Have come across references to this a few times recently & am curious enough to ask the development community whether the NoSQL movement holds any relevance for Koha's future. According to that font of universal wisdom, Wikipedia, "Typical modern relational databases have shown poor performance on data-intensive applications including indexing a large number of documents [= any ILS?], serving pages on high-traffic websites and delivering streaming media. They can be efficient only when they are tuned either for small but frequent read/write transactions or for large batch transactions with rare write accesses, while there are demands for the data stores capable of heavy workloads with frequent updates." Thoughts? How wedded is Koha to SQL in the long run? Cheers, Cab Vinton, Director Sanbornton Public Library Sanbornton, NH
On Mon, Jun 14, 2010 at 11:37 AM, Cab Vinton <bibliwho@gmail.com> wrote:
Have come across references to this a few times recently & am curious enough to ask the development community whether the NoSQL movement holds any relevance for Koha's future.
According to that font of universal wisdom, Wikipedia, "Typical modern relational databases have shown poor performance on data-intensive applications including indexing a large number of documents [= any ILS?], serving pages on high-traffic websites and delivering streaming media. They can be efficient only when they are tuned either for small but frequent read/write transactions or for large batch transactions with rare write accesses, while there are demands for the data stores capable of heavy workloads with frequent updates."
Thoughts? How wedded is Koha to SQL in the long run?
Very wedded for now, and logically so, imho. We don't use SQL for indexing large numbers of (MARCXML) documents. We use zebra for that. Zebra, despite whatever complaints we might have about ease of configuration, certainly is highly scalable. Our documents are still pretty small by rest-of-the-world standards, though I could see this question coming back if features around digital repository functionality are submitted. --Joe
Le 14/06/2010 17:53, Joe Atzberger a écrit :
Thoughts? How wedded is Koha to SQL in the long run?
Very wedded for now, and logically so, imho.
++
We don't use SQL for indexing large numbers of (MARCXML) documents. We use zebra for that. Zebra, despite whatever complaints we might have about ease of configuration, certainly is highly scalable.
++
Our documents are still pretty small by rest-of-the-world standards, though I could see this question coming back if features around digital repository functionality are submitted.
++ no need to say more, Jo is 100% right ! -- Paul POULAIN http://www.biblibre.com Expert en Logiciels Libres pour l'info-doc Tel : (33) 4 91 81 35 08
On ma, 2010-06-14 at 11:53 -0400, Joe Atzberger wrote:
Zebra, despite whatever complaints we might have about ease of configuration [- - -]
*cough*http://debian.koha-community.org/*cough* :) (This is what distributions of free software operating systems, e.g., Debian, are meant for: to take the pain out of installing powerful, complicated software. For whatever reason, most people who write the kinds of sophisticated software that makes you realize you live in the future and reduces near-term sci-fi authors to tears[1] are not so good at making it easy to install, configure, upgrade, and otherwise manage their software. That's where distributions help. To build something like Debian is not just a matter of having a fast machine to compile a lot of source code provided by other people. You also have to put in a lot of hard work to make everything work out of the box, and also get everything to work together.) [1] http://www.antipope.org/charlie/blog-static/2008/03/blindsided-by-the-future...
Cab Vinton wrote:
According to that font of universal wisdom, Wikipedia, "Typical modern relational databases have shown poor performance on data-intensive applications including indexing a large number of documents [= any ILS?], serving pages on high-traffic websites and delivering streaming media. They can be efficient only when they are tuned either for small but frequent read/write transactions or for large batch transactions with rare write accesses, while there are demands for the data stores capable of heavy workloads with frequent updates."
I checked the date and it isn't April 1st! The wikipedia article indicates that RDB performance can start to flag when doing joins on databases that are in the TB and PB sizes (tera- and peta-byte). This is not the case for Koha, yet! Nor for the foreseeable future.
Thoughts? How wedded is Koha to SQL in the long run?
Quite wedded in the data model. Actually, Koha is quite wedded to MySQL specifically. It would be a lot of work to convert the existing code to use a different SQL database, let aone a different style of data store. NoSQL is a storm in a teacup unless you are a Facebook type application dealing with truly enormous datasets. cheers rickw -- _________________________________ Rick Welykochy || Praxis Services "Most folk are about as happy as they make up their minds to be." -- anon
OK, thanks, guys. I suspected as much based on not having seen any discussion of this before, but good to have confirmation from folks who actually know something :-) Curiosity satisfied! Cheers, Cab Vinton, Director Sanbornton Public Library Sanbornton, NH
participants (5)
-
Cab Vinton -
Joe Atzberger -
Lars Wirzenius -
Paul Poulain -
Rick Welykochy