common cause of slowdowns is extending
pagefiles, which is otherwise simply allocating blocks and moving the end
pointer.