Comparison testing ADCs: a real-world approach
Real-world ADC testing: get our recommendations Download
ADC testing: building scenarios that reflect reality
You know what they say: Garbage In, Garbage Out (GIGO).
It’s a saying that’s particularly relevant to testing ADCs. When you test, you need to set up parameters that reflect the real world, rather than a scenario that flatters one vendor’s offering over another.
We realize that this isn’t always easy. Which is why we’ve collaborated with Tolly Group, the leading independent test lab, to develop a methodology for testing ADCs that gets as close as possible to real-world deployment scenarios.
Three tests that keep it real
We suggest using this methodology as a starting point if you’re in the market for new ADCs this year. Here’s a summary of our recommendations:
- For all competing solutions, consider running three tests: simple load balancing, typical web application delivery and secure web application delivery.
- The simple load balancing test should also take account of the need for connection logging, source IP preservation and cookie persistence.
- For the simple load balancing test, we prefer to use a set-up with backend server pools delivering “specialized” content (e.g. one pool for mobile browsers, another for desktop browsers).
- For testing typical web application delivery, we recommend deploying functionality that has become standard in real world conditions, e.g. removing information about the back end servers from headers, inserting client IP addresses into requests so that backend applications can maintain client visibility.
- Ensure that all DUT software is up to date.
- Take care over response sizes. Taking our lead from Google’s analysis of general web traffic, we recommend a mean for response size distribution of 7.19KB (skewed left).
To learn more, take a look at Tolly Group’s detailed report, “Citrix NetScaler ADC: Real-World Application Delivery Evaluation Versus F5 Networks”. We hope you find it useful!