The Vertica Database Forums

Hello Everybody,

How often do you run full backup on your site?
What is your recommendation for 3 nodes prod cluster 1.8 TB DB?
Are there any slowdown performance issues on database when you run full backup?


thanks
Elena

Elena,

The schedule for backups is really up to you but I can speak to the effect on performance. Generally, what you can expect is mostly around disk/network usage as the backup is moving a lot of files underneath. So if you are backing everything up to a local disk on one node then the files are pulled across the back-end network and written on the one node.

Where were you planning on backing up to?
What is the configuration of your private/back-end network?
What is the size of the raw and compressed data on each node?

Thank you for the reply
We are planning backup to the SAN disks.
We are planning to move DBs from Netezza to Vertica approximately 50 TB.
How I can find private/back-end network? We have some Vertica POC installed, I was not involved into.

How I can find the size of the raw and compressed data on each node?
Vertica size of company we just bought is small and we will move it to 10 Vertica cluster servers together with Dbs from Netezza:
dbadmin=> select sum(used_bytes) as total_size from v_monitor.storage_containers;
total_size
---------------
1735709185900
(1 row)

It was question on our project meeting:
1. Estimate disks size for the backup.
What approximately will be size for full backup for 50 TB data size.
2. How often we need to run full backup? Once per week /once per month?

thanks
Elena
Oracle/netezza DBA

Hi Elena!

>> 1. Estimate disks size for the backup. full audit go over all columns, summarize all sizeof(field + 1 byte delimiter).

>> 2. How often we need to run full backup? Once per week /once per month?
You are DBA - so it your decision.
There are no golden bullet/rule - it depends on many aspects:
- how often is data updated
- physical parameters
- ...

* If DB is large volume - so I shall prefer "per month".
* If my network is limited or I have to minimize network throwput - so I shall prefer "per month"
...

Thank you everyone for response
I agree Once per month is good start for the backups


thanks
Elena