[Info-vax] simple HBVS question

[Phillip Helbig---undress to reply]

In article <j7n322$308$1 at usenet01.boi.hp.com>, "Robert A. Brooks"
<rab at aitchpee.com> writes: 

> On 10/19/2011 12:58 PM, Phillip Helbig---undress to reply wrote:
> > Now that I have set up HBMM on some shadow sets, the bitmaps are zeroed
> > some time after the threshold is reached.  The lower the threshold, the
> > faster the MM, though of course there are disadvantages from having the
> > threshold too low; in particular, any reasonable value for the threshold
> > will allow a MM to take place much more quickly than with an infinite
> > threshold, at least for a disk whose content gets completely recycled.
> 
> > Does this mean that a shadow COPY will also take place more quickly
> > (assuming that conditions are such that a minicopy can be performed)?
> 
> No, other than the multiuse bitmaps (which were added post-7.3-2 and I 
> don't think were backported), HBMM and minicopy bitmaps are completely 
> separate.

OK, I'm still at 7.3-2, so obviously I am limited in that respect.

Does multiuse mean that both HBMM and minicopy use the same bitmap?  If 
not, why not?

In 7.3-2, are there separate bitmaps?  If so, can the COPY (as opposed 
to the MERGE) bitmaps be zeroed?

> You can experiment to see how long a merge (either full or mini) will 
> take by using $SET SHADOW /DEMAND_MERGE to force a merge.  My guess is 
> that you'll be shocked at how quickly an HBMM will finish.

I just tried it and it took about as long as I expected (few seconds 
without many blocks having changed) based on experience with minicopy.