Every sysadmin knows that the classic way to set up a backup strategy is to schedule different types of copies (e.g. full, incremental differential) at specific times of the day, week and month. The purpose of this is to create a regular loop of daily, weekly and monthly backups according to the desired RPO and RTO and depending on the criticality and the time required to produce a copy of the data.

This is okay, but we believe that a much more dynamic and efficient approach could better meet modern backup needs. AiRE changes the point of view from a calendar-oriented strategy to a vision always starting from now. Starting from now allows you to set different backup frequencies for the last few hours, or minutes, rather than for 10-year old data or older.

In this scenario, we must stop referring to backup copies as “the last full”, “the month of July” or “Friday the 2nd”. Instead, we should think in terms of “the one from 10 minutes ago”, “the one 7 hours ago”, “the one 3 months ago”, and so on. All the copies are always from the perspective of “now”, and “the one 7 hours ago” changes every hour, so you don’t need to think about what time and date it was taken.

How is this possible?

The AiRE file system allows you to take instant snapshots of a file system. This snapshot needs no time to be taken – the snapshot is immediate, and the data is always in a consistent state within it. It’s always possible to copy snapshots in an incremental manner because each snapshot is always consistent and no consolidation is necessary as in a VMWare environment. Because each snapshot is always absolutely independent, you can delete any snapshot in the sequence without invalidating the remaining ones. Additionally, each snapshot that you keep only allocates storage space in the differential share of the data between it and any other snapshot of the same file system.

This feature is leveraged by AiRE IntelligentFiler’s Data Protection Engine (DPE) to implement a tier system in which you can define different tier periods, each with its own retention number. This allows you to decide how long each tier will last and the backup retention frequency within each tier.

For example, you may want to protect yourself from threats such as ransomware by setting up a tier for the last 4 hours in which you create one snapshot every 5 minutes (equal to 12 snapshots/hour), then a subsequent tier of 20 hours in which you retain only one snapshot every hour. This way, you’ll always have the ability to rollback your data to five, ten, fifteen minutes and so on up to 240 minutes ago, and then to 300, 360, 420 minutes and so on up to 24 hours ago.

In addition, you can define a third daily and a fourth weekly tier to round out your regular backup strategy. For example, you could retain one snapshot a day for 27 days, plus the 28th day made by the first two tiers, resulting in exactly four weeks, about one month. Then you could make one snapshot a week for 48 weeks, plus the four more weeks made by the first three tiers, resulting in about one year.

Finally, you can define a final long-term retention tier of one snapshot a year for the number of years you are interested in keeping your historical data (e.g. 15 years).

The image below shows how to set up this tier strategy in the AiRE graphical interface.

You can also copy snapshots to a backup target, even a remote target, or clone them on another file system. No matter how long they take to copy, the snapshots will never be modified after being taken, so you have all the time in the world to copy them while users continue working on the file system. In any case, there are several copy techniques to save time, including various levels of data compression and deduplication.

After completing the copy of the first snapshot, you can keep it updated with subsequent snapshots using differential copies. This ensures that the process will only copy the data effectively modified since the previous snapshot, saving a great amount of bandwidth and time.

For replicated data, the AiRE Data Protection Engine interface lets you choose either the same or a different tier and retention strategy than the one applied to the original data. For instance, you don’t need to keep as many recent snapshots as defined in tiers 0 and 1 shown above to avoid ransomware attacks: this is applied only to the live original data. On the other hand, it could be necessary to keep the historical data of the last tier for a longer range of time, or we would want to retain only monthly copies instead of the weekly ones, and so on.

Based on our experience, we strongly believe that this new approach is far better than the traditional one, especially once you become familiar with the principles underlying its operations. It has the potential to save on storage utilization and bandwidth while helping you keep just the right snapshots to ensure business continuity and protect you from ransomware. 

We are so convinced that this is the best way to go that we decided to implement it despite the initial difficulty the user could experience in understanding it. What do you think?

All the best,

Luca