Fire Drills and The Art of Cloud Monitoring
In the digital world we all live in today, everyone's worst nightmare is the critical infrastructure issue that turns into a middle of the night fire drill. Whether it is a vital component of your internal business operations or even worse, a customer-facing revenue-generating application, the blood pressures begin to rise, support teams are called out of bed, and the mad scramble to resolve the issue at all costs begins.
At that moment, it doesn't matter what happened, it’s just about restoring the service and listening to a continuous refrain of: “Is it working yet?”, “Why didn't we know about this potential issue sooner?”, “You know I can’t get my work done”, and everyone’s favorite, “This is costing us a fortune.”
Wouldn't it be great if we could prevent situations like these from occurring in the first place, or at least have better visibility into the potential failure prior to it actually occurring?
Just the other day I hopped in to my car, started the engine and my battery light came on literally seconds before the entire car decided to completely shut down. I thought to myself “a little warning would have been nice”. If only my car reminded me that my battery was due for a replacement I could have stopped by the local repair shop and easily purchased a new one. Instead I spent most of my day waiting for a tow truck to take me to the nearest repair shop wasting valuable time and causing much frustration not to mention increased expense. The same analogy can be applied to your cloud resources. No one wants to find out there is a problem right after it happens. I am sure we all would much rather be able to see a potential situation develop in real time so we can mitigate the failure well before workloads are impacted and sleepless chaos ensues. HP Cloud Monitoring can provide that current view into your work load performance so that you can monitor your applications and take proactive measures well in advance of any potentially embarrassing and costly failure.
HP Cloud Monitoring is an open source based monitoring product that provides direct access to compute and block storage performance metrics. The service provides near real-time visibility into critical metrics such as the average bytes per second read from your block store, so you can identify potential latency concerns before they impact the customer; or the number of bytes traveling in and out of a network interface, allowing you to detect potential bottle necks earlier and take action to redistribute network traffic before application performance begins to suffer.
In addition to near real-time metric visibility, you are now able to configure threshold-based alarms that trigger notifications when limits have been exceeded. Email or SMS notifications can rapidly be delivered to mobile devices extending support coverage well beyond traditional boundaries.
A key early design goal was to expose our metrics as efficiently and as easily as possible. We decided one great way to do this was to use an advanced messaging protocol to stream metrics directly to you. AMQP is an open messaging protocol standard that was developed as a highly reliable scalable solution for secure workloads.
Don’t worry though; you don’t need to know anything about AMQP to use our service. You can simply use the HP Cloud Monitoring REST API to create an endpoint and to subscribe to the metrics that matter most. Attach a simple listener to the endpoint to begin streaming metrics at, for example, 1 minute intervals to either your existing toolset or any number of open source visualization tools. We even developed this handy listener app that you can use to get up and running without writing a single line of code.
We feel that our unique approach to stream your workload metrics directly to a managed message queue provides a scalable delivery method that can easily be integrated with most third party tools.
OpenStack® and HP Cloud Monitoring
We are currently working with the OpenStack® community to evolve cloud monitoring concepts and integrate components of the OpenStack cloud monitoring framework as it develops.
Getting Started with HP Cloud Monitoring
With this open architecture, our customers now have the ability to monitor their cloud resources in near real time and detect problems early to prevent performance issues from becoming fire drills. To get started, sign up for free early access to HP Cloud Monitoring. You must already have an HP Cloud account to be part of the program.
Once you receive a welcome email, you can subscribe to metrics and configure thresholds and notifications immediately using the HP Cloud Monitoring REST API.
Finally, your feedback is why we run these early access programs. We would love to hear from you, both the good and bad, so sign up today and let us know what you think. You can reach us on our forum or contact the HP Cloud Monitoring team directly.