The foundational component of any data protection strategy is the backup process. It is the backup system’s responsibility to capture and secure data as it changes and then make sure that data can be recovered in a timely manner. But the backup process is being stretched to the breaking point thanks to rapidly changing environments and ever increasing user/business expectations. As a result the backup process in many organizations is a mash up of multiple software and hardware solutions that are inefficient and don’t achieve their intended goal.
The Evolution of Backup Software
The enterprise backup application can be an amazing source of information. Unlike any other process it gets to “see” all the data in the environment as it changes. It can also provide analytics on how long it takes to transfer data and how quickly the devices receiving the data can absorb those transfers. The backup software application has the potential to know more about the operational characteristics of that data than potentially any other process in the environment, including the application that created the data.
Unfortunately, the potential of this operational metadata generated by the backup process goes largely underutilized. It is time for backup vendors to evolve their solutions so that the backup software and the entire data protection process can be tapped to provide this resource. For example, companies like HP are capturing the backup metadata and are utilizing it to create an adaptive backup and recovery experience for their customers.
Once backup software solutions begin to analyze the information that they are already capturing, they can leverage that analysis to create a backup process that can prioritize backup jobs, predict backup outcomes, visualize the backup process and automate it so it will become self-healing.
The first benefit to come from this analysis is the ability to prioritize jobs. This should allow for better utilization of resources and to ensure protection of more mission critical data assets. Most backup jobs are treated on a first-come, first-served basis depending on available resources, typically the moment the backup window opens up. But this means all the data starts flowing to backup servers and storage at about the same time. The process could potentially flood those storage systems and servers as well as the network. Also, many applications will tend to do some pre-processing work, like a sort or re-index, prior to committing data to the backup process. In almost any case it’s better to stagger backup jobs. Prioritization allows the backup administrator to select which jobs should receive the highest priority but then allow the backup application to leverage the backup resources to their fullest potential. The backup application can learn which data sets are ready to send data first and set priorities based on that knowledge.
One of the key challenges that a backup administrator faces is making sure that backup windows will remain adequate for a particular application as the environment changes. Backups can overshoot their available windows because of data growth but also because other applications or data sets are growing too. Since these all share the same backup infrastructure device contention for those resources often results. The problem is that the first time a backup administrator is made aware of a potential backup window failure is when the backup doesn’t get completed. Fixing this problem requires either re-scheduling or re-directing jobs as well as buying additional backup hardware.
The typical workaround for backup window failure is to buy plenty of excess backup hardware. This approach causes backup and network resources to be wasted until the data set gets big enough to consume them. When the environment finally does scale the administrator is back in the same place of not being forewarned about a backup window that’s potentially too short to complete the scheduled backup jobs. In other words, buying excess backup hardware only postpones the inevitable.
Instead, the adaptive backup solution should monitor this situation and provide warning to the backup administrator when a backup window failure is about to occur. Again, the backup application has the critical variables like overall data growth, time required to write data to the backup device and how those variables are changing. It should be able to leverage this historical data and look ahead to provide notice well in advance of any pending problems with the backup window. The backup application should be able to visually deliver this information so that the backup manager can implement changes.
The next requirement for an adaptive backup solution is to provide recommendations based on its look-ahead capabilities and best practices. These recommendations should take into account historical data growth, infrastructure utilization and the SLA for a given application. By making recommendations to the backup administrator decisions can be quickly made to maintain the backup process and meet SLAs. For example, when the look-ahead capabilities show that a backup window will no longer be met the backup application could then provide some suggested changes, either in the form of backup job rescheduling or the addition of more backup hardware.
Another recommendation could be backup window improvement. For example, if the organization decides that it needs to reduce a backup window from 6 hours to 4 hours the backup application should be able to provide some insight about how to achieve that goal. Again, it could be as simple as reconfiguring backup jobs or suggesting additional backup hardware.
Not Just for Backups
Finally, these recommendations would not be limited to backup analysis. Recoveries could also be analyzed. For example if an application needs to retain a four-hour recovery window the adaptive backup solution should be able to monitor that system’s backups and recovery potential. It should inform when the application has not been successfully backed up and when the application will take longer than four hours to recover.
As a first step the insight that comes from applying analytics to a backup application is extremely valuable, but the end-game should be for the software to automatically adjust to the environment as it changes. In this scenario the backup application can follow the backup and recovery priorities set per application by the backup administrator and take steps during the backup process to make sure those priorities are maintained.
For example, if a database application is ranked as a priority-one application, the backup software should have the ability to automatically adjust the backup and recovery process to make sure that critical application is fully protected and recoverable in the timeframe required. This could include moving the backup job to another device and pausing other jobs. The backup application would of course need to inform the backup administrator of the changes it made.
Virtualization, the pace of change as well as the growth in the importance of mission critical databases as well as the overall explosion in unstructured data is putting a new level of strain on the backup process. This strain potentially overwhelms the backup administrators tasked with keeping everything operating smoothly. Ironically the backup application captures valuable information that the administrator could use to stay on top of the process. But most applications don’t analyze this information nor do they have the capability to mine it.
Companies like HP, which is leveraging its deep expertise in analytics, have significant advantage in their abilities to create backup applications that can inform users of potential problems while adapting to overcome the challenges of an out of control data protection process.
HP is a client of Storage Switzerland