I have been writing for many years about the requirement to manage a computing environment. During a recent conversation with an IT operations manager at a major corporation, the complexity of the problem was put into stark language for me. He pointed out users within his company demand that the applications and services they use to get their jobs done must be up and running consistently and predictably. If a cloud-based service such as the CRM or office applications are slow or down users become furious. “They don’t care whose fault it is. They don’t accept the fact that the cloud service provider is having an outage. They blame my department.” To the users within this business these services – whether they are on premises or in the cloud – are simply computing.
So the crisis that ops organizations face is being able to manage or monitor all the services that businesses depend on. Matters are much worse when an increasing number of employees, customers, and partners are working remotely. A couple of hundred end points has become thousands if not millions of end points. At the same time, the commercialization of AI and Machine Learning (ML) is awaking businesses to find a way to monetize their data across a hybrid computing environment. The data that is required for AI and ML is rarely in one location – it is distributed across on premises data centers, departmental line of business applications and cloud based SaaS applications.
The challenge is to make enable IT Operations (what is becoming known as ITOps) to manage all of these services as though they were one unified computing environment. All of the CIOs and CTOs that I know would love to wave a magic wand and immediately implement a management infrastructure that alerts them to how each service is operating across a multicloud and hybrid environment. For example, these leaders want to be able to switch clouds if there is a performance bottleneck, or to have the flexibility to switch clouds as pricing models change. Managers want to be able to notify their SaaS providers when their services are under performing and get answers so they can manage user expectations.
Unfortunately there are no easy answers. Having looked at many of the proposed solutions no vendor has figured out the magic solution. That being said, there are many point solutions and the beginning of approaches to multicloud and hybrid cloud management are on the horizon. These solutions range from the ability to monitor performance to providing services that can manage Kubernetes clusters.. Other solutions leverage machine learning models that can identify patterns and identify and sometimes automate fixes. Despite announcements and pronouncements from vendors about revolutionary solutions, I predict that it will be several years before true artificial intelligence models are able to monitor, manage, and balance complex hybrid cloud workloads. However, in the real world, users don’t really care about the intricacies of how various cloud services work – they simply want predictability and reliability.