- Image via Wikipedia
In April 2009, Novarum published a comprehensive test that pitted Meru Networks against the two largest microcell vendors, Cisco Systems and Aruba Networks*. Results show that Meru offers superior performance measured by every metric. Meru won each test in the bakeoff, but four results stand out:
- Data Performance: When ten 802.11n clients were connected to a Meru access point, Novarum measured total TCP throughput at more than 180 Mbps. The comparable figures for Cisco and Aruba were at least 40 Mbps lower; a difference maintained in others tests. These included measurements of overall throughput in mixed mode (802.11n with 802.11a/g) and converged networks that carry both voice and data.
- AirTime Fairness: The differences between Meru and microcell vendors are even more extreme when looking at the worst (slowest) client. The slowest of the ten-802.11n clients on the Meru network achieved 15 Mbps, only 20% less than the fastest. The slowest Aruba and Cisco clients both had throughput in the single digits, demonstrating that microcell vendors cannot offer consistent performance even in 802.11n-only networks. In a mixed mode network, Meru let each client transmit at its own data rate. The microcell vendors both starved legacy clients, meaning that upgrading to 802.11n actually degrades 802.11a/g performance with Cisco and Aruba.
- Voice Quality: In a converged network with ten voice and ten data clients, the voice clients on a Meru network averaged a Mean Opinion Score (MOS) of 4.3 – near the theoretical maximum for VoIP, and above the 4.0 generally considered toll quality. Both microcell vendors scored below 3.0, the threshold figure for sound quality intelligible to listeners. Furthermore, Meru achieved the same high quality sound in both directions. Microcell networks favored upstream voice at the expense of downstream.
- Power Consumption: Meru access points consume less power when active than either microcell vendor does in any power state, even idle. In Novarum’s test, a Meru access point needed only 4.7 W when idle and 8.0 W when active. Microcell access points needed at least 9 W when idle and 10 W when active. Meru’s performance per watt of 21.5 Mbps/W is more than twice the microcell average.
Old news? Maybe. But the story is what is the Meru “magic” (our competitors’ term, not ours) that makes this happen. Just like I tell my kids, if you understand the science behind it then you’ll realize its not magic. When you don’t, it can definitely look like magic.
The names of the particular comeptitors doesn’t matter because the “magic” comes from building technology from a fundamentally different philosophy. Thats how we arrive at Air Traffic Control technology. That is what we call the portion of our system that effectively takes control of the RF and coordinates APs, clients, and the backend fabric at a micro-second level and uses patented algorithms make sure performance and security are maximized. For years the competition has tried to disparage us saying the system is proprietary and non-standard. But that couldn’t be farther from the truth. The reality is that our technology uses all the existing, standard 802.11 mechanisms available to it to implement the algorithms that coordinate the traffic.
No magic. Nothing non-standard about the traffic on the network. How did we do it? We approached the creation of our system with two main premises:
- Co-channel interference is inevitable when wireless is deployed pervasively.
- Nothing we do can ever, EVER violate IEEE standards. (Because even Cisco wouldn’t be able to violate a standard and expect to sell a product successfully)
In 2002, before we launched Meru, rather than retro-fitting small office access points with bigger processors and more bells and whistles, we started over. You see, Wi-Fi was designed as a small office, home office technology that became so wildly popular that it made its way into the enterprise, but it wasn’t originally designed with the assumption that APs could communicate through a backend infrastructure. So most of the control of what happens over the air (who transmits, who hands off, where the station hands off to, etc…) was left up to the client. Take that underlying philosophical approach and just deploy with for pervasive coverage and like a house built with poor framing material, you’re going to have a very unstable environment.
At Meru, we simply said from day 1 – “that can’t work as a pervasive network”. The analogy we use around here is the cellular telephone network. Can you imagine what would happen if the phone network were left in the hands of the phones? How many dropped calls would you get then? Instead, the phone network takes control of who transmits when and to where and does it at the micro-second level. And we went on from there.
Now, years and many patents later, our understanding and control of what happens to the RF spectrum in a pervasive 802.11 network allows us to create all sorts of advantages . With the introduction of 802.11n and MIMO, as this testing shows, the control and ability to make micro-second adjustments without the clients getting involved makes the user experience seamless, makes the network more stable, and allows network managers to get back to the job of managing networks, not RF spectrum analysis.
* Download the full report, Enterprise 802.11n Wireless LAN Access Point Benchmark, at: http://www.novarum.com/publications.php
