It seems strange that, in 2018, there should still be a recognizable category of “cloud.” A decade ago, when the first rumblings of what would become cloud computing were echoing around the enterprise IT space, there was a need to define “cloud” as distinct from those other ways of delivering IT resources. Today, in 2018, is there really still a need for that distinction?
The evidence suggests that there is in fact still a need to consider “cloud” as something distinct from “regular” IT. The Cloud Expo Europe event has been running for many years, racking up thousands of attendees, with no sign of people losing interest — this year, 30k attendees are expected. Far from shrinking, additional areas of coverage are being added, extending into security, Big Data, IoT, and data centers.
Why the Cloud is Still Different
The reason for this continuing interest in cloud computing as being distinct from other forms of computing is the mismatch between companies’ historical IT systems, which work (more or less) and are fairly well-understood, with a substantial body of processes and systems to ensure that they continue working.
Processes based on rigid assumptions — that outages would have one singular root cause, that incidents would have one owner, and that the state of the overall IT environment was documented or at least knowable in theory — start to fail in the multifarious, dynamic world of the cloud.
The new forms of infrastructure that fall under the general umbrella of “cloud,” on the other hand, have often been adopted under the radar, without formal input from corporate IT. SA a result, they still lack this sort of process maturity to enable their management by ITOps teams. This is partly because developers and even business users are able to procure and deploy IT capacity without needing to go through arcane procurement processes, shortening the delivery time for new services — and to be clear, this is GREAT. At the end of the day, it lets people get their jobs done, which IT is generally in favor of.
There is, however, a slight problem for the people tasked with keeping all of these systems running. For a while there, it looked like a cloud server could be treated in the same way as a physical one — just easier to get hold of when you needed it. However, it turned out that no, that was not the whole story. People got used to some of the higher-level functions of their chosen cloud provider, and they liked the way they could get things done without having to file a change request in triplicate, or wait for the monthly meetings of the Change Advisory Board.
This meant that more and more activity went directly to the cloud platform, while less and less went through the formal IT procurement process. In turn, this meant that ITOps teams were less able to plan what they would do if — or more realistically, when — something went wrong. How would they even detect that a problem was occurring, how would they diagnose it, and how would they collaborate between different teams to resolve it? Processes based on rigid assumptions — that outages would have one singular root cause, that incidents would have one owner, and that the state of the overall IT environment was documented or at least knowable in theory — start to fail in the multifarious, dynamic world of the cloud. This is where new approaches are required.
Cloud Impact on Traditional IT
There is no point in trying to continue with business as usual. Even if IT were in a position to decline requests for more services, more support, and more capacity, these days business users have myriad options available to them. Thanks to that same cloud, they are able to procure their own IT support without necessarily having to go through a formal IT process. More and more IT is being done outside “the IT department,” whatever that even means nowadays.
The new approach is for IT Operations to become as flexible, as dynamic, and as agile as the systems that they support. The only way to achieve that result is to adopt algorithms and automation right into the process itself: Use algorithms to make sense of the state of the environment, instead of trying to build more and more rules and filters. Break down the walls between departments and enable collaboration across disciplines and specializations. Integrate automation directly into the Operations process via techniques such as ChatOps.
If you want to learn more, come along to Cloud Computing Expo in London. My presentation is AIOps: How AI & Machine Learning Can Help Operate A Future-Proof Cloud, at 13.05 on the 22nd in the DevOps, Containers and Software Architecture Theatre. Moogsoft also has a booth, C1972, if you want to see a demo of this technology in action, or have a more focused discussion. See you there!
