Monitor AWS performance more efficiently by learning how Opsview Opspacks oversee a variety of AWS services.
You are here
An Extension on CloudWatch Monitoring
AWS has been at the forefront of cloud computing since its launch in 2006. I’m sure almost every IT Operations team has considered moving some of their workload to AWS due to its turnkey nature and near unlimited scale.
CloudWatch is arguably one of the least appreciated features within AWS Monitoring. It offers agentless metric collection of many AWS services and the ability to either showcase these in dashboards or take actions using a range of notification services from AWS. CloudWatch is also used for many internal mechanisms such as autoscaling, which is certainly one of the most interesting elements of the AWS platform. There’s also an additional, somewhat hidden benefit to using CloudWatch; one which you never really think about until you really need to. Due to being completely agentless CloudWatch doesn’t have to worry about traffic to cross your network, all of it is dealt with by the platform and is held under Amazon’s responsibility, so it ‘just works’.
Should you look for monitoring on top?
CloudWatch is subject to a few limits that makes it a difficult tool to rely on for stand-alone monitoring. Firstly, it isn’t possible to collect information from sources outside of AWS. Nor are you able to customise the granularity of the storage of metrics. For example your 5 min data will expire by day 64 and your 1hr aggregated metrics are gone after 455 days so this may cause you a compliance headache if you don’t take the time to make sure that you are regularly exporting metrics to storage.
The real downfall of relying on just CloudWatch for monitoring is the idea that it provides a view of just your AWS estate. Seldom does a company manage to get to a point where their entire infrastructure is hosted on just AWS. The end result of this is that you could be adding another random dashboard into your monitoring world, never being able to get the single pane of glass advantage that most monitoring vendors afford you in some way or another.
Improving your AWS Monitoring
What you should be doing is looking to supplement using CloudWatch. By using a monitoring tool that can collect data from CloudWatch you can compare and contrast your different metrics and also create much improved dashboards above what you can achieve within AWS. Opsview can enable this with our Generic CloudWatch Opspack. It may not have the most interesting name, but it can enable you to start collecting and aggregating exactly the metrics you need to see in a platform that can collect more than just AWS data.
By collecting more than just AWS data you have more control to resolve, report and move quickly on issues within the business. This is increasingly more prevalent considering the current hot topic of data-sovereignty - understanding what data you have and where you store it. In essence your monitoring data becomes yours when you monitor on top of CloudWatch. You can store it at the granularity you require and keep copies that you own for the retention period you require for compliance all within an environment that you yourself control. Complete flexibility and complete peace of mind.
You can of course take monitoring on top of CloudWatch further by supplementing the monitoring with additional checks. Simply deploying an agent on your EC2 instances allows you to collect detailed and focused metrics from the hosts with ease, while being able to create and configure your own service checks to return relevant metrics from your custom application. This will let you collect metrics that would not normally be returned via CloudWatch without having to deliver them via the CloudWatch API including creating the tooling to do it.
How to monitor AWS in Opsview
I’ll be the first to say it. If you are fully cloud native and none of your application or business systems exist outside of AWS, your apps are all deployed as per AWS’s best practice and you are confident that everything is covered then go with CloudWatch. But if you have systems that are not turnkey AWS deployments or have any element of a hybrid or multi cloud workload then take a moment to look at what you can do to monitor this hybrid environment. You might find that there is good reason to get some a more specialist monitoring capability, that will allow you to anticipate and resolve issues before they impact your users.
More like this
Oracle performance tuning advice that goes beyond the obvious. Discover how you can make the most from your system settings.
One of the top complaints regarding Nagios is its lack of scalability, an issue that should be a serious concern for IT professionals.