We have workloads in VMWare on-prem and others in GCP. Currently, our admins have to check three different dashboards to see if the system is healthy. Is there a tool that actually provides a unified view of logs, metrics, and traces across both environments without costing a fortune in data ingestion fees?
3 answers
Achieving "Single Pane of Glass" observability is the holy grail of Hybrid Cloud. In 2024, tools like Datadog and Dynatrace are the market leaders, but their ingestion costs can be astronomical if you log everything. A more cost-effective approach is using an OpenTelemetry (OTel) based stack. By using OTel collectors on both your VMWare VMs and your GCP clusters, you can standardize your telemetry data before sending it to a backend like Grafana or an ELK stack. This prevents vendor lock-in and allows you to filter out the "noise" at the source, significantly reducing the amount of data you pay to store in the cloud.
Have you tried using the native "Connect" features of your cloud provider? For example, can Google Cloud Operations (Stackdriver) monitor your on-prem VMWare nodes directly?
Don't underestimate the power of a well-configured Prometheus and Grafana setup. It’s open-source, highly flexible, and there’s a massive community for troubleshooting.
Agreed, Christopher. For many of our smaller hybrid clusters, Prometheus is more than enough and keeps our overhead low while giving us the dashboards we need.
Daniel, we tried that. To answer your question, it works for basic metrics like CPU and RAM, but once you want deep application tracing across the hybrid link, the native cloud tools start to struggle with the "on-prem" side of the transaction, which is why we're moving toward a dedicated third-party tool.