A Practical Guide to Comprehensive Data Protection

Data protection is the preparation for and recovery from data emergencies. Backup and replication technologies contribute to a data-protection solution but a complete solution involves best practices, services, and technology. As might be expected, data-protection is an underlying foundation for disaster recovery and high availability. However, data-protection solutions also mitigate the impact of operational mishaps and intrusions that necessitate data recovery.

Data protection problems and challenges

To understand data protection in today’s business environment, it is necessary to place it in the context of business trends. At one time, backup jobs ran in off hour backup windows when production workloads were idle. As more applications share the same body of data from widely scattered locations, off-hour windows are becoming a vestige of the past. Essentially, it is daytime somewhere all the time. Old data-protection approaches can barely solve today’s problems and do not solve tomorrow’s problems. Figure 1 illustrates the many opposing forces challenging IT organizations.

Some problems are closely related. For example, expert skills are more closely related to complexity than to simplicity. Similarly, data growth and data retention are closely related to IT budgets, whereas backup window and data loss problems both impact availability.

The expert skills problem is of such importance, it is worth more discussion. During a data emergency, people unexpectedly face problems sometimes not seen for months or years. It makes little business sense for an organization to employ full-time staff simply to respond to very infrequent events. Such responsibilities are often given to IT staff that normally performs other duties. When facing emergencies, people experience a period of abnormal stress, which can compound problems. Therefore, responding to a data emergency requires a high level of expertise and practice. However, utilizing these experts to respond to such infrequent episodes would require them to be reassigned from other important projects at a moment’s notice. People with a high level of expertise are usually over-committed and are as likely to be traveling as they are to be onsite when a data emergency occurs. A technical solution without skilled people, who are rehearsed and ready to respond, is of minimal value. Finding, training, and retaining skilled data protection professionals are ongoing concerns for IT management.

As backup windows disappear – a phenomenon that is already occurring in most large international enterprises – simply performing backup operations faster is not a sufficient response. The disappearing backup window is symptomatic of a business environment that demands greater data availability. Various replication and online backup approaches seemingly avoid backup windows, but these approaches often have such an appetite for system resources that they contend with production workloads. Therefore, new approaches must be adopted that do not contend with production workloads and are not bound by narrow backup windows.

Figure 1 – Data protection challenges

The volume of data requiring protection continues to grow while recovery demands become more aggressive. Growth in the amount of data requiring protection is driven by two factors: the volume of data continues to grow 30 to 60 percent per year for most large enterprises and new regulations require companies to retain more data for longer periods of time. Recovery requirements are becoming more aggressive as business dependencies on data become more critical and critical applications increasingly depend on the same body of data.

When multiple applications access the same body of data, the most aggressive application-recovery objective determines the data-recovery objective. Because it has become commonplace for multiple applications to share the same body of data, shared data recovery is the most critical data-protection problem to solve. Most data sharing is accomplished by client/server data access, which delegates a server to provide access to and management of a given body of data. Applications access the body of shared data through a given data server, which acts as a broker. Therefore, recovering shared data to a usable state depends on recovery of its broker – its data server.

Repercussions of doing nothing

Doing nothing about data-protection problems can have severe consequences, such as lost business or productivity, damage to an IT organization’s reputation, blemish on your career, liability for lost data, inability to defend lawsuits, fines for regulatory violations, or bankruptcy and prosecution in extreme cases. Because doing nothing can have severe consequences, you should ask yourself the following questions:

• Have you protected yourself from data emergencies that might result from operational mishaps, hostile intrusions, and disastrous events?
• Have you implemented best practices, do you comply with regulations, and test your state of preparedness regularly?
• Can you retrieve data at both central and remote offices on demand in audit situations or data emergencies?
• Is your data-protection investment adequate to maintain a state-of-the-art environment plus find, retain, and replace skilled data-protection professionals?

