Resources

Monitoring

Without monitoring you don’t know what your system is doing, which means you can’t fix it, and you can’t improve it. Feedback loops require data. Monitoring systems to collect data is fundamental to the process, however, systems can be extremely complicated. As data are interpreted to generate more and more metrics, correlation is needed to troubleshoot problems or plan for growth.

Unfortunately, as the problem has gotten bigger, existing tools haven’t gotten better. Anyone managing hundreds of machines has felt the intense pain of collecting, correlating and making good sense of the information across all of them. Circonus is a large-scale monitoring and trend analysis system designed to avoid the shortcomings.

Collecting simple metrics like CPU, memory and disk utilization are obvious basics. Network switch ports and host interfaces are commonly polled for traffic metrics. Applications expose their own key metrics that can be trended, e.g., number of users, active connections, requests or transactions processed, latencies. A good administrator goes beyond these systemic metrics and starts looking into business-level metrics like registered users, widgets sold, and credit card transaction volumes. The sheer quantity of information is overwhelming and traditional RRDTool-based systems burst at the seams. At Circonus, we’ve seen this break down, and we created a system that eliminates this inadequacy. Circonus is designed from the ground up to perform at any scale.