Software-Defined Shifts: Multi-Cloud Operations and the Mythical Single Pane of Glass

Let’s face it: Mythical creatures exist. There’s proof of the Loch Ness Monster and Big Foot, after all.

Chris Wolf, VP, CTO, Global Field and Industry

All kidding aside, in IT circles, there are many who have similarly doubted the existence of a single pane of glass for all things management.

A pane of glass is a management console that presents data from multiple sources—whether that be across the data center or from the data center to the cloud—in a unified view. Enterprise management vendors have been chasing it since the dawn of IT, and we never seem to get there. If such management challenges were difficult to overcome with traditional data centers, they are now impossible to surmount in the cloud era. And the reason is simple: cloud providers today release several new features every week. It’s not possible for any one vendor to expose every new feature of every major provider while meeting all of the agility demands of its customers.

Bigfoot sits on logDispel the Myth, Define Reality
Acknowledging that there will never be one console capable of everything sets you up for a disciplined cloud operations and management strategy. From there, you will need to determine the functions that should be centralized and those that should be decentralized. Many organizations that I consult already have dedicated AWS or Azure admins who are responsible for lower-level configuration and integration work. That allows the enterprise to take full advantage of provider-specific capabilities.

Most enterprises have near-identical requirements for what they would like to manage centrally. Those include:

  • Intelligent, policy-based multi-cloud brokering
  • Multi-cloud cost management
  • Multi-cloud and multi-data center network fabric
  • Centralized multi-cloud encryption and key management
  • Centralized multi-cloud security policy and distributed firewall management
  • Identity federation
  • Performance and change management dashboard
  • Automated discovery of new workloads and configuration changes
  • Distributed data management
  • Cloud-to-cloud or cloud-to-data center workload migration

Flexible and Operational
In many environments, the multi-cloud operational requirements should be modular in how they are packaged, sold, and integrated. That will allow you to choose the parts that are right for your needs without having to go all in from a single vendor. The broker will provide the central brokering and policy enforcement, while likely working with multiple layers of orchestration to automate various operational tasks.

Many of the technologies listed above are in various states of maturity at a given enterprise. Having awareness of the common operational requirements for these technologies will allow you to begin to build a roadmap for multi-cloud operations, and to work with partners and vendors that are best positioned to deliver on your strategic multi-cloud requirements. Loch Ness monster swims in lake

Sniff Out Fact From Fiction
Use caution when evaluating and selecting multi-cloud management and operational platforms. Some vendor feature lists are as blurry and grainy as the above photos. They will promote every conceivable feature checkbox that you could possibly request, and make those checkboxes work with a massive services engagement—one that leaves you with a cloud management platform that is difficult to upgrade and maintain. Staying disciplined regarding the capabilities that should remain centralized will give you a cloud management and operational framework that is both comprehensive and agile.

During the next several months, I plan to deep dive into the multi-cloud discussion at Virtualization Review. Read the first part of this series, “Multi-Cloud Operations: Letting Go and Thinking Differently.”