Not too bad, but a few things that I noticed were off (and to clarify: I've worked on Hadoop/HBase since 2011): 1. Flume isn't really a Google solution, is it?

Not too bad, but a few things that I noticed were off (and to clarify: I've worked on Hadoop/HBase since 2011): 1. Flume isn't really a Google solution, is it? Sounds like Google is just incorporating Apache Flume into Cloud Workflow. 2. Even if Google abandoned MapReduce a while back (Urs' declaration about the same at IO seemed a bit ambivalent), IMHO Hadoop and its ecosystem are here to stay at least for a bit longer. True, MapReduce isn't perfect, and the Hadoop implementation far less, but (again IMHO) there really isn't anything to replace it at the data scale at which these technologies operate. I mean, there's a reason that Facebook and large non-Google companies still use the Hadoop ecosystem, yeah? :)