As I've mentioned previously, I've been included as a tester in the Covad Blogger Relations program. Part of that testing involved setting up a Covad T1 circuit as my Internet connection. I wrote up some initial performance baseline testing I did here.

Andy Abramson and I did some video comparison testing using SightSpeed for voice and video. In this post, I want to share what we did and what our findings were. This post includes four videos, so you can see each of our tests. The videos include some of our conversation about the link in use at the time and the experience overall. I've also posted some conclusions at the end of this post.

I'll also talk about the idea of telepresence for the masses.

The Network Environment
Andy has two connections - Cox cable and a Covad T1. Ken has three connections - Qwest DSL, Comcast cable and a Covad T1.

Test #1 - Ken's Qwest DSl to Andy's Cox Cable
In this test we are both connected via WiFi in our home office to our T1 access point.

As we see in the video, there are issies with both audio synchronization and video freezing and pixelation.

Test #2 - Ken's Comcast Cable to Andy's Cox Cable
In this test we are both connected via WiFi in our home office to our T1 access point.


While there's marked improvment in both the audio and video quality, the issue with synchronization, frame freeze and pixelation are also present in this test.

Test #3 - Ken's Covad T1 to Andy's Covad T1
In this test we are both connected via WiFi in our home office to our T1 access point.



We were surprised to see that while there was a very noticable improvement in quality, there were still issues. If you watched the entire video, you heard my thoughts on what the root cause might be. This led us to an unplanned fourth test.

Test #4 - Ken's Covad T1 to Andy's Covad T1
In this test Andy remained connected via WiFi, but I shut off the WiFi radio in my laptop and plugged an Ethernet cable in the Linkys router that serves as my access point.




The change, as you can see, was tremendous. Impressive to the point I'll share our conclusions.

About the Technology
DSL is a shared media technology. While the twisted pair to the end user is dedicated, and may provide some level of bandwidth, that guarantee is really only assured to the DSLAM (DSL Access Multiplexor) at the DSL providers point of presence (POP). For DSL users, that's typically in your telephone company's central office that provides your dial tone. In some cases it's co-located by a DSL provider or in another location.

DSLAMs serve a large number of customers. The DLS lines are aggregated and share some higer capacity link upstream to the Internet. There can be contention issues in a heavliy loaded DSLAM. There can also be what are essentially oversubscription issues by architectural design. If a DSLAM supports 128 users and has an upstream link of 10 Mbps and everyone's active at the same time, each user may in principle only get 1/128th of the 10Mbps stream. That means DSL performance may well mirror dial-up modem performance during peak traffic periods.

Cable modem technolgy is a different architecture, but again, the user links aggregate at a cable "headend" for service delivery. Once difference between cable and DSL is that DSL provides a dedicated link over the phone line twisted pair. Cable modem architecture may in fact provide a data stream that passes a number of homes to get back to the cable headend. We share capacity in the coax or hybrid fiber-coax that delivers the cable service. In a cable modem network, your network neighborhood is, in a sense, your real neighborhood.

In both DSL and cable architectures, the bandwidth is typically asymmetrical. That means we get more bandwidth for information coming from the Internet to us that we have for sending information upstream to the Internet. This design made sense in the early development of the technologies. These were designed as consumer services. Most home users recieved far more than they transmitted. They didn't run servers that provided service to other Internet users. With peer-to-peer technologies on the rise, and the increase in sharing audio and video, coupled with advances in gaming technology, this asymmetric approach may not be the best solution any longer.  

The Covad T1 is a frame relay T1 circuit connectied between the customer premise (our home offices) and the Internet. There is no sharing. It's a dedicated symmetric link. It's 1.5Mbps in each direction all the time.

Another Wrinkle - Wireless Spectrum
It's well known that wireless connections aren't as fast or efficient as wired. We use wireless in our home offices for convenience. I know that if I plug an Ethernet cable into my router, I get fast performance that an 802.11 link.

There's also an issue with spectrum contention for WiFi. As you saw in the video, I run four WiFi access points, all in the 2.4Ghz spectrum. And my neighbors also run WiFi. There's a channel capacity and spectrum overlap issue that degrades performance in current wireless technologies. And our tests prove that it's measureable to the human eye and ear.

The great the population density of wireless access points, in an apartmetn complex or tightly packed business park, the greater the likelihood of contention.

If performance is truly a key issue, you may want to consider using wired Ethernet for workstations that need the higher level of performance in some situations.

Conclusions
First and foremost, Covad's T1 service absolutely rocks. When the Covad Blogger Relations program ends, I'll convert the circuit to mine and become a long-term customer. If I take into account what I pay Comcast and Qwest for Internet today, I'm already spending enough money. Neither is business class service, and it shows. But until you really compare side by side, you may think that what you're getting is good enough.

For me, the Covad T1 proved conclusively that my business needs warrant moving the the right technology, and consumer services dont' cut the mustard.

Why? I'm sure you're asking why I can even think I need a T1 in my home office. That's easy - telepresence for the masses. I do a lot of VoIP calls, on a number of services. Even on my own Asterisk box at times depending on how my lab is set up. I also do a lot of video. And I transfer large files, both audio and video, all the time.

In Test #4, it's really obvious that this approach enables telepresence at an afforable level for the small business and many home office-based entrepreneurs. Cisco may have the buzz on the workd telepresence. At least they'll tell you they do. They want to own the concept. But like so many vendors of that scale, they built a solution for the Fortune 500. Cisco's telepresence is for the deep pockets of large enterprises.

I, like any other small business, will never be a Fortune 500 company. We are forever trapped in the Unfortunate 5 Million. We're the great unwashed masses left to fend for ourselves. Cisco's telepresence isn't a tool for me. But Covad and SightSpeed are within reach. And while I'm a SightSpeed afficonado, there are plenty of other video tools that will work too.

Covad enables unfied communications for the SMB market. T1 connections to the Internet have been pretty common for many businesses. There are plenty of offices with 50 people sharing an Internet T1. What Covad's doing is bringing that capacity into the reach of smaller businesses and entrepreneurs. Covad's enabling us to deply unified communications solutions on a scale we can afford at a quality that provides the right level of telepresence for our needs.

I've been using the Covad link as my primary connection since it went in. it's not the first time I've had a T1 to the house, but this is clearly the best experience.



Technorati Tags: Covad, Blogger Relations Program, Covad T1, telepresence for the masses, telepresence, unified communications, video, video collaboration, VoIP, SightSpeed. Andy Abramson, Ken Camp