Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.databases > #340 > unrolled thread

Reaction against RDBMS?

Back to article view | Back to comp.databases

Contents

  Reaction against RDBMS? "Aaron W.  Hsu" <arcfide@sacrideo.us> - 2013-01-21 15:11 -0500
    Re: Reaction against RDBMS? Robert Klemme <shortcutter@googlemail.com> - 2013-01-21 21:40 +0100
      Re: Reaction against RDBMS? "Aaron W.  Hsu" <arcfide@sacrideo.us> - 2013-01-21 23:54 -0500
        Re: Reaction against RDBMS? Robert Klemme <shortcutter@googlemail.com> - 2013-01-22 23:17 +0100
          Re: Reaction against RDBMS? "Aaron W.  Hsu" <arcfide@sacrideo.us> - 2013-01-22 21:05 -0500
            Re: Reaction against RDBMS? Robert Klemme <shortcutter@googlemail.com> - 2013-01-25 21:27 +0100

#340 — Reaction against RDBMS?

Lately there has been a lot of talk against RDBM systems. The NoSQL type 
stuff seems to be getting really popular in some circles or has been for a 
few years. I feel like this is an over-reaction and perhaps misinformed. 
Being a casual user of RDBM systems in my own programming, and understanding 
a few things about Database management, I fail to see the attraction of 
these other Database replacements. 

Is there a real, fundamental justification for this movement, or is it just 
another fad that comes and goes around computer science in general? 

-- 
Aaron W. Hsu | arcfide@sacrideo.us | http://www.sacrideo.us
לֵ֤ב חֲכָמִים֙ בְּבֵ֣ית אֵ֔בֶל וְלֵ֥ב כְּסִילִ֖ים בְּבֵ֥ית שִׂמְחָֽה׃

[toc] | [next] | [standalone]

#341

On 01/21/2013 09:11 PM, Aaron W. Hsu wrote:
> Lately there has been a lot of talk against RDBM systems. The NoSQL type
> stuff seems to be getting really popular in some circles or has been for a
> few years. I feel like this is an over-reaction and perhaps misinformed.
> Being a casual user of RDBM systems in my own programming, and understanding
> a few things about Database management, I fail to see the attraction of
> these other Database replacements.
>
> Is there a real, fundamental justification for this movement, or is it just
> another fad that comes and goes around computer science in general?

Yes and yes.

	robert

[toc] | [prev] | [next] | [standalone]

#342

Robert Klemme wrote:

> Yes and yes.

Would you mind articulating those reasons? I'd like to get a broader 
perspective on the issue. Most of what I find is usually from someone saying 
the equivalent of, "RDBMS stinks, of course X is better."

-- 
Aaron W. Hsu | arcfide@sacrideo.us | http://www.sacrideo.us
לֵ֤ב חֲכָמִים֙ בְּבֵ֣ית אֵ֔בֶל וְלֵ֥ב כְּסִילִ֖ים בְּבֵ֥ית שִׂמְחָֽה׃

[toc] | [prev] | [next] | [standalone]

#343

On 01/22/2013 05:54 AM, Aaron W. Hsu wrote:
> Robert Klemme wrote:
>
>> Yes and yes.
>
> Would you mind articulating those reasons? I'd like to get a broader
> perspective on the issue. Most of what I find is usually from someone saying
> the equivalent of, "RDBMS stinks, of course X is better."

Yes: not all applications need the strict consistency rules (ACID) which 
typical RDBMS implementations of TX handling provide and not all 
applications need immediate access to all basic data fields (i.e. 
retrieval of complete documents is OK for many applications or even more 
efficient).  By abandoning these more performance can be achieved and 
replication becomes easier just to name .  So NoSQL database have real 
benefits for particular use cases (same as RDBMS have as well - just 
other use cases).

Yes: they are hyped which you can see any time when looking in the 
media.  When people start implementing RDBMS like TX handling on top of 
NoSQL databases "by hand" you should be suspicious.

Cheers

	robert

[toc] | [prev] | [next] | [standalone]

#344

Robert Klemme wrote:

> Yes: not all applications need the strict consistency rules (ACID) which
> typical RDBMS implementations of TX handling provide and not all
> applications need immediate access to all basic data fields (i.e.
> retrieval of complete documents is OK for many applications or even more
> efficient).  By abandoning these more performance can be achieved and
> replication becomes easier just to name .  So NoSQL database have real
> benefits for particular use cases (same as RDBMS have as well - just
> other use cases).

Okay, thanks for the analysis. How would you compare component files 
(datatypes which are implicitly serialized to the filesystem but which 
appear just as regular datatypes in the language once loaded), columnar 
databases (such as VectorNova's RDBMS), and NoSQL sort of databases? I also 
recall some databases going through a lot of effort to make XML handling go 
hand-in-hand with the Relational model. I tend to think of XML objects as 
complete, structured documents, so I immediately thought of that when you 
mentioned the whole document retrieval method. Do you think this is a 
misguided idea (mixing XML with the relational model)? Do you think it's a 
completely different use case than what NoSQL is good for?

-- 
Aaron W. Hsu | arcfide@sacrideo.us | http://www.sacrideo.us
לֵ֤ב חֲכָמִים֙ בְּבֵ֣ית אֵ֔בֶל וְלֵ֥ב כְּסִילִ֖ים בְּבֵ֥ית שִׂמְחָֽה׃

[toc] | [prev] | [next] | [standalone]

#346

On 01/23/2013 03:05 AM, Aaron W. Hsu wrote:
> Robert Klemme wrote:
>
>> Yes: not all applications need the strict consistency rules (ACID) which
>> typical RDBMS implementations of TX handling provide and not all
>> applications need immediate access to all basic data fields (i.e.
>> retrieval of complete documents is OK for many applications or even more
>> efficient).  By abandoning these more performance can be achieved and
>> replication becomes easier just to name .  So NoSQL database have real
>> benefits for particular use cases (same as RDBMS have as well - just
>> other use cases).
>
> Okay, thanks for the analysis. How would you compare component files
> (datatypes which are implicitly serialized to the filesystem but which
> appear just as regular datatypes in the language once loaded), columnar
> databases (such as VectorNova's RDBMS), and NoSQL sort of databases?

I don't have enough experience with all of these to make an informed 
comparison.

> I also
> recall some databases going through a lot of effort to make XML handling go
> hand-in-hand with the Relational model. I tend to think of XML objects as
> complete, structured documents, so I immediately thought of that when you
> mentioned the whole document retrieval method. Do you think this is a
> misguided idea (mixing XML with the relational model)? Do you think it's a
> completely different use case than what NoSQL is good for?

Well, as you said: RDBMS with XML support can really do both - tables 
and structured documents.  I haven't seriously worked with XML feature 
in RDBMS to judge fitness of this approach.  It does seem to be a 
general approach to include non typical data types into RDBMS so there 
seems to be some need for this in the market - or at least vendors 
believe there is.

Kind regards

	robert

[toc] | [prev] | [standalone]

Back to top | Article view | comp.databases

csiph-web