I recently wrote an article for the InfoWorld website that pretty much encapsulates everything you need to know about cloud storage. Read it, and you’ll be ready to start building your cloud. Well, not exactly. But the article should give you a handle on the basics—and help you understand why you shouldn’t need to know everything to build a datacenter that delivers effective storage to true cloud environments. Mastering a few storage concepts can set the stage for success.
After you learn a bit about abstracted storage models and how they came to be, you’ll be better prepared to make decisions about cloud storage. Virtual disks that are generated by the abstracted storage, not physical storage protocols, are at the core of how storage works in the cloud. Of course, a physical storage protocol is necessary for storage. However, the physical protocol is decoupled from the storage model in a cloud datacenter.
How is it that storage models and storage protocols can be decoupled? That’s definitely not in line with conventional ways of thinking about storage. To get a handle on the relationship between storage protocols and storage models, read the article. It provides a fast walk through the basics of physical storage protocols. The article also discusses why abstracted storage models emerged with virtualization, and how these models evolved to better suit cloud environments. Then, you’ll get a quick overview of abstracted storage models—instance, volume, and object storage—along with a discussion of the advantage of object storage for many cloud use cases.
What does cloud storage imply for your datacenter? Most storage evolved to work in non-virtualized environments. As you may know from experience with virtualization, you can make traditional storage approaches work with virtual machines. That’s going to be true in elastic cloud environments, too. But using traditional storage introduces hard-to-manage complexity that could reduce the flexibility and scalability of your cloud environment. You’re probably better off looking at storage—such as application-aware storage from Tintri—designed for abstracted models and virtual machines.
Unique control with VM-level actions for infrastructure functions including snapshots, replication and QoS make protection and performance certain in production, and accelerate test and development cycles.