Amazon EBS Snapshot is an easy to use feature that backs up your data on an Amazon EBS volume. This session covers filesystem selection (XFS, ext*, etc.), quiescing of the filesystem, tagging of snapshots, and life-cycle management of snapshots. In this session, we introduce two new OSS tools. One tool manages arrays of snapshots using tagging, making it easier to snapshot and recover a RAID array of Amazon EBS volumes; another tool manages snapshots of root volumes across an AWS account, and automatically snapshotting and applying lifecycle management to root volume snapshots.
2. EBS
• Network-attached block devices for Amazon EC2
instances
• Protected, durable within an AZ
– AFR between 0.1% – 0.5%
• Standard and PIOPS
– PIOPS is for databases!
3. Snapshots
• More durable than an Amazon EBS volume
– Stored in Amazon S3
• Differential (space-efficient)
– First snapshot is a clone
– Pay for only what you use
• AZ-independent
– Clone into any AZ
• Can be copied efficiently across regions
6. Managing Thousands of Snapshots Isn’t Easy
• A snapshot should have a lifecycle
–
–
–
–
Grandfather-father-son
x versions
x dailies, y weeklies, z monthlies
A lifecycle might span regions
• Snapshots of arrays
7. Introducing arche
• arche
– https://github.com/AWSLabs/arche
• OSS, built and maintained by AWS
– Accepting pull requests from the community
8. arche
• Two primary pieces
– arche takes snapshots and manages lifecycles
• Takes snapshots based on a SWF schedule
– Tags snapshots
• Deletes snapshots based on a policy
– Limited to number of days now
– pierus is a CLI for restoring snapshots.
9. arche Uses Tags
• on volumes
KEY
VALUE
arche
1
retention
<int> (days)
array
<str>
dest-region
<str>
10. arche Uses a Lot of Tags
• on snapshots
KEY
VALUE
arche
1
src_instance
<str>
src_device
<str>
expire_date
date
array
<str>
src-region
<str>