The prevailing narrative around cloud today is one of agility and choice. From an infrastructure perspective, cloud theoretically gives businesses the ability to continuously modify their infrastructure to address ever-shifting demand and ever-expanding application portfolios.
From a development perspective, Docker containerization theoretically gives businesses the ability to deploy app code anywhere—while mixing-and-matching code modules to quickly bring new capabilities to market.
These dream scenarios, however, are being brought down to earth by the reality of IT ops. In fact, operational issues are seriously constraining the ability of businesses to reap the promised benefits of a frictionless cloud market where code can run wherever it’s most technically and economically advantageous to do so at any given time.
Five management issues in particular are limiting cloud choice:
Issue #1: Poor cloud visibility
Source: Thinkstock / jewhyte
It’s great to be able to run your code anywhere you want. Unfortunately, that doesn’t help if you can’t clearly see how that code is behaving across your compute, storage, and network fabrics. That’s why many businesses quickly encounter problems with performance troubleshooting and capacity planning as soon as they start running critical services in the cloud. These problems force businesses to put the brakes on cloud adoption until they re-tool for better cloud APM.
Issue #2: Cloud ops sectarianism
Some management tools offer pretty good capabilities for certain cloud platforms—such as AWS or Azure—but don’t support ops excellence equally across “all of the above.” That lack of cloud agnosticism can make the transfer of workloads between different flavors of cloud a practical impossibility. The result: cloud vendor lock-in that can’t be entirely blamed on cloud vendors.
Issue #3: The cloud/datacenter divide
Businesses don’t just need the freedom to move application and database workloads between clouds. They also need to move workloads freely between clouds, their datacenter, and private hosting—depending on which one best fulfills their requirements for scale, performance, cost, and/or compliance.
More importantly, many end-to-end customer- and/or employee-facing services leverage code and data in the cloud and in the datacenter. Even the most cloud-agnostic management toolkit can’t help ops with these challenges if it can’t easily be pointed at their datacenter and all its legacy systems of record. And datacenter tools aren’t much use if they don’t support cloud.
Issue #4: Inadequate flexibility
It’s one thing for a management toolkit to support application workloads in multiple clouds and/or the datacenter. It’s another thing for a management toolkit to continuously and easily support application workloads as they move between those provisioning models. Choice is inhibited if ops has to work too hard or play catch up for too long whenever a workload moves. Choice is enabled, on the other hand, by management tools that automatically follow workloads wherever they go.
Issue #5: Inadequate features
A management toolkit that adaptively supports multiple clouds and the datacenter isn’t much help if it doesn’t support them very well. IT ops teams therefore don’t just need tools that offer broad platform support. They also need tools that capture all necessary metrics, correlate metrics across ever-changing end-to-end application delivery paths, provide pro-active alerts without generating excessive false positives, and more.
Simply put, effective DevOps requires attention to ops as well as dev. Careful attention. In the current industry climate where code is exalted as the be-all and end-all of digital business, it’s easy to forget that.
But code only delivers value if it runs reliably. And cloud only delivers value if its reliability can be properly and universally managed. Those are ops challenges that demand appropriate respect.
Cameron Van Orman is the senior vice president of Product Marketing at CA Technologies.