[rabbitmq-discuss] Riak <-> RabbitMQ integration

[Jon Brisbin]

> >> If you are pushing around binary, what's the method of providing
> >> an
> >> identity to Riak via a message?
> > You can either let it use the exchange name as the bucket and the
> > routing key as
> 
> the Riak key, or you can specify them in a message header
> (X-Riak-Bucket and X-Riak-Key respectively).
> 
> 
> which rabbit tag do "X-Riak-Bucket" and "X-Riak-Key" these fit
> inside?


If using the Java client, you'd pass in arbitrary headers using the "AMQP.BasicProperties.Builder.headers(Map)" method: http://bit.ly/A4CWu3

In Ruby, you'd pass them on the publish: "exchange.publish(msg, :headers => { "X-Riak-Bucket" => "foo", "X-Riak-Key" => "fookey" })"


> > Last time I explored this BerkeleyDB was the solution...
> > now I don't know the space well enough to form a solid opinion.
> 
> 
> It has been a while, then! Things are much different in some ways and
> not so much in others. How's that for an informative answer? ;)
> 
> really really helpful. Once there was "internet time", are you
> measuring in NOSQL-time? In normal time, not that long. ;-)


Yes, NoSQL time is like dog years, only in months. So 7 months in NoSQL time is like 7 years anywhere else! :)

That said, there's lots and lots of things that haven't changed in 30 years. When you get down to it, it's still a key/value lookup and some things are very similar to what we were doing many years ago writing COBOL programs on CICS/mainframes. A few years ago, I worked pretty closely with some RPG programmers (AS/400) and they were pretty frustrated with our boss, who was trying to get them to learn to embed SQL into their programs (which they really hated...can't say as I blamed them...it was much uglier and didn't work as well as using a "set lower limit" and CHAIN) and just about the time they're starting to get the hang of that, I started talking about using NoSQL in our apps and they rightly recognized that that's exactly what they'd been doing for decades!

The new kids think this stuff is all flashy and fancy and uber-modern and us old fogeys (that is, those of us just barely over 35 ;) see this as somewhat of a cyclic return to a simpler and more effective way to handle data that isn't altogether different from what it was like when we first started writing PC apps using FoxPro and dBase. :)

Thanks! 

Jon Brisbin 
http://about.me/jonbrisbin