Thu Sep 19 07:00:00 UTC 2024: ## Network Observability 1.6: Enhanced Insights with Reduced Resource Usage

The Network Observability Operator has released version 1.6, offering a significant improvement: **network insights for your Red Hat OpenShift cluster without relying on the Loki log aggregation system.** Previously, Loki was required to store network flow data. This new feature unlocks performance gains, reduces resource consumption, and simplifies deployment.

**Key benefits of disabling Loki:**

* **Faster Query Performance:** Prometheus, used as the data source when Loki is disabled, boasts significantly faster query times, ensuring near-instantaneous results even for extensive data ranges.
* **Reduced Resource Utilization:** Tests demonstrate a 45-65% decrease in memory usage and a 10-20% drop in CPU utilization compared to deployments with Loki.
* **Cost Savings:** The need for object storage for Loki is eliminated, lowering overall costs and streamlining operational efficiency.

**A Trade-off: Granularity Limitations:**

While offering performance advantages, disabling Loki limits the level of detail available. Users seeking per-flow or per-pod granularity for troubleshooting may need to re-enable Loki or consider alternative solutions, such as:

* **Enabling specific Prometheus metrics for granular visualization.**
* **Utilizing dedicated tools for detailed flow analysis.**

**Leveraging Prometheus Metrics for Real-World Scenarios:**

The new feature unlocks the potential of Prometheus metrics for proactive issue detection. For example, users can now set up alerts for DNS latency issues, quickly identifying workloads impacted by slow DNS resolution.

**Overall Impact:**

The ability to disable Loki provides a more lightweight and cost-effective solution for network observability in Red Hat OpenShift. This enhancement empowers users to gain insights into their network, diagnose issues efficiently, and set up customized alerts, ultimately bolstering the stability and performance of their OpenShift clusters.

Read More