Optimizing Application Performance with Change Management Improvements - Part 1
May 02, 2017

Mike Cuppett
Author of "DevOps, DBAs, and DbaaS"

Share this

This blog is an excerpt from DevOps, DBAs, and DBaaS by Mike Cuppet.

End-to-End Response Time

Network packet collectors such as Netmon or Wireshark capture network traffic and provide information about when a request was sent until the response was received — from the network perspective as the capture is taken at the computer's network interface. The gap is the presentation layer, the processing that happens between the browser and the network on the computer. For that, using a product such as Fiddler, HttpWatch, or the Inspect function in certain browsers provide the truest timings.

Combining a packet capture and a browser-level timing makes determining slowness on the user's computer relatively easy, although only a very minute percentage of application performance problems happen at the local computer. But just because multiple users have reported slowness, you should not disregard the client infrastructure. Many companies distribute the same image to many computers, which allows for the possibility that a client-side change has caused a problem. Figure 6-3 demonstrates a packet capture, showing packet sequence, time elements, and packet descriptions.


Note: The packet capture can be analyzed further to determine transaction times.

Figure 6-4 shows a proxy capture using HttpWatch . Notice the http calls, execution duration bars, and time data.


The duration bars make it easy to see which calls take the most time. Imagine the example data being from your application with timings ten times as high. Improving application performance with just this information is possible by investigating each long bar http call.

Figure 6-5 was captured using the browser's Inspect tool , which is accessed by right-clicking the web page in a location without links (a blank space).


Note Again: The long statement durations have to be investigated further for potential performance enhancements.

Similar analysis should be done as part of the DevOps pipeline to hunt and kill performance-draining beasts. As code is tested in the pipeline against the development and integration ecosystems, collecting performance data sets enables a deep analysis of each code statement's impact on performance. Obviously, statements found violating predetermined baseline performance values must be tuned prior to production release.

Although I simply collected data using a browser call to www.oracle.com to provide simple examples, code performance analysis may require including code within the application that captures performance information or start and stop times for each execution. By now, you may realize that the types and count of the possible checks are vast; your goal is to determine which checks, done where, bring the best results.

The infrastructure example exists in real life because customers visiting the company web site expect great response times. What happens behind the curtain is the DevOps team's responsibility to deliver. If the application reaches out for credit card authorization or to validate the visitor's membership needed for a discount, that time needs to be analyzed before and after changes are made.

As shown, digging out exact timings provides a much better understanding of the application performance — broken into manageable chunks for analysis and correction.

Read Optimizing Application Performance with Change Management Improvements - Part 2

Mike Cuppett is a Business Resiliency Architect for a Fortune 25 healthcare organization, and the author of "DevOps, DBAs, and DbaaS: Managing Data Platforms to Support Continuous Integration."
Share this

The Latest

May 13, 2021

Modern complex systems are easy to develop and deploy but extremely difficult to observe. Their IT Ops data gets very messy. If you have ever worked with modern Ops teams, you will know this. There are multiple issues with data, from collection to processing to storage to getting proper insights at the right time. I will try to group and simplify them as much as possible and suggest possible solutions to do it right ...

May 12, 2021

In Agile, development and testing work in tandem, with testing being performed at each stage of the software delivery lifecycle, also known as the SDLC. This combination of development and testing is known as "shifting left." Shift left is a software development testing practice intended to resolve any errors or performance bottlenecks as early in the software development lifecycle (SDLC) as possible ...

May 11, 2021

Kubernetes is rapidly becoming the standard for cloud and on-premises clusters, according to the 2021 Kubernetes & Big Data Report from Pepperdata ...

May 10, 2021

Overwhelmingly, business leaders cited digital preparedness as key to their ability to adapt, according to an in-depth study by the Economist Intelligence Unit (EIU), looking into how the relationship between technology, business and people evolved during the COVID-19 pandemic ...

May 06, 2021

Robotic Data Automation (RDA) is a new paradigm to help automate data integration and data preparation activities involved in dealing with machine data for Analytics and AI/Machine Learning applications. RDA is not just a framework, but also includes a set of technologies and product capabilities that help implement the data automation ...