This message came in after posting the SUMMARY:
Alan has given info in an interesting manner (to my understanding, at least)
which follows:
----------------------------------------------------------------------------
--------------------
The 8 KB is the fundemental allocation block size, not
necessarily the transfer size. If you manage to allocate
the database to be contiguous, AdvFS won't interfere with
whatever transfer size you want to use, until you get to
around 64 KB. Even then I think AdvFS will allow larger
transfer sizes, but LSM or the underlying device might get
in the way.
----------------------------------------------------------------------------
--------------------
My comments are:
How come HP/UX is capable of modifying the block size, and not DU?
Isn't the block size an important factor for boosting large reads/writes,
just like the chunk size when stripping with HSZs?
Thanks Alan and all members of this list.
BTW, his email address is alan_at_nabeth.cxo.dec.com should you want to
mail him.
____________________________
Alexander Sakellaris
Sr. Systems Administrator
Operations Department / IT Division
Network/Systems & Support Sector
COSMOTE Mobile Telecomms S.A
Athens / Greece
email: asakelaris_at_cosmote.gr
Voice: ++30/1/6177480
Mobile: ++30/97/277480
Fax: ++30/1/6177594
____________________________
Received on Fri Nov 13 1998 - 19:05:21 NZDT