Knowing what information and data types that need to be protected is important. by Ron McCabe, Communications News, September 2007, www.commnews.com
Demanding requirements for a workable disaster-recovery solution cause IT managers to be faced with some cold realities. Recent studies have shown that, on average, IT managers only have a 30 percent data recovery rate when failure or disaster occurs. Combined with the expanding scope and volume of information that needs to be protected, this presents a daunting challenge to most IT managers.To develop a disaster-recovery solution that works, understanding what information needs to be protected is important, as well as the data types. In some cases, a single, simple solution will suffice. More common is that IT managers will find several distinct types of data with different characteristics, as well as different recovery-point objectives. For example, backing up critical documents only on a daily basis may be acceptable, even though changes that occur between backups may be lost as a result. In this particular case, a daily file-level backup effectively protects the critical information.An organization may also have structured applications like databases (DBII, Oracle and SQL), or semistructured applications like e-mail, which present more complex scenarios. These data types often have near real-time recovery requirements, as well as unique recovery characteristics.There is a spectrum of needs in the document backup environment. If files are relatively static and have a low rate of change, most solutions will work. As file change increases, however, along with the relative rate of activity, these standard solutions start to fail.This is due to what is often referred to as the “open file problem,” which becomes a problem when a backup occurs. The backup software often “skips” this reserved open file and tries to return later to back up this file when it is closed. If this file is not closed before the backup software is finished, the file is never backed up and can create an opportunity for file corruption upon recovery.Because of their data change rate and level of activity, structured and semistructured applications often require more advanced techniques like real-time volume replication to provide effective recovery. These solutions operate at the volume level rather than the file level, avoiding file-level issues while providing a short window of exposure and rapid recovery.When evaluating different products, IT managers should focus on recovery. The key is to understand that almost all of the available replication solutions will work if all the data or packets get to the remote location. This is often referred to as needing to allow the data to “settle” or “acquiesce.”The majority of replication solutions will have failure conditions if packets are in transit when a disaster occurs. This is especially true for structured and semistructured applications, as their file-level and block-level data-recovery journals are interdependent.
In order to ensure that an effective recovery mechanism is implemented, IT managers should verify that the solutions deployed are able to recover in a variety of compromised environments. When researching solutions, check the actual customer recovery rate.
