Fans of Star Trek: The Original Series know that The Enterprise was not capable of going past Warp Factor 8 for long periods of time. That would be all the power Scotty could give the starship before it started breaking up. I know that you love Star Trek (who doesn’t?) but you might be wondering: what does this piece of Trekkie trivia have to do with cloud metrics?
The answer is “limitations.” Traditional on-premise computing is kind of like The Enterprise: the hardware has limitations on processing power and storage. With cloud computing, however, these limitations virtually disappear. Cloud computing is allowing companies to recognize the benefits of scalability in ways no one ever imagined.
Picture a scenario where your company needs more processing power. Traditional configurations would require you to plan, purchase, and allocate those resources, which could take weeks or even months. By then, the requirements may have already changed. Some companies even purchase for future scaling, which causes them to spend before they need it.
Cloud computing releases that burden of knowing how much processing power will be needed in the future. You simply ask for more (or have your configuration ask for you, triggering resources when it needs more power).
Due to its highly scalable nature, monitoring cloud computing is very different from monitoring on-premise servers. The cloud vendor may have tools you can use. However, if they fall short of your monitoring requirements, you need to seek alternative solutions. In this article, you will discover the metrics to track and how to select the right monitoring tools for your situation.
There’s a vast number of metrics regarding cloud services that could overwhelm your organization. Therefore, you need to determine which metrics are the most important to track and find the tools that will report those metrics.
Obviously, your needs will be different from other organizations and perhaps even other departments. However, some common metrics should serve as a baseline for your monitoring activities:
While this covers several layers, determining where bottlenecks are happening in your applications is essential. In many cases, cloud vendors provide a minimum acceptable level of performance, which might not be enough when you want to improve the performance of your applications. Luckily, competition for cloud services continues to heat up. This is forcing vendors to pay more attention to the performance of its applications.
One best practice when monitoring performance is to view all the metrics in one central location. Otherwise, you will have to constantly cut-and-paste information to match up your metrics. Further, when dealing with hybrid cloud solutions, the monitoring tools must adjust to give reliable reports. Automating as much of this as possible is smart.
When it comes to performance measuring, are there more specific metrics you should track? As it turns out, yes, there are. What metrics are those?
For instance, identifying slow queries is essential to performance monitoring. Regarding that, important metrics to keep track of, per query, are:
- Calls. The number of calls to that specific query.
- Average Time. The time it takes, on average, for the specific query to be executed.
- URLs. The URLs in the web application that make use of the query.
In general, you should save money when adopting cloud technologies, but that doesn’t mean you are free to spend like it’s an open checkbook. When you allocate more resources on the cloud, you can expect this to cost more money. How much more will depend upon the agreements you have with your cloud vendors and what functions you want to include.
One of the best ways to manage costs is to monitor the usage of your resources. Scale back any resources that are deemed unnecessary or don’t meet the objectives of the business.
If certain aspects of your architecture are failing, you want to know this as quickly as possible. You’ll need a tool that gives detailed information about this will help get to the bottom of the situation quickly. Your tool should also help determine the frequency of failures.
Knowing which metrics to track helps you make better decisions for your architecture. Conversely, vanity metrics could lead you to make the wrong decisions. Try to focus on the metrics that affect important aspects of your system.
The above factors (performance, cost, and uptime) are good places to start when doing an audit of your metrics. Remember, you aren’t locked in. Add or subtract different metrics as your needs and applications change.
More specifically, you can use metrics such as logging rates to help you with uptime monitoring. If you monitor the logging rates for a certain period of time, and there’s a sudden lack of data, that clearly indicates a system outage. You can subtract that period to obtain the application’s uptime. If you use a tool that provides monitoring alerts, grouped by their severity level, you can also use them to obtain the app’s uptime.
Finding the Right Cloud Metrics and The Right Tools
To determine specific metrics to monitor, start with the ones you were tracking before your decision to switch to the cloud. Then, look for cloud monitoring tools that will report as many of those metrics as possible. If the tools don’t report every metric on your list, you will have to let the less important ones go. Vendors are adding features all the time. Ask potential vendors how frequently they add new features and whether customers have a say in the process.
When you have a solid understanding of your essential metrics, you should learn what features cloud monitoring tools will provide that will match your criteria. There is some give-and-take. However, your requirements will change, and you’ll revisit the ones you dropped.
I hope you have a better grasp of what is needed to select monitoring tools for your cloud solution. Determine what metrics are important for your organization and seek out vendors who offer solutions that come close to providing those metrics.
- Top Kubernetes Tutorials & Resources for All Levels - January 20, 2020
- Key Cloud Metrics to Monitor When Scaling - December 18, 2019
- Web Application Performance: 7 Common Problems and How to Solve Them - April 26, 2018
- Addressing Roadblocks in Cloud Computing - April 24, 2018
- Performance Engineering: The What, Why, and How’s Explained - April 19, 2018