May 2009
6 posts
“The app slaves will monitor the health of the app master and if the master...”
– InfoQ: Flex: Engine Yard’s New Cloud Offering
Interesting approach to load-balancing instead of the usual “1 master + 1 failover-master + N slaves” setup.
How are big, enterprise data warehouse products... →
The paper [pdf] documents the kinds of things that are possible with expensive data warehouse products like GreenPlum including how they’ve integrated map-reduce (specifically how it’s being used at Fox Interactive Media). Warning: the paper sounds more like a GreenPlum whitepaper than UC Berkeley CS research.
Writing a template compiler for Erlang →
Well written article by one of the authors of ErlyDTL walking you through the steps involved in writing such a template language compiler for Erlang.
“Even with the new more aggressive shuffle, minimizing the number of transfers...”
– Hadoop - Petabyte sort
Looks like the bottleneck and optimizations happened this time in the shuffle phase of Hadoop M/R.
“In 2007, Amazon opened up its global fulfillment network to other merchants,...”
– Why Jeff Bezos Is Obsessed With Waste
CAS - good idea or not? →
Discussion on the redis mailing list about whether check-and-swap (CAS) operation is a good idea in a network-based KV-store like redis. memcached has one.