Pitfalls of using Snowflake Time Travel as your database backup strategy

Snowflake Time Travel enables accessing historical data at any point within a defined period. You can go back up to 1 day in a standard edition and up to 90 days if you are on an enterprise edition. It's a powerful feature to see how the data looked at an instant of time. However, it cannot be a replacement for your database backup strategy. Here's the reason:

If you alter table to drop few columns, all the time travel data for those columns are lost. You cannot historically navigate to look at the data in those columns

Time travel works as long as the structure is not altered. So be mindful of this tiny drawback before you consider your data warehouse backup strategy. It's always good to take a clone instead of relying on Time-travel.

What do you use for backups in your organization?

nVector

posted on 20 Mar 19

Enjoy great content like this and a lot more !

Signup for a free account to write a post / comment / upvote posts. Its simple and takes less than 5 seconds




AzharuddinK16-May-19

We did Snowflake Migration recently and ran the DDL commands for creation of the table in the old instance itself instead of new instance. And we lost all the data. Time Travel also didn't help in this case. What is your startegy for backup of your databases. ?

nVector16-May-19

True, once the structure changes / overwritten, we lose the time travel capabilities. Its good to take daily clones and monthly file exports to s3 buckets would be an additional layer of protection