Why don’t most enterprise WiFi access point vendors tell you what’s inside their AP? They don’t publish which WiFi chipset the AP uses, or the CPU specification. At best they state the amount of RAM. When you evaluate APs for your deployment, you should consider hardware components. Hardware components and the software running on it will impact the AP’s performance and user experience. The test results below demonstrate this.
An AP’s hardware platform is the foundation for all it abilities. Both design and components can affect performance. Two of the most critical hardware components for performance are the CPU and the WiFi chipset.
If the CPU doesn’t have enough horsepower, performance will suffer and/or it will have limited capabilities. It is critical that there is enough CPU headroom for all of the application processing the AP needs to do for the life of the AP. A just-barely-capable CPU today will falter as current features are implemented and new applications are added.
There are only a few manufacturers of WiFi chipsets used in enterprise APs: Qualcomm, Broadcom, and Marvell (sort of). The chipset must change when new technology is introduced (e.g.: 802.11ac Wave 1 and Wave 2), but also to fix bugs, increase performance, or reduce cost. (Often, all of the above.) Of course, it’s hard to have the best-performing AP without the latest generation chipset.
An AP could have the best hardware components, but if the software isn’t architected properly to use it efficiently and effectively, it will not perform well. As hard as it is to determine the hardware components used in an AP, it is almost impossible to determine how well the software is architected. This is why performance benchmarks are so important. Benchmark tests expose weaknesses that will impact real world performance.
Comparing Qualcomm-Based APs
We tested the best Qualcomm-based enterprise class 802.11ac Wave 2 4x4 APs available at the time from each major vendor. As I was reviewing the benchmark test data, I was struck by the big difference in performance for similarly-specced APs.
All APs used the same standards, the same settings, and were optimized for performance. They were all tested on the same network infrastructure, in the same environment, using the same clients at the same distances.
Of course, not everything is the same. APs use different generations of WiFI chipsets, hardware components, and software architecture. Here is a summary of the AP specs:
The test we are highlighting here is one of pure performance. It uses 30 802.11ac Wave 2 1x1 clients downloading data as fast as possible. This exercises the whole system, hardware and software, to expose AP weaknesses.
The WiFi chips are clearly not the whole story. The Aruba IAP-325 and the Ruckus R710 use the same Qualcomm chip (QCA9990) but get very different results. This marked difference could be because of lower performing hardware component(s), software overhead, both, or something else altogether.
What is clear is that the Mojo C-120 with the latest generation Qualcomm chipset (QCA9994) with a dual-core CPU wins hands down. It has outperformed every AP we have tested to date, in almost every use case we have tested.
It takes good WiFi chips, good hardware components, and good software architecture to make a great AP. We have all that.
Don’t take our word for it. Test one for yourself.