incubator-general mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Federico Strati <fede.str...@gmail.com>
Subject [PROPOSAL] RDBMSSR: RDBMS Software Redundancy Proposal
Date Thu, 24 Jan 2013 07:52:09 GMT
Hello,

I had no answer to the first post of this proposal, but I sent
it during xmas holidays.

So I send it again, I really would like to discuss the ideas
proposed.

Maybe you may suggest me the appropriate forum as well.

Cheers

Federico


------ Short Description ------
RDBMSSR: RDBMS Software Redundancy Proposal.

PROBLEM / GOAL: avoid any loosing of up-time service of an application 
using either
commercial or non-commercial databases due to yet uncovered and undiscovered
bugs in database middleware.

PROPOSED SOLUTION: Do make use of a redundant approach in the use of the 
database
middleware by diversifying databases, schemas in the databases, 
relational or non relational
data models, or using explicitly modifiers of any kind to SQL statements.
The “Multiple Paths” model is proposed: do use multiple different paths 
and algorithms
to access the very same data beneath them all.

FILES: all the complementary files described below are available at the
following URL: https://sourceforge.net/projects/rdbmssr .
------ Short Description ------

------ A bit Longer Description ------
The basic idea is to generate from an unique input DataModel of the RDBMS
hundreds of multiple paths per SQL statement.

The way we generate the multiple paths is described in the powerpoint file
"RDBMS_Software_Redundancy.pptx" (other format is 
"RDBMS_Software_Redundancy.html").

The basic idea is to use multiple databases (commercial as well as non 
commercial),
to use multiple different schemas in order to model the same data 
(relational
and non relational), to use the modifiers of any kind available for a given
RDBMS platform in order to generate different SQL statements to access 
the same
underlying data.

Then we may attach a probability and a time-out to each path
and do compute explicitly for all paths, summing up contributions.
------ A bit Longer Description ------

------ Proposed Solution ------
The Proposed Solution will have (minimally) the following features:

A) An abstract DataModel generator -> The input DataModel is entered and 
checked
and from it, iteratively if needed, the whole number of different 
“paths” is generated.

B) A software interface consisting of OR the DBO (Dynamic Business 
Object) model
(presented as a very simple and short example in the "dbo.zip" file in C++)
to represent data-types (intrinsic architecture)
OR a pre-compiler able to adapt to existing concrete objects
and able to generate adapted concrete classes (extrinsic architecture, 
no example given).

C) A software interface consisting of the DB pool managers and handlers
specific to commercial and non-commercial databases alike (as well as 
platforms).
------ Proposed Solution ------

------ PUBLIC DISCLAIMER ------
The proposed idea is in its very early stage of conception, so any 
discussion,
even if completely against it (but with good reasons), will be appreciated
as well as any pointer to related on going efforts in any project.

The author of this email claim to make the Proposal to the APACHE "brand"
for an open discussion of such topic, ideas against and for, and any 
other help
that may come from experienced open source developers.
------ PUBLIC DISCLAIMER ------

Thanks in advance

Federico

---------------------------------------------------------------------
To unsubscribe, e-mail: general-unsubscribe@incubator.apache.org
For additional commands, e-mail: general-help@incubator.apache.org


Mime
View raw message