hawq catalog is on master and it uses local file system. For hawq, catalog only sits on master. GPDB is different. catalog is on all the nodes.
you can reference paper here for more details: Lei Chang et al: HAWQ: a massively parallel processing SQL engine in hadoop <https://github.com/changleicn/publications/raw/master/hawq-sigmod-2014.pdf>. SIGMOD Conference 2014: 1223-1234 On Wed, Oct 28, 2015 at 5:08 PM, sequoiadb <[email protected]> wrote: > where is hawq's catalog > Is it using master’s local file system or if it’s on hdfs? > I’m not very familiar with GPDB or HAWQ, each segment has its own catalog > ( including local tables, indexes, stored procedures, etc… ) or all sitting > at master node? Is there any detailed architecture document I can read? > Thx >
