Backup and Maintenance Automation
Backing up ScienceOps protects you from potentially catastrophic data loss. While the chances of a catastrophic loss might not be very great, it's still strongly encouraged as a best-practice. Remember: Backing up is the only way to protect your data!
Additionally, backups of ScienceOps are useful for routine administrative purposes, such as copying an instance from one cluster to another, setting up additional failover servers, or archiving for posterity.
Disaster Recovery / Prevention
It's possible that at some point your ScienceOps system will experience downtime. There are many possible reasons for this: a data center outage, a user "fat-finger" mistake, application bugs, or some other act of God. Whatever the case, it's important that you have a well-planned backup and recovery strategy to protect against data loss and long outages.
ScienceOps provides 2 methods for avoiding downtime: Failover and Restoration. For high availability use-cases, it's recommended you use the Failover strategy. For other non-production use-cases, the Restoration method can be a sufficient (and typically cheaper) alternative.
|Strategy||Recovery Time||Work loss exposure||Recover to point in time?|
|Failover||Immediate||Typically none||Moment in time immediately preceding failure. This will not include models deploying during (or causing) failure.|
|Restoration||15 minutes||Changes since most recent backup||Yes, though you will be limited to restoring from the most recent backup.|
Performing full backups of ScienceOps is easy to do and can be configured and scheduled according to your company's IT policies. For more information on performing backups of ScienceOps, please read the Backup docs.
You can find more detailed information about ScienceOps backup and maintenance automation by clicking on one of the sub-topics below: