Amazon Web Services is taking its CloudWatch monitoring service to the next level, announcing on Friday morning a half-dozen new features that automate actions when certain metrics are hit or certain alerts are triggered. They are Basic Monitoring, Elastic Load Balancer Health Checks, Alarms, Auto Scaling Suspend/Resume, AutoScaling Follow the Line and Auto Scaling policies. In addition, AWS customers can now use Auto Scaling with their Virtual Private Cloud and Cluster Compute Instances. Cloud monitoring is becoming big business, so anything AWS can do to keep those dollars in-house is probably worth the effort.
Probably the most-important thing about this round of updates is that most of them involve the service actually taking action: the type of automation for which cloud computing has come to be known. Policies for stopping, starting and scaling new instances based on performance metrics takes CloudWatch a step beyond what many monitoring services currently offer, especially for cloud-based resources. Most services, such as Cloudkick or Nimsoft, can provide detailed monitoring and visualizations of numerous cloud platforms, but they can’t automatically act on what they discover.
Actually, the Basic Monitoring feature, which produces performance metrics at five-minute intervals, isn’t an automation feature at all, but it is free. When added to the existing suite of free or reduced-price AWS services, it creates an even fuller experience for developers testing out the cloud or running minimally important applications. Although there are plenty of free cloud-monitoring products, such as Cloudkick’s basic service and Hyperic’s CloudStatus, there’s something to be said about the simplicity of a one-stop shop during the learning phase.
As with most AWS features, the new CloudWatch capabilities take the commodity out of cloud computing, to a degree, and truly differentiate the AWS offering. A developer choosing between cloud providers might lean toward AWS because it doesn’t require going to a third party for advanced monitoring features. In fact, CloludWatch now offers automation capabilities beyond what most third parties offer.
Of course, CloudWatch also doesn’t tie into on-premise systems the way that many of the third-party, cloud-monitoring tools do, nor does it provide granular performance metrics and visualizations. CloudWatch certainly doesn’t monitor other public clouds. There’s a reason why CA paid $350 million for Nimsoft, VMware is pushing Hyperic, and Cloudkick has raised a few million: They’re betting businesses won’t be content running everything in the cloud, or even with a single provider. It seems only likely that as cloud-capable monitoring services and systems-management software integrate within large vendors, AWS-style automation could be possible across a variety of cloud platforms. That means there’s plenty of money to be made, and AWS can’t have all of it.
Heart monitor photo courtesy of Flickr user brykmantra.
Related content from GigaOM Pro (sub req’d):