To Fix Broken Backup Focus on Service Level Agreements

In scores of conference rooms at companies in nearly every industry, IT staff and executives are meeting to talk about the future of their internal backup infrastructures.  In most cases, these discussions are prompted by an internal sense that backup is broken.  The stakeholders around these tables might note the explosive data storage growth their companies are experiencing, making their backup windows endangered, if not extinct.  They might discuss the impact of virtualization, with its smaller number of servers and increased stress on available input/output (I/O).  Inevitably, they will talk about the downward pressure on their IT operating budgets.  Above all, these countless players discuss how their data backup challenges affect their service level agreements (SLA).  That sound focus on service provides the key to creating a better approach to data backup.

How Virtualization and the 24/7 Business Model are Changing the Way Enterprises View Backup

Data protection has become one of the highest priorities of IT managers responsible for virtualization deployments, but backing up and protecting virtualized servers has been problematic, to say the least.  In a physical server infrastructure, the utilization rate of server and network resources sits at an average of 20 percent, which leaves plenty of available CPU cycles and network bandwidth to satisfy the needs of compute- and network-hungry backup processes.  However, in a virtual server environment, the abundance of resources doesn’t generally apply.  In many cases, the consolidation and density of server infrastructure increases the server and network resource utilization rate to as much as 70 percent, leaving little room for backup.  Therefore, the process that typically took a long time to finish in the past in a physical data center takes much longer in a virtualized environment and may not even be possible.  Standard backup applications and concepts fail to address the data protection needs of virtualized environments and exacerbate the availability and recoverability of such deployments.

There are additional factors that seriously affect the way enterprises look at backup processes and capabilities.  First, the natural growth in the data center adds pressure on existing backup windows, and many organizations today are not able to finish their backups in a timely fashion.  More importantly, however, is the status of the backup window itself.  The Internet has transformed the way businesses cater to customers.  Online services allow enterprises to continuously reach their customers and to break the geographical barriers that they may have had in the past.  In the new 24/7-business model, the concept of a backup window does not even exist.  There is no downtime, no acceptable period in which a company can say, “service not available.”  Therefore, a new method of data protection must be applied to ensure the continuous availability of services.

A Data Backup Model that Looks at the Service as a Protection Unit

Traditionally, the data protection industry focused on how to best protect files and blocks of data to make sure they were recoverable in case of loss or corruption. However, this question doesn’t align with what IT organizations deliver to their customers.  CIOs are responsible for a sustained delivery of services and business applications to their users.  They operate based on service performance and availability, not files and blocks.  For many years, the industry looked at what the infrastructure could deliver and defined SLAs according to IT – not customer – needs.  All the while, vendors were trying to understand the interdependencies of systems and applications and how to deliver a cohesive service to users.