Forecast capacity and optimize resource utilization

This article is Part 5 of A practical guide to AIOps series. This article focuses on leveraging AIOps to forecast capacity and optimize resource utilization.

With digital transformation, IT organizations are increasingly facing growing and unpredictable workloads. For the "born in the cloud" organizations with services on public cloud, it simply means to scale up, down or out as needed. The reality is that most organizations undergoing digital transformation are adopting hybrid, and unpredictable workloads are causing operational challenges like -

• Increasing costs to manage the workload uncertainty
• Customer experience and application performance fall below expectations during high demand period
• Low utilization of infrastructure capacity

Many of these operational challenges can be addressed by improving capacity planning and optimal workload placement. Advanced analytics and machine learning can help improve capacity planning by helping develop forecasting models with large number of long and short term factors including seasonalities and other leading indicators. With the knowledge of daily, monthly, quarterly and annual forecasted trends of workloads, AIOps can also now help in identifying optimal workload placement.

The key success factor for leveraging AIOps to improve capacity forecasting and planning is to enable your forecasting solution with both business and IT metrics. E.g. if your shopping cart application experience high demand during a sales promotion, then developing forecasting model with your sales promotions data can help improve capacity forecasting.

Time and again, capacity forecasts have been proven wrong primarily resulting from extrinsic factors. Having an exception management process to respond to such situations can significantly reduce impact to your customers without deteriorating application performance.

The ideas, opinions and research presented in this article are my personal views on this subject. Please stay tuned to read the conclusion of this series on AIOps use cases. Please leave comments and share your experiences, thoughts and participate in the conversation.

You can also read about the other AIOps use cases from this series :

1. Identify, predict and isolate problems faster (Part 2)
2. Analyze large amounts of operational data for root cause analysis (Part 3)
3. Prioritize problems and recommend solutions (Part 4)
4. Forecast capacity and optimize resource utilization (Part 5)
5. Autonomous IT Operations (Part 6)