This is a very good question. Below, you will see exactly why mirroring alone doesn’t cut it for complete data protection.
It makes sense why individuals are confused about all of this. After all, mirroring to another system or site provides a full backup copy of data, which can, in turn, be restored anytime…correct?
Well, this is not entirely accurate. In fact, believing this way sets the stage for a potential problem regarding the process of restoring.
Many companies mirror their data to another system, but actually want three copies of their data, with the third being a transportable one.
Clients want this data to be stored on portable media so they can take it with them anywhere.
This way, there is increased flexibility to restore to various data centers, should they find one to be full.
In speaking of portability, it is good to note that both tape cartridges and virtual tape can be made portable.
Another important point in regards to mirroring is the fact that during the mirroring process, data that is corrupted will get mirrored as well. This third copy not only offers a portable solution but a corruption free copy of the data as well.
The good news is, the undefiled data allows clients to restore from a backup that was created prior to any corruption taking place.
Here are a few real-life examples what will help explain why mirroring alone is not enough to protect data.
Hurricane Katrina devastated much of the South in 2005, causing great damage to not only physical structures such as homes, roads, and buildings but also to virtual structures as well. A heartbreaking example of this was when a particular company lost both of their relied upon sites during this disastrous storm.
Why did this company lose everything?
They were relying on their main site, which was located in New Orleans, along with their mirrored site, which was located in Alabama, and both sites were destroyed during Katrina.
This close proximity in mirroring offers a perfect situation for data loss. Best practices regarding the mirroring process state that there must be an east-west practice. In other words, the mirrored site must be east of the primary site. If this point was put into practice, it is likely that at least one site would have survived.
Another example of the folly of simply depending upon mirroring happened in 2011 at Google.
A particular technician was fixing a problem in Gmail. This resulted in an accidental deletion of over 100,000 accounts, and they were ultimately gone forever.
Why? Because Google depends mostly on mirroring, and this technician’s mistake hit all mirrored sites as well. Here is an example of a big time “oops.”
Google apologized for their mistake, but obviously, even the best apology cannot bring data back after it’s lost. Here is the main point in all of this: mirroring is NOT enough to offer complete data protection! Mirroring can only do so much! It is important that consumers view mirroring in the appropriate light so they won’t expect more of the process than the process can offer.
Mirroring is great on many levels. In fact, it need not be criticized for what it does best, which is creating a smooth canvas for continuity in operations.
The fact is, mirroring cannot offer historical recreation of data, thus, we must not depend upon it to do so. In the event of needing to recreate data, mirroring places the responsibility upon the programmer, who may ultimately fail at providing appropriate safeguards for their application due to human error.
The other point to keep in mind is that mirroring is not a frugal process. The more money-saving alternative might be to limit mirroring to the most important storage groups while backing up less critical applications.
In summary, mirroring is not all sufficient when it comes to protecting data. In fact, it is simply one small piece of the puzzle towards data protection. Clients must consider a supplemental backup and recovery plan along with mirroring, in order to avoid recovery issues. Major mainframe environments should pay close attention to this point as highly critical data is stored in these particular environments for banks, and insurance and government agencies.
Here are some steps to help create a suitable backup and recovery plan:
1. Switch to automated for point-in-time products that replicate data.
2. Many versions of your historical backup ought to be made, in order to facilitate restoration from any time frame, and without corruption of data.
3. A transportable copy of your backup should be obtained. This way, you can use it anywhere.
4. Ensure recovery granularity is in place. This will allow a restore to occur from
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.