Len,

> Well, I don't have to trust you.  The master database has to be exclusively
> on SQL database, any other products like Imail that need data from, or need 
> duplciate data from, will get synced to that, trying to sync in both 
> directions is crazy.

So...you  do  agree  with  me  that going back from Registry-->ODBC is
unfeasible?  Your initial "I don't have to trust you" is unnecessarily
combative, then. IAAP, that's why I ask for your trust on API-specific
matters.

> that  still splits the database between some it on Ipswitch/registry
> and other bits on SQL.

You  didn't  read  my  post  correctly.  In  this second section, I am
referring  to  the  fact  that  most  people  who  complain  about SQL
performance  vs.  Registry haven't even tried to tweak SQL, so they're
not   yet   qualified  to  complain.  Many  people  who  jump  on  the
Registry/SQL  hybrid  idea--which I still think is a good idea, that's
why  I'm  proposing  its  productization!--usually  haven't done their
homework  first  with  the  existing  tools.  Tweaking  includes smart
indexing,  hardware  rightsizing, Winsock Direct--just as it would for
*any* performance-critical database. Once they tweak it, queries could
be  many,  many times faster. Still slower than the Registry? Yes. But
groan-worthy?  Maybe  not  in  nearly  as  many cases. And if Ipswitch
publicized  a  bunch  of  hints and optimized their client-side calls,
most systems wouldn't need the hybrid system at all. There would still
be  some  with  a  deal-killing  need for it, but they would be in the
minority, as opposed to the norm.

> I didn't say it would be easy, but there has to be a central "customer
> database" on SQL as the foundation, master database, and other apps get the 
> bits they need from there.  Simply because SQL is a better general purpose 
> database technology than a registry database.

Like  I  said,  I have a proof-of-concept architecture that adheres to
your   specifications--with   the  *strong*  caveat  that  a  separate
management  interface  would  need  to  be built to preclude orphaning
account settings. If anybody who has done their due diligence with SQL
optimization is interested in this product, please let me know. I just
don't  want  to go building it, only to find that people's performance
is  still bottlenecked by a non-optimized back-end database. Let's get
that out of the way first.

> I really can't imagine running 200k users out of the registry as a 
> database.  (I remember at one point that MS admitting NT4 SAM pooped out at 
> 14K users.)

I  don't  know  what in my post prompted this comment, but you need to
look  at  the archives. In a message that I posted about a year ago, I
established  that this old chestnut only relates to the NT SAM and has
absolutely  nothing  to do with Imail. You yourself, in fact, referred
to  my  little white paper as a "keeper." Guess you didn't follow your
own advice!

Anyway,  I  don't  understand  why you, who are promoting its use as a
high-speed  cache,  would  start  attacking the Reg's fitness for this
purpose!  Nobody's  arguing  (I never have) that its query language is
suitable   for   multivendor   systems.   But   its   performance   is
kickass...that's  benchmarked  and  non-debatable...under server-level
load.  Throwing  in,  out of left field, doubts about its stability is
just  plain  silly--do you really want to run your cache off something
you  think  isn't  stable? It's still the primary point of contact for
Imail  in  that setup, and if it gets corrupt or what-have-you (you've
offered no proof of this, though), how do you propose that Imail knows
it's getting bad data and refetches from the main DB?

Sandy


Please visit http://www.ipswitch.com/support/mailing-lists.html 
to be removed from this list.

An Archive of this list is available at:
http://www.mail-archive.com/imail_forum%40list.ipswitch.com/

Please visit the Knowledge Base for answers to frequently asked
questions:  http://www.ipswitch.com/support/IMail/

Reply via email to