Re: MapReduce?

From:

=?ISO-8859-1?Q?Arne_Vajh=F8j?= <arne@vajhoej.dk>

Newsgroups:

comp.lang.java.programmer

Date:

Wed, 05 Aug 2009 18:07:04 -0400

Message-ID:

<4a7a0282$0$296$14726298@news.sunsite.dk>

Roedy Green wrote:

On Tue, 4 Aug 2009 22:03:47 -0700 (PDT), jlc488 <jlc488@gmail.com>
wrote, quoted or indirectly quoted someone who said :

Someone told me about the Hadoop Map/Reduce. Just wondering, Is It the
right solution to use according to this kind of situation?

Usually these sorts of tools have a free trial.

Hadoop is open source.

Then you have to decide if the tool is sufficiently powerful to do
what you need. If it looks as if it will run out of steam, you might
as well right custom code from the start.

Conversely, if it looks vastly overpowered and expensive, chances are
you can whip up something custom more cheaply.

It is free.

One way to get a lot of fancy searching/consolidating power is to
write a module to export your logs into SQL. This also offload the
problem of clever efficient algorithms on the SQL manufacturer.

For a problem that really needs the map reduce approach, then
a relational database and SQL is hopeless.

Arne