In today's IT world, technology lets people "set it and forget it" in many areas that once required constant attention. Cloud backups for public cloud and Infrastructure-as-a-Service (IaaS) are perfect examples. During a cloud deployment, it's easy to pick a backup option without much thought, and this can cause problems later when it really counts. Users need to understand their cloud backups, ensure they have the right solution, and know what to do if disaster strikes. Before we move on, let's take a step back and review the current types of backups typically used. The backup method provided by the vendor - or managed by your company with another solution - can affect network and storage costs. It can also impact your Recovery Point Objective (RPO).
Brief Summary of Popular Backup Techniques
Differential Incremental - Often called incremental backups, these start with a full backup. Then, only data that changes each desired period gets copied. If you were on a weekly rotation, you would end up with a full backup, plus 6 more having only the changes from each day. This approach saves storage and bandwidth, but if an "incremental" is damaged, you can only do a full restore to the day before the corrupt backup. Cumulative Incremental - In this model an initial full backup is also created. In the weekly example again, each incremental backup would have changes for that day plus everything that changed in the days before it going back to the original full backup. This means you use more resources but have greater reliability. Software overhead to recompile both types of incremental backup methods goes up over time and restores become increasingly slower since you need each incremental backup along with the full backup for a restore. Progressive Virtual Full Backups (PVF) - This is a more sophisticated approach. An initial full backup is taken and a rotation period's worth of cumulative incremental backups are created. Once the rotation period ends, the original full backup and incrementals are re-compiled into a new updated version of the entire dataset. After that, data that changes during the interval period (daily for example) will continually be recompiled to create a new, up-to-date version of itself. This ensures the original data is preserved, and changes are accurately captured with fewer resources. To learn more, read PVF Backups: The Best of Most Worlds.
Many cloud providers offer "snapshots" as their backup method. Snapshots copy Virtual Machine (VM) disks at regular intervals and can be used revert VMs back to a point in time, or to clone VMs onto new instances. Although a valuable tool, snapshots have limitations and may not be adequate for all purposes. If your vendor uses these, be sure to understand if they will fully meet your needs. For more information, read "Backing up Your Cloud - Are Snapshots Really Backups?"
Effective Steps for Managing Cloud Backups
1) Know What You're Getting If you subscribe to a vendor's cloud backups or they are included as part of your service, be sure to learn exactly how they work and how they will be managed. Pay attention to details such as frequency, recovery times, exclusion features, and so on. And, be sure to understand if the backups are at the file level, system level, or both. 2) Ensure Adequate Coverage Be clear about your recovery objectives and make sure you have cloud backups that will meet your needs for both the data and applications you rely on. Some just assume that the vendor's offering will suffice - this may not be the case. Be sure your data will be backed up often enough and can be restored quickly enough to meet your goals. When core OS and/or application changes are being made, it's wise to do a full system backup just prior to those changes so you can roll back if need be. Be sure your backup frequency is short enough to accommodate this, or you have the capability to do a manual backup when needed. 3) Document the Process Learn what the exact steps will be when you need to recover. Usually, this is fairly simple if everything has been set up correctly, but take the time to understand it clearly and document how the backup and recovery process works. This can be transcribed from vendor FAQs and other sources, but it should be re-written in your IT policy manual. Make sure this information is readily accessible for those who will need it. In your documentation, you should also include the internal procedures and staff contingencies for your backups and recoveries. 4) Periodically Audit Cloud environments can grow and change, and they usually do just that. Be sure to include your cloud backups on the list of other critical IT audits your company maintains. If the requirements have evolved over time, make the necessary adjustments to your cloud backups. 5) Test and Test Again Finally, don't assume your cloud backups will work as advertised. Make a point of testing them early and then on a periodic basis after that. If there is a significant change to your infrastructure, data or applications, be sure to do a test shortly afterwards. Cloud backups, like most IT Disaster Recovery (DR) programs, are like insurance policies. They can become easy to forget until you need it in a critical situation. If you follow these steps, it will go a long way to ensuring that when that moment of absolute panic takes over, you will recover your organization's valuable data with precision and ease.