Why Composable Analytics Matter for Multi-Cloud AIOps

An Intellyx BrainBlog by Jason English for CloudFabrix

There’s plenty of loaded terminology and buzzword bingo when it comes to the latest advances in cloud application delivery.

Especially when it comes to multi-cloud – which should merely mean multiple cloud instances when modern cloud applications really leverage multiple hybrid IT operating models, atop both existing business silos and newer microservices application workloads.

I watched a recent webinar featuring CloudFabrix VP AI and CMO Shailesh Manjrekar, and Intellyx President Jason Bloomberg titled: “The Secret to AIOps for Hybrid Multicloud: Composable Analytics” and came away with several insights for overcoming the diversity of today’s distributed application architectures. 

Jason Bloomberg warned viewers to beware of siloed operating models. The people who need to manage cloud infrastructure will constantly change, depending on the multi-faceted nature of today’s Hybrid IT architecture – consisting of public clouds, private clouds, SaaS/PaaS resources, on-premise datacenters and edge servers and devices.

“Siloed operating models can create a management headache,” said Bloomberg. “But they create a huge problem when you have an application asset that spans multiple deployments, in multi-cloud, or hybrid, or edge…ITSM was never intended to support managing separate operating models.”

Dozens or even thousands of streaming events, from service providers, acquired systems, container images, CI/CD pipelines, Kafka clusters, and agents will generate massive loads of data – more than any number of IT professionals can handle.

AIOps – yet another hot buzzword – offers to solve this problem through machine learning technology that can automatically spot anomalies in the data stream, filter alerts, and speed resolutions. 

The problem? Most ITOps management and observability tools were taught on machine learning data from the perspective of only one operating model, or one part of the topology such as networking or a given hyperscaler, or one dimension of metrics or alerting that doesn’t meet the observability needs of all roles.

There are few rote ‘best practices’ for hybrid cloud application management. What works for one company, or one role or user profile within a company, often fails to deliver the insights needed for the next.

What we need is an end-to-end control plane where you can conduct management, observability and automated deployment in a form that matches the flexible nature of these hybrid IT environments, as well as the people responsible for them. 

This is where composable analytics come in. As Manjrekar said in the webinar, “Observability with composable analytics is really all about the user being able to ask the right questions.” 

Shailesh continued, demonstrating how their CloudFabrix platform can fetch telemetry data from multiple leading commercial observability platforms and open source monitoring tools, clearly showing a map of dependencies, with alerts and recommendations that are bubbled up by severity or source to a custom dashboard to the needs of a user in a BizOps or DevOps or SRE incident manager role.

Consolidated, composable analytics based on AIOps filtering and recommendations can match the visibility needs of specific personas, so any role can immediately gain insight into the performance of a distributed system, and collaborate with flexible incident management teams to not just solve problems, but deliver on key administrative objectives such as resolution time, productivity, or cost optimization.

All in all, a lot of cloud ground covered was covered by our experts in this hour-long session, and you can watch the video archive on LinkedIn here: https://www.linkedin.com/video/event/urn:li:ugcPost:6973683916884905985/

©2022 Intellyx LLC. Intellyx retains editorial control of this document. At the time of writing, CloudFabrix is an Intellyx customer. Image source: Photo credits: Dayne Topkin on Unsplash, Analytics screenshot, CloudFabrix.

Jason Bloomberg
Jason Bloomberg