Mainframe Cloud Storage: Ready or Not, Here it Comes

By Glenn Wilcock

Just a few years ago, any mention of storing z/OS data on cloud object storage was met with scoffs and scorn. z/OS data and cloud storage were seen as oil and water. Yet today, many clients are exploring or have already introduced cloud object storage into their z/OS environments.

So, what has changed? The reasons vary from needing to react to an executive’s edict of, “Thou shalt use Cloud Storage,” to an enterprise team fully understanding the intrinsic value of cloud storage. 

Before any enterprise determines if cloud storage is right for their environment, they must dispel the myth that cloud storage is just a big, multi-tenant, off premise, public data warehouse. While this is certainly one instance, it is not the only instance.

So, what is cloud object storage? All storage has three basic components: An API, a server (controller) and physical storage media. For example, virtual tape systems (VTS) use a tape API, have a tape controller, and implement various combinations of spinning disk and/or physical tape. Applications that interact with a VTS do not need an awareness of the actual physical media type since they simply interface with the server through the tape API. The tape server externally provides tape mounts, block reads and writes, seeks, and rewinds regardless of the physical media used internally.

Cloud storage utilizes the same basic components while leveraging a very simple RESTful API. Four basic command constructs—‘put object’, ‘get object’, delete object’, and ‘query’, enable applications to efficiently store, retrieve and manage their data. (Each provider of cloud object storage has a unique variation for authentication protocol and API). An object is the data itself. 

Additionally, the simple RESTful API and object data format enable data to be accessed from any platform and any physical location. This means that the actual server and media can be installed right next to your z System, or on the other side of the world. It can be privately or publicly managed, single or multi-tenant, or any combination thereof. If desired, your z/OS data can share the same cloud storage instance as your distributed data. The actual cloud storage implementation is driven by your data’s access, availability and cost requirements. 

Wary of using a multi-tenant, public cloud storage offering? Start with a private, on premise cloud storage instance that is completely owned and managed by your enterprise. Later, leverage a third-party, off-premise cloud storage offering that enables a fixed, low cost solution for your z/OS encrypted, offline data.

Hopefully, this article gets you thinking about the possibilities. We’ll follow up with other articles on the many z/OS cloud storage solutions that are available today and the unique value that they each provide.  Until then, peruse the SHARE archives for the various cloud storage sessions that have been offered by various vendors at recent conferences. My personal favorite is “Cloudy, with 100% Chance for Gain”!

Glenn Wilcock is a Senior Technical Staff Member with the IBM Systems Unit, located in Tucson, Arizona. As a z/OS DFSMS architect, his primary focus is to create Data Availability and Information Life Management solutions. These solutions leverage the latest storage technologies to provide leading edge functionality for z Systems clients. He has over 25 years of experience with storage management technologies.

Recent Stories
Using IPCS to Solve CICS Failures

Tips and Tricks: The Transition to Remote Work

Master the Mainframe 2005: Anna Clayton's Springboard to Success