Tape vs. Cloud

In this days it is hip to store all kinds of data in the cloud, thus on a storage reachable over the internet and housed in a more or less huge data center. This trend is also seen in terms of data protection. The idea behind this is to store the data that was previously saved to magnetic tapes to offsite storage to comply with requirements of external storage.

This idea that can be quite interesting for companies of any size, gets simplified by the use of Optimized Deduplication within Backup Exec.
In the past, many attempts to copy data over an internet link failed because of the amount of data compared to the available bandwidth.

By using Optimized Deduplication however, it is now possible to transfer the whole stack of data once to the storage provider (i.e. by using removable storage disks).
Afterwards, only the changed blocks will be transported over the network link during the backup time.

Let’s make a sample calculation here:
We assume a company has one terabyte of data to be protected during a full backup. An incremental backup can be calculated with 15% of the amount of a full backup. The local deduplication will reduce the data by a factor of, let’s say 75%. This means that we have to transport something around 37.5 GB of data over the internet line to the storage provider. If we can assume that the internet line is usable for backup purposes from 11pm to 7am, we have eight hours to transport the data which means that we’ll need an upload of approximately 15 Mbit/sec.

And what about a full backup at the weekend? There will be much more data, of course. But the local deduplication will discard so much of it that the amount of data that needs to be transported over the internet will be nearly the same as during an incremental backup. In addition to this we do have a disproportional larger backup window during the weekend.

So, in case you do not have enough bandwidth to store your backup in the cloud every day, you could think of replicating only the weekly full backups to the provider and keep the incremental backups only on premise.

This way of doing backup is quite common. Many companies do copy only their weekly full backups to tapes and keep the daily incremental backups solely on disk.
You could also think of replicating dedicated backups to the provider on a daily base, i.e. the backups of your mail system.

Before you can start to use this kin d of backup, all your data has to be copied to the storage provider once. This is done by a so-called “preseeding” and is done by simply creating a backup to removable storage (i.e. USB drives) and transporting them to the provider. The provider will then make a local copy of the backups to the deduplication storage dedicated to your company. If this is done, you can immediately start your backups sing Optimized Deduplication.

One point that can be quite challenging in such a scenario is the attempt to restore a larger amount of data.

Retrieving a single file or folder is easy, as the primary backups are done on premise and you can just restore from your local copy. Ut if you have to restore data that is older than the timeframe you keep on premise, you’ll need to retrieve the data from the providers storage. As it may be not a good idea to move those larger restores over the internet line, the provider should offer a service to extract the data needed from the deduplication storage to a removable disk and send this to you.

Which way ever your provider will offer to you, please keep in mind that restoring data will take much longer as if it was done on premise.

Leave a Comment