John's been trying to run this backup for quite some time and always comes up against the same problem: the backup window is too short and full backups do not complete correctly. Knowing that there is no life insurance policy for his critical data weighs heavily on his shoulders.
John is in charge of protecting his company's large scale-out NAS. For the last 10 years he has backed up these growing volumes of data and he knows where the pain points are. On the strength of his experience, he is perfectly positioned to define an ideal petabyte NAS solution.
The 5 ingredients for a perfect petascale NAS backup solution:
- Leverage snapshots to rapidly list all changed files and stop long and painful scans of the entire file system which can last for days, making it impossible to run daily incremental or weekly full backups. Atempo rapidly builds lists of new, changed and deleted files.
Optimize scanning time and start actual data movement early on.Why waste time not moving data even if the scan is still in progress.
Our solution starts actual data movement early and balances the files and load intelligently between multiple nodes. This way, having billions of small files or some very large files is not a show stopper.
Parallelize data movement across multiple nodes on the NAS filer. Ideally, the solution will offer control over which node to use and the period the backup. In this way, production is not impacted.
Provide faster restore than with traditional NDMP technology not be limited to tape-only backups. Open full access to object storage and cloud targets.
Faultlessly manage the protection of millions or even billions of small files and also the backup of very large files >500 GB by multi-threading file batches and optimizing I/Os on the NAS.
It's important to be able to combine all the above to manage fast backup solutions that deliver efficient data protection. We need to respect customers constraints for their incremental and total backup windows.
And by way of a bonus, our large-capacity NAS backup specialists would like to add two other points to John's list:
- Leveraging snapshots is not enough, good backup must also identify the files deleted on the NAS so the full backup image that is restored is accurate.
- It's time to abandon the Full/Incremental backup cycles. From now on, it's incremental only!
John may also be surprised to learn that there is a solution which can deliver a full restore by doing only incremental backups – without performing new full backups.
This approach combines the capacities of a file system or an object storage and the native capacity of our solution to limit the number of versions backed up for each object. Add a certain backup know-how (something we have plenty of here at Atempo!).
There is life after NDMP which combines smart and flexible technologies
So back to John and his question and his peace of mind.... Yes, we can reliably and rapidly backup large scale NAS to, for example EMC's Elastic Cloud Object Storage (ECS) or to another disk-based storage and/or to tape or most major Cloud providers (AWS, Azure, Google, S3).
Let's hope that in his now-restful sleep, John can dream of incremental backups to infinity (and beyond!).
Rather than a thousand and one words, you might want to take a look at what our customer UNIL (University of Lausanne in Switzerland) has to say about using our backup solution to protect their replicated Isilon NAS.