As traffic volumes rise, applications become more complex and the enterprise workforce grows more distributed, it becomes more challenging to deliver a high-quality experience for every user in every location. To ensure optimal application performance, customers need application-aware ADC capabilities far beyond basic load balancing.
The ADC selection process now includes making sure that the solution can handle simultaneously managing numerous traffic profiles dominated by higher-layer protocols, as well as providing the required processing power for a multitude of advanced features from IPv6 to SSL offload. But it can be hard to evaluate the impact of this intensive usage on the ADC’s overall performance when vendor specifications fail to reflect real-world usage—as is the case with F5 ADC products.
F5’s approach to packet processing relies on two different modes that are incompatible with each other:
- ‘Fast Path’ (FastPath Virtual Server), a dedicated high-speed technique using their PVA ASIC and CPUs with dedicated code to perform basic load balancing and other rudimentary tasks
- Standard CPU processing to run all other supported features and feature combinations
F5 reports performance levels of close to 100 percent of rated specifications when their appliances are used in Fast Path—but a large majority of real-world implementations for ADCs require the use of advanced packet processing for tasks such as IPv6, SSL offload, compression, caching and virtual server authentication. These tasks don’t function in Fast Path mode. Instead, they rely solely on iRules using standard CPU processing—at a much slower speed. In other words, while F5 talks about the performance made possible by Fast Path, your real-world experience will be determined more by F5’s standard CPU processing performance.
In contrast, independent ADC testing from the Tolly Group comparing the performance of Citrix NetScaler vs F5 in a battery of tests reflective of customer deployment and traffic behaviors showed that NetScaler delivers:
- Up to 4.8 times the performance of F5’s BIG-IP in real-world testing scenarios
- Significantly better performance when scaling multi-tenancy or multi-instance configurations
For more information, download the official Tolly Group report detailing how NetScaler outperformed F5’s BIG-IP.