AIOps is a trending topic within the hybrid and complex IT landscape. And in fact, many companies already started their journey to AIOps. They have more and new requirements for IT operations.
From our experience working with China and ASEAN Pacific customers, we have found that uptime, consistent interactive application response time, resource efficiency are more important than ever for our customers. And data analytics for z/OS and IBM Middleware logs and performance metrics helps with outage avoidance, performance monitoring, real time tuning, resource efficiency, to allow enterprises to meet the high demands of their business applications more consistently.
To achieve the AIOPs goal, customers are using both IBM and other analytics products, but also, some customers have significant internal development teams, that develop customized IT analytics. But in all cases, customers are craving a good set of data sources, that are configurable, flexible, performant, cost effective. Let us dive into the key requirements resonating with customers:Reliable data collection that can cover all kinds of operational data from single data gatherer.
For traditional mainframe customers, they usually have various kinds of workload running on such as zOS, CICS, DB2, MQ, IMS, etc. To monitor system health there involves different tools to collect operational data and to analyze. For example, CICS Performance Analyzer, CICS statistics, OMEGAMON portfolio, DB2 Report and so on. The tools bind data source and data analytics together in each tool itself. The ideal way is to separate data collection from data analytics. There is demand to centralize data collection from single data gatherer to collect different kinds of operational data, rather than work with many individual tools, and their internal data sources.
After data collection is centralized, customers need data collection to be configurable. This is reasonable since different customers have their own workload pattern and AIOps goal is different. They need the ability to configure the data type to be collected or even data fields in specific data type to be collected. This makes the data collection more effective. What’s more, the data collected can be sent to a couple of popular data analytics platform using a subscription model. This requires widely tolerant data format or configurable format for data output.
The data analytics product is the brain and core processor for AIOps. We need to keep the data analytics function flexible to support different flavor of analyzing. For example, instead of providing an existing or fixed analysis report or graphs to the customer, a more flexible way is to provide API to the customer. The customer can choose which model to use, can customize the input data of the model, training interval, aggregation key etc. Also, making data analytics open source is a good way to enrich existing data report and analyze model. We have seen some customers with strong development teams that are already writing their own data analytics platform. Flexible data analytics product will reduce the effort and bar for customer to customize their own IT operational platform.
And what’s more important, all these above are in the scope of IBM z AIOps direction.
Depending on where you are on your journey to adopting more AIOps best practices we have developed the following resources:
And finally, to research our IBM Z products that are implementing AIOps technologies to improve operational resiliency visit our product portfolio page