S3 Usability and Use cases

Copied the question from Colin Baker in the tf_chat_farmers.

Hi Guys,

A bit of a broad question I know, but how reliable is our S3 storage solution now and is the setting up of it relatively bug-free?

Looking to onboard our first test client and want to give them a very clear idea of what to expect - I’d much rather under promise and over deliver.

1 Like

The S3 solution we have improves reliability over the reliability provided by the underlying hardware. Based on how you configure the s3 solution setup you can get from 50% to 100% improvement.

But this all depends again on your setup for example the following factors must be taking into account:

  • The data to parity shards ration (in other words how many nodes you can lose before you can’t re construct your files)
  • If you setup a TLog (transaction log) or not. This will protect you against metadata disk failure.
  • Did you setup multiple read-only heads or not.

The solution has proven to be “bug” free, but the measurement of reliability is way more complex than a yes/no answer.