Learn about monitoring clouds to understand their performance.
- [Instructor] So let's talk about cloud performance monitoring. What's important here is that we understand that this type of monitoring goes toward the cloud-based systems meeting performance expectations. So, people expect performance to occur out of their computers. It's the single most common complaint that I get... My cloud is running too slow, my database is running too slow, my systems are running too slow. It's closely linked with cloud heath monitoring, but focuses on response... Refocuses on the response that comes back to the end users or that applications that are requesting the service.
So, things have to happen in a timely manner. In fact, there's some legal agreements that we have from time to time, called SLAs or Service Level Agreements, which are contracts between us and the public cloud provider to ensure that their going to live up to a minimum response rate that they have to actually legally live up to or they own you money. So back to the foundations of monitoring and operations. We have performance, which is right next to governance and above security.
So performance is kind of a relative concept. People have different ideas of what performance means. So, my expectations in terms of how my computer performs is very different than someone else's expectations that may be on a factory line, for instance. So they may be used to five second response times and that's perfectly fine for them as to how they do their job, but I need sub-second response time for what I do. And other people may need better response time. For example, real-time data acquisition systems.
In fact, machine-to-machine communications is absolutely dependent on good performance, and if that performance isn't there, then those applications are going to fail to run. So cloud performance entails what's running, just like cloud health. In other words, what databases, what storage services, what virtual machines are currently running on the cloud, and how are they performing? Are we meeting performance expectations? So, somewhere we're setting limits. Somewhere we're setting that if, for instance, the user interface response time goes less than five seconds, then we're out of compliance for the performance expectations that we set, and something has to be corrected.
Some user is notified to go in there and fix something or something is definitely wrong, either with the application or the cloud platform itself. Then any corrective actions that need to be taken. You'll find that some performance issues are relatively easy to solve. It's just a matter of tuning or allocating additional virtual machines, or virtual servers, or virtual storage systems to the particular account so they have more capacity. Where other performance issues, such as things that are built within the program, are difficult if not impossible to solve without doing significant rewriting of those applications.
So the technology needed again is trending. We need to understand what's trending and what's occurring in terms of performance. Is it getting worse? Is it getting better? Is it staying the same? We need to do testing and benchmarks within the system. So every once in a while, instead of the public cloud provider providing you with a test, test yourself. Run some benchmarks. See how it's doing from the end user standpoints. Automation in terms of correcting issues. The ability to kind of kill process. The ability to use some sort of technology that's going to go through and improve performance automatically.
The ability to correct the issue. For example, we have not enough service running, and additional services are spun up automatically. And then alerting. The ability to let someone know... get an email, get a text... that performance is below expectations and needs to be corrected.
- Cloud health, performance, security, and governance monitoring
- Cloud monitoring analytics
- Cloud monitoring costs
- AWS CloudWatch
- Librato CloudWatch
- Cloud Cruiser
- Microsoft cloud monitoring
- Rackspace cloud monitoring
- Creating a cloud monitoring and operations plan
- Defining cloud monitoring operations patterns