NoSQL presentation

209 views
Skip to first unread message

repeatt itself

unread,
Nov 12, 2010, 9:57:15 AM11/12/10
to NOSQL
I'm planning a presentation on NoSQL for an upcoming event.

I had the idea to break it up into the following sections:
- What does NoSQL means (in relation to relational DBs :P)
- What's the use
- Who uses it and for what
- What's the future, and alternatives that may arise

Any help, links, comments, insults, etc. are welcomed.
Thanks.

Cliff Moon

unread,
Nov 12, 2010, 1:19:21 PM11/12/10
to nosql-di...@googlegroups.com
NoSQL is a bullshit marketing term coined to insight ire in the heart of
DBA's. Each individual database under that umbrella term was written to
solve a different problem and they don't really belong in the same
category. For instance, Cassandra was written to solve inbox search at
Facebook, Riak was a backend for Basho's earlier forays into CMS
software and CouchDB was written so that Damien could get a job he
didn't hate.

If one were to attempt a classification system I'd imagine it would end
up something like a 3 dimensional cube. Distribution Model vs. Data
Model vs. Disk Data Structure. For instance cassandra would be
classified as (token ring dynamo, column store, log structure merge
tree), riak would be (vnode dynamo, key/value, pluggable) and couch
would be (offline replication, document w/ incremental mr, btree). The
rest I'll leave as an exercise for the reader.

As for what's the use it basically comes down to 2 things: if you have a
ton of data then you must use one of the distributed datastores and your
requirements for how you want to query that data will most likely drive
the choice of which database to use. If, however, you do not have a ton
of data then the choice basically comes down to personal preference.

edlich

unread,
Nov 13, 2010, 4:17:38 AM11/13/10
to NOSQL
To answer these questions I have started to gather material since
summer 2009 here:
http://www.nosql-database.org/links.html
There are many articles answering your questions (e.g. the Gupta
ones).

This great site from alex also has more info you can ever read ;-)
http://nosql.mypopescu.com/

Best
Stefan Edlich

Cliff Moon

unread,
Nov 13, 2010, 4:33:08 AM11/13/10
to nosql-di...@googlegroups.com
nosql-databases.org has the editorial veracity of a google alert set to
"nosql". Either fix your shit or stop pointing people at it. Also: way
to yoink the nginx logo for your masthead. The mynosql blog isn't much
better. I'm surprised that Alex's cmd-c and cmd-v keys haven't worn out
by now.

Truth is, if you want real information you need to be in IRC with the
people writing these systems. They don't spend their time collecting
links and writing 2 line blog posts. Those are activities for the
takers. The makers are writing code and if you want the real story you
talk to them.

Bill de hÓra

unread,
Nov 13, 2010, 7:54:05 AM11/13/10
to nosql-di...@googlegroups.com
On Fri, 2010-11-12 at 10:19 -0800, Cliff Moon wrote:
If one were to attempt a classification system I'd imagine it would end 
up something like a 3 dimensional cube.  Distribution Model vs. Data 
Model vs. Disk Data Structure.  

I agree with Cliff. I always use the first two and calling out the disk structure is a good idea I wish I'd thought of.

I would maybe add licensing, as licensing can directly impact how you scale/organise your system.



If, however, you do not have a ton 
of data then the choice basically comes down to personal preference.

Also what you can operate in production.

One the thing to be aware of is with this crop of systems is there is a fair chance you'll be the first person to find that bug especially. Unlike RDBMSes where bugs are fewer and generally found by someone else.

