Data Organization

---

Disaster recovery planning and contingency plans are now examined specifically with regard to data. Although alternate premises and physical logistics are vital elements, data is the heart and soul of any business. Here are some important points to take into account in order to cope better with data preservation and recovery.

Data Quantity

Although companies may have various replication scenarios in place, in the event of a disaster these plans may become overwhelmed by the sheer quantity of replication that has to be accomplished at once. You may have drilled replication with regard to a specific server, but what if the entire server farm has gone down? Any DRP, then, must include provisions for how exactly to replicate what might be tens of terabytes worth of data in a short time. The storage site, for example, had better have enough bandwidth to relay the data at the required speed. And, once the data has been received, can you use the images immediately or do you have to rebuild databases and cobble together multiple transactions until everything is complete?

Data Priority

Next is priority of data; that is, which systems and databases/servers should be restored first? If a disaster occurs, everybody will want this data restored now; therefore, it is important to have determined which data is most important and should be recovered first in order to get the business going again. Also, set up several phases of data recovery so that, even if the required bandwidth is lacking, at least the first phase can be pulled in immediately and effective action taken to minimize the delay in getting the company functioning. As part of this prioritization, determine an acceptable recovery window for each network or system. Some systems can perhaps be down as long as 48 hours without catastrophic affect on the business, whereas for other systems one hour may be too much downtime. Further, be realistic in your estimates. While it might be unreal to avoid any transaction losses, perhaps you can limit losses to as small a window as possible, but this depends on the type of business, which in turn determines the amount spent on minimizing transactional losses. Whereas a bank may have duplicate recordings of every transaction, a manufacturing company may decide it can afford to lose a day's invoices because it should be able to recover the information based on a few phone calls to the sales staff.

Data Medium

Next we consider the medium on which the data will be recorded. Most DRPs call for daily backup to tape, with a subsequent shift of these tapes offsite. The idea is that, in the event of a disaster, these tapes can be taken to alternative headquarters and the data loaded onto the servers, with the aim of getting the business up and running within a day or two. This is the preferred method for most American companies. Alternatives include electronic vaulting and mirroring. Electronic vaulting is backing up data across a network to an offsite location; this is typically far more expensive than tape backup but yields a faster recovery window. Mirroring data to an identical system can cut the recovery window to less than an hour, provided a remote operations center is available in the event of a disaster. Mirroring, however, significantly adds to cost. As a result, surveys of American enterprises indicate that 19 percent routinely use remote mirroring or continuous data backup procedures for enterprise servers. Gartner Group surveys found that 70 percent of Fortune 500 companies use replication-based backup systems for 10 percent of their data.

Hardware Issues

It is important to pay close attention to hardware issues such as machine types and configurations. Many elements, including disk capacity, device names, peripheral devices, RAM, and many more, must be considered regarding the subject of configuration. In addition to recording everything, it is also essential to consider exactly how the hardware is going to be replaced, detail by detail. According to Nickolett, another issue is deciding whether to use an existing preconfigured machine or to completely configure a machine (load the OS, initialize and configure the disks, determine the TCP/IP configuration and SCSI addresses, etc.). His recommendation is to plan for the worst case (i.e., complete rebuild). It might also be possible to reconstruct a new machine using a tape backup. Although this approach is somewhat inflexible in terms of configuration, it is a lot faster than a manual system reconfiguration.

Networking Issues

It is no good replicating all the hardware and software configurations if the necessary networking points are not taken care of. Are you running a LAN, WAN, or VPN? If so, what types of software and connections are necessary to replicate critical parts of the production environment in order to get back to work?

Software Issues

Software encompasses all manner of operating systems, applications, third-party software, customized code and applications, etc. It is important to have a complete inventory of all of these things, as well as a listing of what is really needed to resume business activity. Also, look into licensing. What happens if it is necessary to reload everything? Do your current licenses make it easy or difficult to do so? Find out what is involved and act before rather than after any emergency to ensure the process is smooth.

---