Skip to main content

Time taken to complete Copycluster in Vertica?

  • March 12, 2019
  • 3 replies
  • 4 views

Nithesh
Forum|alt.badge.img+1

We have a 10 node cluster with about 25TB of data. We want to create a DR with same configuration and use copycluster to transfer the data.

How long would it take for the copycluster to complete over 10Gbps network?

In the Vertica Documentation its mentioned for Bulk Loads:

Vertica can load about 30-50GB/hour/node for a 1-Ksafe projection design

Does the same apply for CopyCluster too? Is this irrespective of the Network bandwidth?

3 replies

Sharon_Cutter
Forum|alt.badge.img
  • Participating Frequently
  • March 12, 2019

Based on copyclusters that I've done, I will guess that 25TB over 10 nodes will take about a day for the first full copy. After that, subsequent copyclusters will be incremental with only the new or modified storage containers getting copied over, so it depends on the churn, but likely a couple of hours. All copyclusters also need to do a "database snapshot" as the first step, the duration of which depends on the number of objects in the database.


Nithesh
Forum|alt.badge.img+1
  • Author
  • Participating Frequently
  • March 13, 2019

Hi Sharon,

Thanks for your answer, the information i gave might have been misleading as i didn't clearly specify the data size, the Data size on disk is 25TB and raw datasize is 54TB.


Sharon_Cutter
Forum|alt.badge.img
  • Participating Frequently
  • March 13, 2019

Hi Nitesh - I based my reply on the assumption that it's 25TB compressed, so my estimate stands :-)