Bill

dwight_mongodb

unread,
Nov 13, 2010, 1:55:12 PM11/13/10
to NOSQL
The name is a misnomer, but I think we are stuck with it now. I
usually describe the space as follows:

- NoSQL means non-relational next generation operational datastores
and databases

What the products have in common:

- no joins (joins are hard to scale horizontally)
- light transactional semantics (complex transactions are hard to
scale horizontally)

What varies from product to product in the space:

- data model (Document oriented, Columnar, Key/value, ...)
- consistency model
- data distribution / partitioning model (e.g. bigtable style
partitioning vs. consistent hashing)

b

unread,
Nov 13, 2010, 2:06:23 PM11/13/10
to NOSQL
On Nov 13, 10:55 am, dwight_mongodb <dwi...@exch.alleycorp.com> wrote:
> The name is a misnomer, but I think we are stuck with it now.  I
> usually describe the space as follows:
>
> - NoSQL means non-relational next generation operational datastores
> and databases
>

I usually describe the space as follows:

- NoSQL means non-relational next generation operational thetan
datastores and databases and data processing and enthusiasm and a gosh-
darn-it-we-don't-need-to-understand-this-because-our-users-don't-
either can do attitude.

I'd repeat Cliff's list of parameters, but I see you've already done
that.


b

William Newport

unread,
Nov 13, 2010, 8:12:42 PM11/13/10
to nosql-di...@googlegroups.com
I'll keep on the right side of cliff then ;) I can take some questions
if anyone is ever interested, I am the lead on IBM web sphere extreme
scale, a partitioned distributed hash map with partition scoped
transactions. I support ca and ap for wxs.

Sent from my iPad

> --
> You received this message because you are subscribed to the Google Groups "NOSQL" group.
> To post to this group, send email to nosql-di...@googlegroups.com.
> To unsubscribe from this group, send email to nosql-discussi...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/nosql-discussion?hl=en.
>

Alex Popescu

unread,
Nov 15, 2010, 3:00:09 AM11/15/10
to NOSQL
On Nov 13, 11:33 am, Cliff Moon <moonpolys...@gmail.com> wrote:
> nosql-databases.org has the editorial veracity of a google alert set to
> "nosql".  Either fix your shit or stop pointing people at it.  Also: way
> to yoink the nginx logo for your masthead.  The mynosql blog isn't much
> better.  I'm surprised that Alex's cmd-c and cmd-v keys haven't worn out
> by now.
>
> Truth is, if you want real information you need to be in IRC with the
> people writing these systems.  They don't spend their time collecting
> links and writing 2 line blog posts.  Those are activities for the
> takers.  The makers are writing code and if you want the real story you
> talk to them.
>

As far as I can tell:

- not everyone is creating a new persistence engine (are you?)
- not everyone is using all new persistence engines (are you?)
- not everyone is already informed about all new persistence engines
out there (are you?)

These are usually the reasons for having such websites/blogs out
there. So, I guess the only way I could understand your comment is if
you meant something like: "I don't like the way he's doing it or it
doesn't fit my needs of finding out about these systems" in which case
there would be nothing bad about either your opinion nor my/Erich's
approach.

bests,

:- alex

repeatt itself

unread,
Nov 16, 2010, 11:05:13 AM11/16/10
to NOSQL
:)

Thanks for all your help.
It's nice to see the Interweblinkmediainfo still has a side that likes
to laugh of itself.

Angel Java Lopez

unread,
Nov 16, 2010, 12:19:09 PM11/16/10
to nosql-di...@googlegroups.com
Hi people!

Links I used in one of my presentations:
http://ajlopez.wordpress.com/2010/07/19/nosql-resources/

Angel "Java" Lopez
http://www.ajlopez.com
http://twitter.com/ajlopez


Achim 'ahzf' Friedland

unread,
Nov 20, 2010, 10:14:58 AM11/20/10
to NOSQL
On 13 Nov., 10:33, Cliff Moon <moonpolys...@gmail.com> wrote:
> Truth is, if you want real information you need to be in IRC with the
> people writing these systems.  They don't spend their time collecting
> links and writing 2 line blog posts.

There is only one problem: This approach does not scale and NoSQL is a
lot about scaling ;).
So why not use some Alex- or Stefan-Caches to improve the read
performance?
May be their latency is even worse than using HTTP proxies, but it
might still be a much
better solution than having overloaded NoSQL developers.

Cheers,
achim
Reply all
Reply to author
Forward
0 new messages