Has the cloud changed your monitoring strategy? If it hasn't, you may want to rethink your approach. The more places you put stuff, the harder it is to keep track of, relevant to almost every company that's neither entirely in the cloud or in its own data center any more. The cloud means more tiers and wider tiers, with more nodes than you'd likely have in your own data center.
It's also true that the more something moves around or changes, the more you need to keep an eye on it, and that's the case with cloud: it's a dynamic environment where usage varies from day to day and workloads move between cloud and on-premises servers. This means identifying issues like the most heavily loaded servers isn't necessarily helpful in maintaining a high-performing environment.
The impact on your network and monitoring your network also is changed by the move to the cloud. The performance of your WAN and the cloud provider's network may have more impact on the performance end users experience than your local network. Understanding the network traffic patterns is also critical to ensuring adequate application performance.
Define the Meaningful Performance Indicators
To achieve effective monitoring, you'll want to establish key performance indicators (KPIs) in to get an overall picture of your system. KPIs should cross all areas of your cloud environment, including application, databases, servers, virtual machines, and storage devices. Typical KPIs include:
- application KPIs such as transactions
- CPU KPIs such as utilization and load
- network KPIs such as number of bytes transmitted within a specific period
- storage KPIs such as number of write operations and free space percentage
- system KPIs tracking downtime and failures
Ideally, you should have all monitoring data visible via a single pane of glass dashboard. In addition to gathering performance metrics, it's important to be able to correlate performance metrics across the components of your infrastructure. Analytic tools can help with analyzing the data to find the relationships. Predictive analytic tools can help you identify potentially developing problems before they have a visible impact—as long as you have an effective process in place for responding to alerts.
Build a Team that Can Respond to Alerts
Indeed, it's the capability to respond to the data that is the most critical element of any monitoring process. Your response shouldn't simply react to alerts, but should utilize the data to help create plans for growing or altering your architecture in response to changes in demand.
Cloud providers typically provide monitoring capability, such as AWS Cloudwatch, but monitoring effectively requires more than simply turning on such software. Because of the distributed nature of today's computing, you may require monitoring tools both in your data center and in the cloud. Pulling it all together and using the data effectively is often made easier through services such as dcVast's Managed Amazon Web Services, which offers 24x7 support. Whether you're just getting started with AWS and want to get off on the right foot, or you need help understanding your current environment's metrics to solve a current problem, our certified engineers implement monitoring that will make your metrics meaningful. Contact us for a free assessment.