Amazon S3 Glacier vs. MailShelf

If your organization has a lot of archival data that it needs to be backed up for safekeeping, where it won't be touched except in a dire emergency, Amazon Glacier is perfect. In our case, we had a client that generates many TB of video and photo data at annual events and wanted to retain ALL of it, pre- and post- edit for potential use in a future museum. Using the Snowball device, we were able to move hundreds of TB of existing media data that was previously housed on multiple Thunderbolt drives, external RAIDs, etc, in an organized manner, to Amazon Glacier. Then, we were able to setup CloudBerry Backup on their production computers to continually backup any new media that they generated during their annual events.

Incentivized

• Cheap storage of backup data.
• Can be used as a part of the entire suite of tools from Amazon, without requiring you to leave the familiar stack.

Incentivized

• Sometime due to slow drives there are operation failure noticed by us in testing
• Cost of restoring data is high and if you have regular restoring them it is not good option and slow as well.
• While we were setting up the system we took some support from AWS and in many cases their answers were not up to the mark.

Incentivized

It is difficult to delete the data as you have to wait for inventory and then bucket modification has to expire.

Incentivized

Since the rest of our infrastructure is in Amazon AWS, coding for sending data to Glacier just makes sense. The others are great as well, for their specific needs and uses, but having *another* third-party software to manage, be billed for, and learn/utilize can be costly in money and time.

Incentivized

• We seldom need to access our data in Glacier; this means that it is a fraction of the cost of S3, including the infrequent-access storage class.
• Transitioning data to Glacier is managed by AWS. We don't need our engineers to build or maintain log pipelines.
• Configuring lifecycle policies for S3 and Glacier is simple; it takes our engineers very little time, and there is little risk of errant configuration.

Incentivized