Organizations often monitor virtual machines (VMs), which are stored in their private datacenters, using third-party tools or Windows Performance Monitor. This helps to ensure that they are running as intended.
The need to monitor workload performance doesn’t disappear just because a workload is moved to the public clouds. CloudWatch is a tool that Amazon Web Services (AWS), provides to administrators to monitor VM performance.
CloudWatch supports two types of monitoring for EC2 instances. These are basic monitoring and detailed monitoring. Figure 1 shows that I have created an instance and logged into the Elastic Comput Cloud (EC2) console. After selecting the instance, I clicked the Monitoring tab in bottom half of screen. This tab displays four charts that display CloudWatch metrics. However, if you look closely, you’ll notice that the Monitoring column for EC2 instances shows a Disabled status. You will also see a link right above the charts that allows you to enable detailed monitoring.
[Click on the image to see a larger view.] Figure 1: CloudWatch provides basic and detailed monitoring. Basic monitoring capabilities are included in the cost of EC2 instances. Data is available in five-minute increments without additional charge. AWS charges extra for detailed monitoring. However, detailed monitoring can be performed in five-minute increments. If you use auto scaling, detailed monitoring can provide aggregate performance metrics for the entire auto scaling group.
The CloudWatch metrics displayed by default are helpful, but the EC2 console provides much more information. Figure 2 shows 14 metrics for the current EC2 instance. These metrics are primarily related to performance but there are also metrics that monitor CPU credit.
[Click on the image to see a larger version.] Figure 2: You can monitor a variety of metrics for your EC2 instance. Click on a metric to see more detail than the one shown in Figure 2. Clicking on a metric will increase the chart’s size to show more detail. The resulting window has a number of drop-down options you can use to adjust your chart view, as shown in Figure 3. You can select the type of statistics displayed in the chart, as well as the time range and sample period.
[Click on the image to see a larger view.] Figure 3: Click on any chart to see a larger and more detailed view. Although Figure 3 is a useful view, administrators don’t spend much time looking at performance charts for individual EC2 instances. While the charts can be helpful if you want to quantify the performance for a specific instance of EC2, they are not practical for monitoring large numbers. CloudWatch alarms can be set up by AWS. Click on the Create Alarm button shown in Figure 3.
I won’t spend too much time on alarms as I have already discussed them in depth in recent columns (see here, here, and here). However, the basic idea is that an alarm can send an alert to an administrator or take corrective action if a threshold value is exceeded. Figure 4 shows that alarms can be matched with any of the metrics in Figure 2.
[Click on the image to see a larger version.] Figure 4: You can use alarms to alert you of performance issues and take automated corrective actions. CloudWatch has many other uses than monitoring EC2 instances. CloudWatch can be accessed from the Management Too.

Oswald