Facebook: we need big data loads to work every time (not just most of the time), which is why we built Presto Raptor.

Facebook: HDFS just isn’t up to producing consistent predictable loads of very large data sets.

Facebook uses Presto to scan trillions of rows and several Petabytes every day.

Having a great evening meeting developers at Boston’s first Presto big data meetup, sponsored by Teradata.

Summary of Microsoft event: forget that post-PC stuff; phones, tablets and notebooks really just want to be PCs.

Assuming the thermal design is decent, does Surface Book mean I can now make a 10-node Hadoop cluster that’s only a couple inches thick?

Sony, Asus, Lenovo, Dell and Toshiba are all yelling “Hey we’ve been making Windows convertibles for years!”

The Apple Macbook design team is sitting in a room shouting “Bring it!” at the Surface Book guys.