Comments on: The NoSQL Movement: The Object – RDMBS Incompatability http://halwebguy.danziger.net/2010/02/25/the-nosql-movement-the-object-rdmbs-incompatability/ following the crossroads of media and technology products Fri, 13 Aug 2010 14:59:24 +0000 hourly 1 http://wordpress.org/?v=3.0.1 By: Mickey Rahman http://halwebguy.danziger.net/2010/02/25/the-nosql-movement-the-object-rdmbs-incompatability/comment-page-1/#comment-190 Mickey Rahman Fri, 13 Aug 2010 14:59:24 +0000 http://halwebguy.danziger.net/?p=162#comment-190 In your example, are you saying its not effective because it doesn't map to exactly to the objects in code, or because its just slow and not efficient? Or because its just not intuitive? I think there are many ways to make it faster(other than what I stated above), at least fast enough to it not be "not effective" anymore.... In regards to it being a natural fit - I would argue whether how ALL your data is stored in its repository (be it a RDBMS, or any other system) actually needs to mirror your object models in code - they have different philosphy on how they work, and is stored the most efficient way so that the individual systems can work to their fullest capability. I mean, as programmers, its our job to integrate them in the best way possible (or so I believe). And in regards to it being intuitive - hey, can't win em all :) In your example, are you saying its not effective because it doesn’t map to exactly to the objects in code, or because its just slow and not efficient? Or because its just not intuitive? I think there are many ways to make it faster(other than what I stated above), at least fast enough to it not be “not effective” anymore….

In regards to it being a natural fit – I would argue whether how ALL your data is stored in its repository (be it a RDBMS, or any other system) actually needs to mirror your object models in code – they have different philosphy on how they work, and is stored the most efficient way so that the individual systems can work to their fullest capability. I mean, as programmers, its our job to integrate them in the best way possible (or so I believe).

And in regards to it being intuitive – hey, can’t win em all :)

]]> By: Mickey Rahman http://halwebguy.danziger.net/2010/02/25/the-nosql-movement-the-object-rdmbs-incompatability/comment-page-1/#comment-189 Mickey Rahman Fri, 13 Aug 2010 14:51:19 +0000 http://halwebguy.danziger.net/?p=162#comment-189 I've seen this problem solved in many ways before with a combination of things - no round trip is neccessary because you can do the recursive loop to get all the children in one set of sql statements, and then depending on the situation, cache it so that you don't have to go get it everytime. But there are much better algorithms to solve this in a RDBMS that are much faster, but not very intuitive - doesn't map to the objects very well. Hal, I'm partly from the same dinosaur tribe that you are :) but I know the current trend is to have developers do as little DB developement as possible, mainly for the reason that you stated, that they would have to write in a second language. Personally I don't consider that a tragedy, but I can see how that can become a problem. But thats what DBAs and DB developers are for, no? Although I hear that in the trend to separate the programmers from DB as much possible is making those positions obsolete.... In reference to the second comment - I see a few issue with that solution in most scenarios, but maybe you can clarify them: How does the XML repository store the data? In a text file? How do you maintain that, and what is the concurrent access problems you will see in that? I suppose you can store the whole blob in the DB as well in a text/large varchar field - but also, if its a category-sub catgeory hierarchy, how would you relate this to the rest of the data that its associated with? There will be no referencial integrity if its all in a blob.... I'b be curious to see more solutions to this problem in a non-RDBMS system.... I’ve seen this problem solved in many ways before with a combination of things – no round trip is neccessary because you can do the recursive loop to get all the children in one set of sql statements, and then depending on the situation, cache it so that you don’t have to go get it everytime. But there are much better algorithms to solve this in a RDBMS that are much faster, but not very intuitive – doesn’t map to the objects very well.

Hal, I’m partly from the same dinosaur tribe that you are :) but I know the current trend is to have developers do as little DB developement as possible, mainly for the reason that you stated, that they would have to write in a second language. Personally I don’t consider that a tragedy, but I can see how that can become a problem. But thats what DBAs and DB developers are for, no? Although I hear that in the trend to separate the programmers from DB as much possible is making those positions obsolete….

In reference to the second comment – I see a few issue with that solution in most scenarios, but maybe you can clarify them: How does the XML repository store the data? In a text file? How do you maintain that, and what is the concurrent access problems you will see in that? I suppose you can store the whole blob in the DB as well in a text/large varchar field – but also, if its a category-sub catgeory hierarchy, how would you relate this to the rest of the data that its associated with? There will be no referencial integrity if its all in a blob….

I’b be curious to see more solutions to this problem in a non-RDBMS system….

]]> By: Dominique Rabeuf http://halwebguy.danziger.net/2010/02/25/the-nosql-movement-the-object-rdmbs-incompatability/comment-page-1/#comment-65 Dominique Rabeuf Thu, 25 Feb 2010 16:39:11 +0000 http://halwebguy.danziger.net/?p=162#comment-65 Your sample is based on use of PK/FK (Primary and Foreign keys). Mechanism largely used in RDBMS. XML Schema provide with key/keyref elements, analogous but most powerful than PK/FK in relational use case. With a simple XQuery and a very little recursive function one can retrieve nested hierarchic collections Your sample is based on use of PK/FK (Primary and Foreign keys).
Mechanism largely used in RDBMS.
XML Schema provide with key/keyref elements, analogous but most powerful than PK/FK in relational use case.
With a simple XQuery and a very little recursive function one can retrieve nested hierarchic collections

]]> By: Rob http://halwebguy.danziger.net/2010/02/25/the-nosql-movement-the-object-rdmbs-incompatability/comment-page-1/#comment-64 Rob Thu, 25 Feb 2010 15:22:32 +0000 http://halwebguy.danziger.net/?p=162#comment-64 Take a look at GT.M's natural hierarchical sparse persistent arrays. These give you two solutions to your problem. Firstly you can model many hierarchies to directly using n subscripts, but also take a look at the XML DOM example I've used in the M/Wire tutorial (http://www.mgateway.com/mwire.html) where you'll see the array structure can be used to very simply model the arbitrary level of tag nesting you get in an XML document. Is this the kind of NoSQL solution you have in mind? Take a look at GT.M’s natural hierarchical sparse persistent arrays. These give you two solutions to your problem. Firstly you can model many hierarchies to directly using n subscripts, but also take a look at the XML DOM example I’ve used in the M/Wire tutorial (http://www.mgateway.com/mwire.html) where you’ll see the array structure can be used to very simply model the arbitrary level of tag nesting you get in an XML document.

Is this the kind of NoSQL solution you have in mind?

]]>