Not really, not as such.  There's not much call for partitioning
across machines unless you have a fancy high-speed interconnect, or
unless they are so nearly independent that they might as well
be separate tables.  Then, to make use of it properly you need
a lot of Star-like machinery on top to put the pieces together.

I don't see this happening unless someone comes along with a
proposal and eager hands to do the work.  (A contribution would
be entirely welcome, though.)

Karl

I think DATAllegro may have done all this sort of work already.

Hello Sam (an other),

Very intersting mail, I follow you in that direction.

What you explain here is in the cluster scope, and what you say about saving
money is, from my point of view, true. Cluster is a large subject and I
think many people confuses the concept and the technology (system cluster
and from the Ingres side things like VMS cluster or functionnality announced
with Ingres2006).

What you also need to include here a knowledge layer about star, perhaps
replicator if you want to put some high avaibility option in your
consideration and for database administration. Dealing with 1 to 50 ingres
installation is one thing if they are all the same (I did that in the past),
dealing with 1 big system composed with 20 or more Inres installation is a
other stuff.

I have in mind a french customer (french army) with a very big replicator
configuration. They manage the distribution of information and its
confidentiality with replicator and at the end they have a very complex
scheme of replication. To deal each day with it, they have developed some
particular houskeeping scripts and procedure and finaly it work "well". I
say "weel" because if they can find a solution for each issue they have they
must maintain people with a very good knowledge of their applicaton, Ingres,
replicator, system, and something like a hacker spirit to deal with those
complex issues (and they have some people like that, it is due to the fact
they are working for the army and then are not thinking salary or career but
only to the service they must provide to their "customers"). Saving money on
hardware is one thing, saving money on all the project is an other thing.
What if your system is sometime down for some unfindable reasons ? Or if a
patch break something somewhere ? But even in a more simple configuration
you can have issues like that ...

I already discuss of this kind of subject with many people, in some formal
and less formal context, and few of them think it was doable et maintainable
in a regular basis. My point of view is all of this is a question of
discipline and rigour ... and tests. I see in my life only one customer
doing some real tests with users doing sometimes the same job on 2 system :
the production system and the next production system. This customer is a
bank and due to the amount of money captured by the information system they
have this heavy environment. All the other customers in the indusctry think
first to the bill of hardware and software, some others of the complete bill
(hard + soft + knowledge).

I hope you will acheive your project in the mood you explain and you'll find
all the help you need with Ingres, his support (premium or not), prehap's
this list and the future community site Inges announce. But sure I will read
your post with the first interest, don't care. If I can help you of any way
I will not hesitate ... (your 10TB table is too much interesting to miss a
knowledge on this subject :-))

Kind regards,
Jean-Pierre
15 Dec 2006 12:22:19 -0800, samantha_fein19@yahoo.com <
samantha_fein19@yahoo.com>:

It's not clear that plain, "old-fashioned" Ingres rule-based partitioning
alone wouldn't be a huge benefit to you if you spend a little time setting
it up.  Have you already concluded that would be no help to you?

But otherwise, you are describing a Datallegro DW appliance.  Go to
www.datallegro.com.

Roy