EV-DO 'A' From The San Francisco Bay: My, My, My...Watch The Bits Fly!

Greetings from Woodside, CA. Earlier this evening, as previously promised, I re-ran my EV-DO Rev. 0 and A testing, this time on both Sprint and Verizon's service. I was in downtown Mountain View, CA, at Dana Street Roasting Company. For the full effect of the (generally) increased performance here versus last night in Sacramento, make sure you read both this writeup and my prior report.

First off, here are the BroadbandReports results of my three consecutive tests of Sprint's EV-DO Rev. A network, again run against MegaPath Networks' San Francisco server.

Ack latencies for pings of Yahoo's server averaged ~180 msec.

Next, let's look at Verizon's EV-DO Rev. A service, which as you'll see from the numbers below, was definitely operating in Mountain View.

The first ping-to-ack latency in this case was consistently ~500 msec, but subsequent ack latencies were only ~110 msec.

For comparison sake, let's look at Sprint's service at that same location, with my prior-generation EV-DO Rev. 0 card

Ack latencies here averaged ~280 msec, with random and occasional pushouts beyond that time

And finally, here's Verizon's EV-DO Rev. 0 results

Here, latencies were consistently ~170 msec.

Finally, since I was in the heart of Google's municipal broadband Wi-Fi footprint, I decided to re-run my rudimentary testing of it, first conducted last summer. I was underwhelmed then, and I'm no more impressed this time around. Keep these results in mind when, in the near future, I talk about wireless VoIP.

I was a half block from downtown Mountain View, and from a Wi-Fi access point mounted on top of a service pole at an intersection, yet I only (and only occasionally) got one bar's (out of five max) worth of reception strength from inside Dana Street Roasting Company. Stepping outside, I achieved between 3-4 bars of signal strength, and I snapped this photo of the access point to prove how close I was to it, and what my line-of-sight proximity was to it.

Here are the BroadbandReports results:

Downloads were reasonably interruption-free, judging from the activity pattern presented on Task Manager's Networking tab. Uploads, however, were hit-and-miss, as the above results exhibit. The ping tests of Yahoo's server were equally erratic, generally between 10-20 msec but with frequent excursions to 50 msec and beyond.

With all of the above results, I'd encourage you to not get hung up on the absolute numbers. On different days, at different times of day, at different locations and with different EV-DO cards, the numbers you get might be somewhat different. However, if you compare the EV-DO Rev. A results with their EV-DO Rev. 0 predecessors, I think you'll agree that the newer technology delivers tangible advantages, both in upload and download speed and in latency. I'm not sure why my Sprint results in Mountain View this evening were so much faster on average than they were in Sacramento yesterday; I'll ask my Sprint PR contacts for clarification and will report back anything interesting I uncover.