Again I'm not familiar with the implementation of Image B-tree indices,
though I understand they're in the HFS, and would guess (bet?) that they're
stored as filenames.
At any rate, I think this stands as another testament to the efficiency of
FIRSTPartyIndexing: By maintaining a KSAM file in concert with an Image
Master OR Detail, I can move all the overhead of b-trees to KSAM, which is
very efficient as long as that file has only one key. And finding all the
entries that begin with "MA" is trivial.
Of course it WON'T find all the entries which CONTAIN "MA". So assuming
that the B-tree indexing scheme WILL fulfill the second request (it will via
the same mechanism used in :LISTFILE if implemented as I suspect), that
scheme genuinely represents an enhancement to Image, and TPI is not needed
at all if I'm willing to pay via an automatic AND the index, right?
Is this the reasoning behind the b-tree enhancement?
> -----Original Message-----
> From: [log in to unmask]
> [mailto:[log in to unmask]]
> Sent: Wednesday, August 09, 2000 9:23 AM
> To: [log in to unmask]
> Subject: RE: SQL Access to a Master Dataset
>
>
> You are correct Tracy, you should be able to just create a
> b-tree index on
> this field. You are also correct that it must be an image
> key...so if it's
> in a master..you may be out of luck if it's not already a
> key. If it's in a
> detail,you can create an automatic master on this field, then
> add a b-tree
> index and you're all set. If you need the b-tree on a
> non-key in a master,
> I think you're stuck with Third Party Indexes. (sorry, can't
> verify that
> last sentence..at the moment. Does it complain if you try
> ADDINDEX on a
> non-image key?)
>
> RAT
>
> ______________________________________________________________
> _________
> Rich Trapp "RAT"
> Managed Business Solutions [log in to unmask]
> http://www.mbsnav.com
> Assigned to Design Automation Support at Agilent Technologies
> Telnet or 970-679-2221 [log in to unmask]
> Loveland, CO USA
>
> ______________________________________________________________
> _________
>
>
> -----Original Message-----
> From: Tracy Pierce [mailto:[log in to unmask]]
> Sent: Wednesday, August 09, 2000 10:15 AM
> To: [log in to unmask]; [log in to unmask]
> Subject: RE: SQL Access to a Master Dataset
>
>
> Truly hoping I'm all wet re the capabilities of native image
> b-tree indices,
> please correct my (mis)understanding, which is that: the key
> fields of a
> Master set can be effectively sorted by building a (b-tree) index.
>
> If that's the case, a SQL index update could know about that
> index because
> it's part of Image. But if LastName is not a key to a
> Master, the only way
> to build an index without 3PI is via 1PI: do it yourself. If
> that's all
> true, how is SQL gonna know about the index (a ksam file, of course!)?
>
> TIA
> K Tracy Pierce, Systems Programmer
> Golden Gate Bridge, Hwy & Trnsp Dist
> PO Box 9000, Presidio Station
> San Francisco, CA 94129-0601
> mailto:[log in to unmask]
> 415-923-2266
>
>
>
> > -----Original Message-----
> > From: Rich Trapp [mailto:[log in to unmask]]
> > Sent: Wednesday, August 09, 2000 8:29 AM
> > To: [log in to unmask]
> > Subject: Re: SQL Access to a Master Dataset
> >
> >
> > Todd,
> > If you've got Omnidex or Suprdex indexes already on the
> > last name field,
> > just enable the database for third party indexing (TPI) and
> > do an "update
> > statistics for table <tablename>;" on the SQL side (so it
> > knows about the
> > indexes) and you should be all set.
> >
> > If you create indexes on the TurboImage side, you can
> > accomplish the same
> > thing without the third parties. You may have to do the
> > update statistics
> > after that also.
> >
> > RAT
> >
> > ______________________________________________________________
> > _________
> > Rich Trapp "RAT"
> > Managed Business Solutions [log in to unmask]
> > http://www.mbsnav.com
> > Assigned to Design Automation Support at Agilent Technologies
> > Telnet or 970-679-2221 [log in to unmask]
> > Loveland, CO USA
> >
> > ______________________________________________________________
> > _________
> >
> >
> > -----Original Message-----
> > From: Matteson, Todd [mailto:[log in to unmask]]
> > Sent: Wednesday, August 09, 2000 9:25 AM
> > To: [log in to unmask]
> > Subject: SQL Access to a Master Dataset
> >
> >
> > Although this may seem obvious to some, I have yet to become
> > a novice on
> > ODBC/ISQL/SQL and access to IMAGE DB's using SQL. My
> > question is this: I
> > have a Manual Master dataset with DBE attached. I wish to
> > produce a query
> > for a subset of the manual master. For this examples sake:
> >
> > Employees is my Manual master. SS# is the key. The DB is
> enabled for
> > IMAGE/SQL and the query is working (although slowly).
> >
> > In an effort to improve access speed online, I want to do an
> > SQL through
> > ODBC that doesn't read all records in the IMAGE DB serially.
> > For example,
> > all last names starting with 'MA'. In IMAGE, this would
> > require a serial
> > read.
> >
> > Having said that...On the HP3000 (all on the 3000) I could
> > accomplish this
> > through a 3rd party indexing product such as Superdex or
> > Omnidex and build
> > my key as I see fit. My programming would take care of the
> > rest of the
> > details. I've done this before and response times are good.
> >
> > With ODBC using an SQL statement, I am afraid that I might
> > not be able to do
> > this in an efficient manner. Testing this method seems to
> > read all records
> > serially. Do I need to read the Allbase/SQL(Create Index)
> > manual or where
> > would I go to research how to get to my desired result. If I
> > were to create
> > an index using ISQL, is this index maintained in the DBE
> > only? Any broad
> > direction on where to look for research material would help.
> > Thanks in
> > advance.
> >
> > Todd Matteson
> > Business Application Manager
> > Huffy Service First
> > 8521 Gander Creek Drive
> > Miamisburg, OH 45342
> > E-Mail:[log in to unmask]
> > Phone:(937) 436-5141
> >
>
|
