Hacker News Reader

Comment by frisco | Hacker News Reader

I don't have a good reason to believe that this is real, but if it is the most surprising part to me is the "mapreduce" rule definition in there. As far as I know the only group with a C++ mapreduce implementation called "mapreduce" that also uses protocol buffers (the "proto:" block is protocol buffers) is Google. This seems to say to me that the NSA is using a Google implementation of map reduce. That can't possibly be right, can it?

View on HN

Well, if the NSA were among the group of organizations with a C++ mapreduce implementation (developed in-house) the code probably wouldn't be on github or otherwise divulged to the public... seems like a rather large leap of logic to assume they are using Google's code.

They're not using the Google implementation; it looks like they rolled their own.

Your profile says you work at Google... are you saying that the code snippet here is inconsistent with Google's mapreduce?

I used to work at Google - this looks absolutely nothing like a Google MapReduce specification or code would look like.

Not a Google employee - but the linked file looks more like a simplified DSL for analysts.

That's because they're using a derivative of Accumulo (which was developed internally at the Agency).

Interesting video about them using openstack https://www.youtube.com/watch?v=NgahKksMZis

While mapreduce isn't my area of expertise, my guess is it's probably easier for them to use mapreduce/ and Hadoop than invent their own wheel in some cases.

Reply on news.ycombinator.com