While public cloud storage systems have enjoyed incredible growth and success, these architectures have not yet enjoyed anything close to the same success in the traditional data center. Perhaps it’s because these organizations don’t think they need it, or perhaps they feel object based storage systems are too complicated, so they decide that their scale-out NAS can do the job for now. Perhaps if they had a greater mandate to implement private cloud storage, and if that storage was easier to install, the trajectory of private cloud storage might change.
Object storage systems are the tool of choice when building cloud storage for three primary reasons: ease of management, low cost, and built-in data protection. Ease of management is a broad term that includes things like how easy it is to scale object-based storage and how easy it is to maintain the system over time by adding new nodes to and retiring old nodes from a cluster. Management is also easier due to how it deals with components via the integrated data protection features of object storage. Whether an object storage system uses erasure coding or replication, it is able to withstand many more simultaneous failures than a typical scale-out NAS system, including failures across multiple nodes or even multiple locations. These protections also enable organizations to put their trust in commodity servers and storage, which drives down hard storage costs.
Yet still, the problem many people have with object-based storage is their applications don’t yet know how to write to it. They know how to write to any POSIX-based file system, but not to something like the S3 API. This will change over time, but it hasn’t changed yet. Which is why it is difficult for many people to use object-based storage; they still need NFS or SMB access.
This is why one of the most popular products right now is NFS/SMB gateways to object-based storage systems, or object-based systems with native NFS/SMB capabilities. Customers get the benefits of NFS or SMB access while also getting object-based storage on the backend. This way they get the manageability and scalability features of object storage, as well as its integrated data protection features, but with the ease-of-use of NFS or SMB.
Another feature that might not be immediately obvious when using an NFS/SMB front-end with an object-based storage backend, is customers who decide they want to use a hybrid cloud approach have the functionality already built-in. Since the NFS/SMB component already knows how to write to object-based storage, it could facilitate the move of data between private and public cloud and back.
Private cloud storage is on the rise, but it is not enjoying the same popularity as public cloud storage. Increasing the ease of moving both older and newer data into the private cloud system should help increase adoption of such systems in companies of all sizes. Having a product that can talk both NFS/SMB and object storage protocols would also help companies to adjust their cloud strategy over time to include both public and private cloud.