This article is part of the series about continuous data protection (CDP). In the first article I introduced the concept of CDP. I recommend reading the article first, before you continue. This article discusses Microsoft Data Protection Manager 2006 as one example of a CDP system. Today, I will focus on its advantages over traditional disk-based backup systems. In the next article of this series I will talk about the disadvantages. I recently published an article about the DPM in the German magazine iX. My German readers can get the complete review there.
Maybe the DPM is not a good example to use in discussing the advantages of CDP, since it only supports “near continuous data protection”. There has to be at least one hour time difference between two scheduled shadow copies and DPM can only create eight shadow copies a day. Thus, on average the time difference between two backups will be even greater than one hour.
Of course you can also have eight disk-based backups a day with any other professional backup software. So why do you need DPM after all? The point is that DPM specializes on frequent backups:
1. Continuous Backup
The DPM agent logs continuously all changes on the file sever. This causes an extra load on the server of about 5% which is comparable to an on-access virus scanner. If you backup a file server eight times a day with conventional backup software, you will have much more additional load on the file server. A traditional backup tool scans the whole server for files where the archive bit has changed. This reduces the performance of any read/write access to the hard disk noticeably.
2. Data block replication
The DPM agent doesn’t copy the whole file to the backup server, if only some bytes have changed. Only those data blocks, which changed since the last creation of a shadow copy, will be replicated. This saves network bandwidth and backup space.
3. Manual shadow copy creation
DPM only allows scheduling of eight shadow copies a day. However, the data replication to the backup server can be scheduled hourly. An administrator can use the replicated data to create a shadow copy manually any time. Therefore, if you realize early enough that a restoration is necessary; you only will lose the data a maximum of one hour. To achieve the same RPO (Recovery Point Objective) with conventional disk-based backup software you have to run a backup job every hour.
4. Server state restoration
With conventional backup software you usually work with a combination of full and incremental/differential backups. If you have to restore a folder or the whole file server, you usually will restore files which were deliberately deleted by the users after the full backup. The DPM agent logs all changes on the file server. Hence, it also recognizes if a file is deleted. This change will be replicated to the backup server and the deleted file will not be a part of the next shadow copy. Therefore, if you restore the complete file server you will get the exact state of the file system at the time where the shadow copy was created.
5. End-user recovery
This feature of the DPM is not directly related to the CDP technology. I am mentioning it here because it is an advantage of the DPM over most conventional backup solutions and it is also typical for other CDP systems. End-user recovery with the DPM is quite user-friendly. One uses the Windows explorer to navigate to the file or folder. The properties of the file show the older versions. One can then preview the file before restoring it. The procedure is exactly the same as with the shadow copy feature of Windows Server 2003. Actually, one can combine shadow copy for file shares with the DPM. For the end-user, there will be no difference whether the file is restored from the file server or from the DPM server.
This post only covers the advantages of the DPM over conventional backup systems. In the next article in this series I will discuss some of its problems and disadvantages. Stay tuned!