At this point, cloud computing is a well-established--some might even say boring--realm within the IT industry. Innovation within the cloud has slowed down, and it has become hard to imagine many revolutionary steps forward in the way we build or manage applications within the cloud. With that said, if I had to pick one concept that is poised to exert major change within the cloud in the near future, I’d put my money on AIOps and AIOps tools.
Although the impact of AIOps is by no means limited solely to the cloud, AIOps is one of the few forces right now that is on track to disrupt the way we interact with cloud infrastructures. Here’s what you need to know about AIOps and the role AIOps tools and platforms may play in your future.
AIOps is a term coined by Gartner in 2016 to refer to the application of Artificial Intelligence to IT operations work. Basic AIOps involves using AI to help interpret or analyze data within IT environments. More advanced AIOps use cases center on leveraging AI to automate IT management tasks (like restarting a failed server or updating firewall rules in response to a newly detected threat) that staff would traditionally have had to perform by hand.
The AIOps concept existed before the term was invented. In fact, if you used machine learning or data analytics tools to assist in application monitoring or security testing at any point in the past 20 years, you were using AIOps.
But AIOps and AIOps tools have exploded in popularity during the past several years. That trend reflects the growing sophistication of AI, as well as the ever-increasing scale and complexity of modern IT workloads (challenges that AIOps helps to address by using AI to automate and systematize IT workflows).
AIOps and the Future of Cloud Computing
AIOps has promising applications across the IT industry. Yet, in many respects, cloud computing is one of the areas where AIOps is poised to offer the greatest benefits. That’s because AIOps can solve some of the most complex challenges in the cloud--challenges that other techniques or technologies have yet to address adequately. Following are four main examples:
1. Cost optimization
Running workloads in the cloud is easy enough. Running them in a cost-optimized way is much harder. Cloud vendors don’t exactly go out of their way to help their customers spend less on their platforms. And although various third-party tools are available to help predict and manage cloud costs, most require a fairly significant amount of manual effort on the part of IT teams to set up and use. You have to tag your cloud resources carefully and spend time manually interpreting the cost-savings recommendations that the tools give you. Many of these tools also make after-the-fact recommendations based on past usage, rather than suggesting cloud configuration changes that you can make in real time to save money immediately.
AIOps promises a new level of automation and real-time insight into cloud cost optimization. Not only can AIOps tools make recommendations about where companies are over-spending in the cloud, but they can take the additional step of automatically reconfiguring workloads to save money. An over-provisioned virtual machine instance could be automatically migrated to a lower-cost one by AIOps tools, for example, or data that is stored on a more expensive object-storage tier than it needs could be migrated to a more cost-effective tier instantly.
2. Cloud migration
In some senses, cloud migration is harder than ever. Although the rise of multicloud, and the advent of platforms like Kubernetes and Anthos, has made it easier to integrate workloads running on one cloud with those hosted on another, public clouds have in other respects become more nativist. If you adopt a framework like Azure Stack or AWS Outposts to help build out your cloud workloads, you end up highly dependent on your cloud provider, without an easy way to move your applications, data and configurations to another public cloud without having to rebuild everything from scratch.
AIOps could emerge as a solution to this challenge. In cases where IT teams would otherwise need to rebuild from the ground up to migrate from one cloud to another, AIOps tools could automate the process by leveraging AI to rewrite configurations for the new platform. In other words, instead of having to manually recreate IAM policies, API configurations and so on for a different cloud, IT teams could let AIOps tools do the heavy lifting for them. The result would be a world where cloud migration is smoother, even as different cloud platforms become more complex and unique in their service offerings.
3. Cloud architecture planning
One key challenge that IT teams face when working with cloud environments is that there are so many cloud services to choose from--and so many different configuration options for each service--that identifying the best type of service for each workload is daunting, to say the least.
Will a given application perform the best (and in the most cost-efficient way) if you deploy it in a virtual machine, a container or using serverless functions? Which cloud region or regions will deliver the best results for a given workload? If you want to take advantage of edge computing, where exactly should workloads reside: on cloud gateways, on devices or on a combination of both?
These are the sorts of questions that IT architects grapple with constantly in modern cloud environments. Traditionally, the only way to know which arrangement worked best was to manually test different options and analyze the results.
With AIOps, it becomes easier to predict which architectural patterns and configurations are the best fit for a given cloud workload. Using data about workload requirements and the performance and cost of each potential architectural solution, AIOps tools can make recommendations that are more powerful and systematic than those that IT teams can devise manually.
4. Administering diverse workloads
Along similar lines, the fact that public clouds now offer a litany of different services means that some admins struggle to master them all. That’s understandable; it’s hard to expect a single engineer to command the expertise necessary to manage Windows VM instances, Kubernetes clusters, cloud-based NoSQL databases, SaaS analytics platforms and serverless functions (to name just some of the diverse types of workloads that organizations typically run today in the cloud) all at the same time.
Faced with this challenge, businesses have traditionally had to either hire top-notch IT talent to acquire the skills necessary to cover so many different types of cloud services at once or rely on large IT teams with personnel broad enough in their areas of expertise to manage diverse cloud services.
With AIOps, however, it becomes easier for engineers to manage multiple cloud services even if they don’t have extensive expertise in each one. AIOps tools can analyze and help manage workloads hosted with virtually any type of cloud service, lowering the burden placed on human engineers.
To be sure, AIOps is not a panacea. It can’t solve every challenge in the cloud, and, on its own, it will not usher in a totally new age of cloud computing. Still, more than many other modern technological trends, AIOps is on track to solve several key challenges in cloud computing. In that respects, it’s poised to breathe some new life into what has become a somewhat stale part of the IT ecosystem.