IT planners are increasingly becoming concerned about service levels. These objectives are often thought of as the performance that the storage system will provide to the applications whose data it will store. But there is often an important missing ingredient… data protection. As most IT professionals know, protection of data on a storage system is far more than just RAID and snapshots. It involves copying data on to a secondary storage device and eventually storing that information on a device in an offsite location for disaster recovery.
Managing data protection to service level objectives (SLOs) allows the process of protection to become more focused on the needs of the applications rather than the needs of the protection process. Given the size of an enterprise protection process, job failures are almost a given. In fact, every iteration of the backup process will generate some failures. The challenge is, the backup administrator typically has limited visibility into which failed job should be addressed first and in most cases, they don’t have the time in the day to address every failed protection job. The result is they address failure in a first in, first out method as their schedule permits.
SLO driven data protection, on the other hand, allows for the better prioritization of addressing failed backup jobs. Now the backup administrator can focus on the failed parts of the protection process that are causing an application to fall out of SLO compliance. It also allows the backup administrator to declare a “stop everything” moment so that a mission critical application can be brought back into compliance.
Instead of forcing everything to fit a “one size fits all” enterprise backup policy, the SLO driven data protection plan acknowledges the fact that there are many data protection features available to the application. This can include RAID, mirroring, snapshots, replication, application driven backup, enterprise backup application protection, archiving and vaulting. The protection SLO drives how each of these protection capabilities is leveraged.
What Is The Protection SLO?
The protection service level objective is the agreement between IT planners and application owners to how much time may transpire, if there is a storage system outage, before a given application must be returned to full operation. There are three primary components to the protection SLO.
- Recovery Time Objective (RTO) which specifies the maximum duration for how long an application will be off-line in the event of an outage.
- Recovery Point Objective (RPO) which specifies how much data will need to be re-keyed in the event of a failure.
- Retention Objective (RO) which specifies how long will data need to be retained for an application, and how quickly will that retained information need to be recovered. It is assumed that retained data does not need to be as readily available as recovery data.
See our article “Backup Basics: What is RPO, RTO and RO?” for a detailed discussion on these three letter acronyms and why they are critical to backup design.
Establishing a protection service level is a manual process that requires interacting with the application and data owners to define what their specific needs are. Then by taking this information and analyzing the data protection tool(s) at the organization’s disposal, a strategy can be developed to help meet end user expectations. It may involve purchasing a reporting tool that can help you consolidate into a single dashboard all the data protection events running against that SLO. EMC has this capability via its Data Protection Advisor (DPA) product.
Automatically Provisioned Protection SLOs
As EMC’s Stephen Manley pointed out in his recent blog “Tech Prediction 2014 – A Battle Cry For Protected Storage”, in 2014 IT planners need to begin to look for vendors that can provide primary storage systems that can integrate with data protection solutions to provide a policy driven data protection SLO. The goal would be to take some of the manual labor out of the process. There are early signs of this happening through companies like EMC providing primary storage snapshot management directly from within their data protection applications; as we discussed in a recent briefing note on NetWorker 8.
In the future, when storage is provisioned for an application, part of that provisioning process needs to be configuring SLOs. A few storage systems now have the ability to set performance and availability SLOs; like what type of storage media that a particular application’s data will reside on and how many IOPS a given application will require. But no storage system suppliers provide the ability to set the various protection parameters described above at the point of volume provisioning.
Storage Swiss Take
While we agree with Manley that in 2014 “The battle cry will be: “I want to provision protected storage”, today it is a manual process but it can be successfully monitored from an SLO perspective with tools like DPA. That manual effort, however, should deliver increased data protection, reliability and confidence as well as simplification of the backup process. As the future unfolds, those companies that begin to articulate and then deliver on a message of provisioning protected storage should have a distinct advantage over those that cannot.
EMC is a client of Storage Switzerland