Skip to main content
Question

Should I keep Vertica and HDFS on separate nodes?

  • April 1, 2020
  • 1 reply
  • 7 views

rjs_docs

In the presentation, you mentioned that Vertica should be separated from other services (tools, etc.). What about HDFS?
@rbankula @bat

From Optimizing Query Performance and Resource Pool Tuning.

1 reply

mcellio
  • Participating Frequently
  • April 1, 2020

If you're using Vertica with HDFS storage locations, you can either run separate clusters (treat HDFS like any other shared file storage) or co-locate Vertica nodes on your HDFS nodes. Co-locating nodes allows Vertica to take advantage of data locality, but requires that you carve out space from your HDFS nodes for Vertica. See the Cluster Layout section in the Hadoop Integration documentation for more information.