|
No matter how well we design, a lot can happen in the real world of wireless: Spectrum interference will be different than predicted; the stateful behavior of wireless clients requiring a high level of protocol acknowledgments will mask network issues at higher layers; and on and on.
And the more complex things get, the more difficult it is to achieve desired end-user Quality of Experience (QoE), and the more we need sophisticated software mechanisms to be designed in from the beginning.
Of course it's not possible to anticipate everything that might go wrong from the lab. But strategic testing throughout the design and development of wireless networks goes a long way in minimizing the pain of deployment and user disruptions when things go live.
The challenge is to test as thoroughly as possible while at the same time speeding rather than delaying time to market; and, to decrease rather than increase overall cost.
Traditionally, two aspects of real world wireless performance have been particularly difficult to assess from the lab: scalability and how various applications will interact with each other and with the network. And for the next 18 months, an additional challenge will dominate; namely, the introduction of 802.11n-complaint devices alongside existing a/b/g equipment.
While the home environment will smoothly move from a legacy wireless environment to 802.11n, the real promise—and challenge—for 802.11n is the business environment.
Here, the wireless network must deliver a quality of service sufficient to guarantee support of mission critical applications. And with wide area wireless networking on the horizon (WiMAX, LTE, or other), fixed mobile convergence will need to seamlessly integrate these WWANs with WLANs.
Again, no amount of testing can completely eliminate surprises when new technology and new equipment enter a live network. Fortunately, there are a few sound strategies that can maximize QoE while minimizing surprises.
Scalability: Don't limit yourself to laptops
Even now, wireless testing too often revolves around hooking up a handful or so of laptops, PDAs, IP phones and other mobile devices and blitzing the network with traffic. This is informative, but gives no visibility into the nuances of large networks.
In particular, the three things to test when designing for scalability are:
- The effects of a large operational network on the level of service a client receives.
- The degradation, or possible complete disconnection, of wireless sessions when the network load increases rapidly.
- Proper traffic differentiation and enforcement of quality of service mechanisms when a wide range of applications (such as web browsing, e-mail, CRM, scanner traffic, voice calls, streaming video and audio) all demand network services.
The optimal method of testing relies on mixing real devices with a test platform that generates large numbers of clients and enables configuration of traffic conditions as well as automating repetition of the test with the same or varying conditions.
One of the biggest challenges in wireless networks is the continuous change in the physical environment. To properly design the infrastructure network components, a repeatable and tightly controlled environment needs to be established.
Only such an environment can be used to properly subject the network to a broad range of operating conditions with a constantly changing number of connected users, and a constantly changing mix of applications used.
|
