Often times, people wonder why a backup and recovery
program is necessary when mirroring is taking place.
This question is not at all unreasonable, and in fact, is quite common.
After all, mirroring data to another system (or site for that matter)
will provide a complete backup copy to be available, which can
in turn be restored whenever is necessary.
But is that exactly accurate? Can data be restored when
mirroring alone is relied upon for protection purposes?
Below, you will see that this is indeed not always true.
You will learn that many issues can and likely will arise
in regards to restoring data when mirroring is solely
relied upon for complete data protection.
Let’s start with a few basic examples as to why mirroring should not be the sole source of protection. It has been noted that many consumers do already mirror their data, but that they ultimately desire 3 copies of that particular data, and they want it to be stored on portable media. One of the benefits of this portable 3rd copy is that it allows the customer to restore their data to various data centers should they find a certain one to be full. Another important point here is that data that is corrupted also gets mirrored during the mirroring process, so this 3rd copy would essentially be clean, allowing for a restore to take place from backup where there is no corrupted data. These 2 important points should ultimately be enough to demonstrate why the practice of reliance upon mirroring alone is not wise, but below you will see even further that best practices must be used when it comes to mirroring, and that it is absolutely vital that more than simply mirroring data must occur for guaranteed protection of data.
Remember Hurricane Katrina? Of course you do. It was devastating to both the land and the people occupying it. A prime example as to why mirroring alone is not enough when it comes to data protection became incredibly evident during this huge disaster. Here is the reason why. The main site of a particular company was located in New Orleans and also had this site mirrored in Alabama. Can you take a guess as to what happened during this awful hurricane that swept through the South? Both sites were majorly hit, and thus both sites were LOST. Just like that….the data was gone as a result of one major storm. This example clearly shows how fragile the whole system is, and thus certain protocols must be followed in order to boost protection. When it comes to mirroring, best practices demand that mirrored sites be directionally east of the primary site. This IT disaster is precisely why we cannot rely upon exclusively mirroring for complete data protection.
Not yet convinced? Here is yet another keen example. A few years ago, Gmail was receiving some maintenance from a Google employee. Keep in mind that Google has traditionally been very reliant upon mirroring practices. During this particular maintenance session, the technician inadvertently deleted over 100,000 Gmail accounts. Because of Google’s mirroring practices, these accounts simply could not be recovered. This was a tremendously difficult day for those on the other end, whose data was lost for good. Here is the moral to be taken away from these examples: mirroring can only do so much. If it fails to give data protection completely, then it cannot be the only source of protection.
Keep in mind, this information is not intended to put mirroring in a bad light! Indeed, mirroring is not all bad! On the contrary, it is a fantastic method which helps to create smooth operations regarding business. However, mirroring cannot recreate data, and so when it is necessary to do so (and if one relies solely upon mirroring) the provision of data backup falls upon application programmers. Remember that these programmers are human, and therefore prone to error. Because of that fact, these application programmers might fail to create the necessary and appropriate historical safeguards into their applications.
Expense is another consideration when it comes to breaking down why one should not rely solely upon mirroring for data protection. It has been noted that costs could be driven down if mirroring was kept for various key storage groups only, by prioritizing the most important applications. In turn, storage groups which are less critical-both in time and in importance-would simply be backed up.
All in all, it is vital to remember that full and complete protection of data does NOT come from mirroring alone. Mirroring must be paired with a sufficient backup and recovery plan in order to avoid many recovery problems and disasters. Mainframe environments need to keep this important information at the forefront of their game plans regarding data protection! Why? Because CRUCIAL data is kept here for high profile consumers such as large banks, insurance companies, and government agencies. Imagine if important information in these institutions were lost! It would NOT be good, to say the least.
So, where do you go from here? Below you will see a few tips and tricks based on the above information which will help you create a well-functioning backup and recovery strategy.
You should:
- Ensure that particular products that offer point-in-time replication are made automated.
- Make various versions of historical backup. This will allow a restore to
occur during any time-frame and it will be WITHOUT corrupted data. - Keep a backup copy handy that is portable. You will then be able to use it anywhere.
- Keep recovery granularity in mind, which will grant you the option
to restore from things such as full volume backups.
Mike Johnson is a Technical Writer for Rocket Software. He writes on topics like data protection, backup monitoring and reporting software. He holds a Bachelor of Science Management degree from DeVry University.