Public cloud storage is highly scalable and very cost effective. It also happens to be located next to an almost infinite supply of compute. Why then do most primary storage vendors pretend the resource doesn’t exist? The answer is quite simple, embracing the cloud will cost them sales of storage capacity. The reality is there are several use cases for a cloud instance of an on-premises storage system that will be very beneficial to those vendors’ customers.
The Use Cases for Primary Storage To Cloud Integration
The most obvious use case for a primary storage system to integrate to a public cloud storage provider is to be able to use the public cloud as a cold tier. As data ages, it moves from the primary storage system to the cloud. But it is still available for transparent recall if it is accessed. While this curtails primary storage capacity growth, user experience is not compromised nor is the administrator’s job made more difficult. While the storage vendor might lose revenue placing physical hardware, it could easily find a way to charge fees based on the amount of cloud space the customer consumes. Using the cloud as a cold tier would require no in-cloud presence, the on-premises storage software could simply convert data to S3 Object protocol on the fly as it is set.
Another somewhat obvious step is to use public cloud storage as a disaster recovery point. While most storage vendors provide replication software, not every organization has a secondary site, or at least a secondary site with a data center and staff to manage it. A cloud replication feature would entail instantiating an instance of the storage software in the public cloud and then replicating data from the on-premises storage system to the virtual cloud-based storage system.
The next step would be to allow that secondary copy of data in the cloud to be acted upon by cloud compute resources. Taking this step would enable the public cloud to be more than just a replication target, it could actually be the data center in the event of a disaster. For the storage vendor, enabling cloud compute is a higher level of integration and one that might take some work. But clearly the precedence is there since so many disaster recovery as a service (DRaaS) vendors provides this feature. Enabling the storage system to provide a similar functionality eliminates the need for introducing a whole new vendor into the mix.
The ability to operate on replicated data opens up whole new sets of possibilities around copy data management. If the replicated copy can be snapshot and set to read/write, then the snapshot copy could be used for testing and development, reporting and analytics processing.
The coup de grâce would be to make the cloud just another node or controller to the data center storage system which would enable live bursting of operations to the cloud. If a peak workload were to occur, compute could be started in the cloud to run extra instances of an application. While it may take a few minutes for the final changes to be sent, the ability to start additional processing in the cloud would dramatically change the way organizations not only scale, but reduce the build-out requirements of their on-premises data center.
The use cases for integration to the public cloud are numerous, and each has specific benefits to overwhelmed IT organizations. Also, except for archiving to the public cloud, none of the use cases reduce the potential for the sale of on-premises storage. In fact, the primary storage vendor has an additional revenue opportunity by reselling the public cloud provider’s capacity. The primary storage providers that move first and move completely by covering each of the use cases stand to have a significant competitive advantage, as do their customers, by providing a practical application of cloud infrastructure.