Of course, few organizations can answer all questions affirmatively. But what is good enough for your company and your career? If you are dissatisfied or unsure of your responses, further exploration is advisable. Some of the many symptoms that might signal data-protection problems are:

• Excessive data recovery incidents caused by operational mishaps
• Excessive time elapses before recovery can begin
• Compounding of data recovery problems from operational mistakes
• Resolution of data protection problems drag on for weeks to months
• Testing of data recovery procedures known or perceived to be a potential disaster
• Patch levels are far out-of-date or data protection products are out-of-date by more than two release levels
• Backup jobs frequently delayed until the next available backup window
• Inadequate or nonexistent recovery for database servers, file servers, and mail servers

Anatomy of a comprehensive solution

Although technology is of major importance to a data protection solution, people and processes are of equal importance. Technical aspects are most closely related to solution design, solution implementation, and operational tasks. People and process aspects are most closely related to planning, best practices, and ongoing testing. A comprehensive solution combines technology and services into a cost-effective solution with the following benefits:

• Reduces the amount of application downtime caused by data emergencies
• Meets recovery objectives that support even the most critical data
• Cost-effectively backs up and retains massive amounts of non-critical data
• Raises backup and recovery success rates well above industry standards
• Mitigates constraints imposed by tight or disappearing backup windows
• Minimizes the gap between the current environment and a state-of-the-art environment
• Mitigates the attrition or loss of skilled data protection professionals
• Minimizes and manages IT operational risks

The major technology goal is to design, implement, and maintain a state-of-the-art data-protection environment, which displays the following functional characteristics:

• Protects data in a manner that meets the recovery time objective (RTO) with no more data loss than specified by the recovery point objective (RPO). The solution might employ combinations of replication and backup to this end.
• Supports recovery of all classes of data such as database, flat files, and email. Recovery means restoring data to its normal operational state.
• Employs the most cost-effective protection methodology for each class of data without compromising recovery objectives. For example, files with very relaxed recovery objectives would be handled differently than mission-critical databases.
• Recovers data servers to an operational state. Recovery is not complete until applications can access their data.
• Avoids or eliminates tight backup windows thereby providing flexibility in backup schedules and increasing success rates. Ideally, offloads CPU-, memory-, and I/O-intensive operations from application servers to management servers.
• Uses tape media and tape drives efficiently to contain costs. Costs associated with tapes are a major factor in a backup solution.
• Uses the replication network thriftily to contain network costs. Studies have shown that the network is a major cost factor in employing a replication solution.
• Requires no more than a skeletal onsite staff. Eliminating the need for a large staff at production locations saves onsite staffing costs, leverages offsite services for multiple production locations, and mitigates the impact of a local area disaster.

Aligning technical approaches to data protection objectives

There is no single technical approach that both delivers aggressive recovery and meets low cost objectives. Therefore, any solution must be a hybrid, where each technical approach addresses part of the problem. Figure 2 shows how technical approaches align with recovery objectives:

Figure 2 – Alignment of Technical Approaches to Data Recovery Objectives

Backup is normally the last line of defense when all other data-protection approaches fail. Tape backup addresses the low-cost problem where recovery objectives are least aggressive; a day or more. Disk-based backup can support data-recovery objectives from a few hours to a day but at a higher unit-cost than tape backup.

Replication approaches can support data-recovery objectives of minutes to hours but at a higher unit cost than disk-based backup. Whereas periodic replication can reduce recovery time to a few hours, asynchronous replication can reduce recovery times to a few minutes. However, incremental backup using a disk-based approach can closely approximate periodic replication while taking advantage of inexpensive tape storage for old backup images.

Synchronous replication is the only way to achieve zero data loss; a zero recovery point objective (RPO). As expected, purely synchronous replication approaches have the highest unit-cost, especially over long distances. Skillful hybrids of synchronous and asynchronous approaches can deliver zero RPO over long distances, in all but extreme cases, at unit-costs similar to asynchronous approaches.

An overriding problem is that a cost-effective, integrated solution does not come ready to use. It requires expert design, skillful integration of technology, rigorous testing, and a long-lasting commitment to maintenance. Anything less does not protect data.

Conclusion

IT organizations should think about how they will create or acquire a complete data-protection solution that meets current needs and evolves to meet future needs. The business environment is changing and data-protection must change with it.

Because ignoring problems can have severe consequences, it makes sense to do something about them. IT organizations should not take a piecemeal approach to fixing problems of such a critical nature.