Share this post:
Interested in adopting application performance monitoring? Let’s take a look at how you can embrace automation in the style of one of my favorite things to read: a classic, therapeutic newspaper advice column.
It was a beautiful spring Sunday with great weather outside. But it was ruined once again with an alert. My customer was having issues with their web application. The ticket said that the app was loading slowly but doesn’t elaborate further. As a network admin, I had to then go pull wireshark traces and TCP dumps and then comb through those logs. But I was still unsure of the root cause. So I had to guess that it was the network’s fault and take the blame. I have gone on one too many Sunday hunts for a needle in a haystack. Please help.
— Another Network Admin Buried Deep in Haystacks
Dear Not-Just-Another Network Admin,
Those of us with network responsibilities often worry about application deployment and delivery. But many of us desperately lack architectural innovation and access to real-time telemetry.
For future incidents, I recommend you research application performance monitoring technologies. This will equip you well for when incidents occur in future. And they will.
Here’s a simple, three-step methodology that will help you get started.
Some may try trick you into believing that you can achieve the results you seek with traditional approaches to cloud computing. But you need infrastructure that not only captures real-time telemetry but also can process millions of data points in real time without any performance impact.
Solutions built on software-defined principles separate the data plane from the control plane. This gives you flexibility. The data plane can just capture real-time application traffic telemetry and feed it to the off-path control plane. Your control plane can analyze these metrics and present the insights in a visual dashboard without impacting performance.
Of the various elements of application traffic that you can measure, you need to identify the relevant metrics. Then you can configure your tools to collect real-time telemetry from your application instances.
You will need insights into:
- End-user performance
- Page load times
- Media and files accesses
- URLs and URIs accessed
- Response codes
- Client analytics such as location, device types, operating system versions and browsers
Together, all of this can average millions of data points per second. Traditional computing models can neither scale nor process potential petabytes of data without performance degradation.
I’m reminded of an IT joke: “Automate painful processes and now you do stupid things faster.”
We adopt cloud-native architectures to achieve flexibility, agility and continuous delivery. Automation plays a critical role in achieving these benefits. Based on the insights you get from real-time application analytics, your network team can automatically scale their resources to mirror traffic patterns. Application teams win too – they can automate application services thereby shortening development life cycle.
With these three steps to get you and your team started, you will notice that your teams and your infrastructure solutions walk in sync.
And before you know it, sunny days ruined by alerts will be a thing of past.
PS: To learn more about APM, join us at IBM InterConnect March 19 – 23, 2017. Or download the DevOps APM for Dummies ebook to learn how teams work together to continuously deliver secure, available, application insights.