Verizon's EV-DO: Addictive Speed
Continued from '2.5G and 3G (Finally) Achieve Critical Mass'….
As mentioned in my prior post, Verizon recently sent me a Kyocera KPC650 EV-DO card to review. Unlike Sprint, who offers a single $80/month unlimited data class of service (1xRTT now in most areas of the country, migrating to EV-DO over time), Verizon offers two classes of service. NationalAccess is their branding of 1xRTT and costs $60/month. BroadbandAccess is their EV-DO brand; BroadbandAccess-compatible hardware and software automatically down-throttle to 1xRTT in areas of the country where EV-DO isn't yet offered. BroadbandAccess normally costs $80/month; Verizon recently announced that they were cutting the price to $60/month for customers who are also Verizon cellular voice subscribers (Sprint predictably responded by matching Verizon's $60 price).
As you'll see in the paragraphs to follow, and as you should keep in mind when reading others' reviews (such as this in-progress series), the speed you'll experience with any cellular data network is dependent on a number of factors and therefore any conclusions drawn from a single data point (or small set of data points, for that matter) should be viewed with appropriate skepticism. Fundamentally, of course, a reviewer should first determine if he/she is connected to a 1xRTT or EV-DO tower, and should reveal if he/she is stationary or moving (and if the latter, at what speed and what direction relative to the tower) at the time the performance benchmarking is done. The distance from that tower, along with other interfering factors, determine signal strength and therefore modulate both peak and sustained bandwidth. The reviewer should also factor in how heavily loaded with other data and voice traffic that cellular tower at the time he/she does the test, and therefore how much raw speed is available to tap into. Finally, the reviewer should ensure that there's adequate bandwidth available at whatever server is on the other end of the cellular data connection, along with intermediate points along that connection, and determine what sort of data is being transferred and whether or not it's being adaptively recompressed by an intermediary proxy server.
To that last point, Verizon (like Sprint) offers the ability to on-the-fly re-compress JPEG and other image files en route to the EV-DO-equipped computer, in order to boost perceived bandwidth (at the expense of degraded image quality). However, whereas Sprint's service automatically handles this re-compression, Verizon's EV-DO employs Venturi Wireless's compression scheme, which combines proxy server-based software with a client-based utility. Venturi's client package is user-configurable in four different speed-versus-quality options, and the re-compression can also be completely turned off. All of the testing results I'll be describing in this writeup were obtained with the Venturi client disabled, to give the purest-possible sense of the service's speed potential. And anyway, DSLReport's Speed Test doesn't employ image file transfers for its 300 and 768 KByte file download, along with 50, 100 and 200 KByte file upload, benchmarking.
Last Wednesday in my Sacramento, CA home office, I installed the software on my Windows XP-equipped Dell Inspiron 700m. Here's the screenshot of the Verizon EV-DO results. Without moving the laptop, I unplugged the EV-DO card and plugged in my Sprint 1xRTT card (a Novatel Wireless Merlin C201). Here's the 1xRTT screenshot. The speed improvement of EV-DO is clear; 369 Kbps downstream and 118 Kbps upstream, versus 110 Kbps downstream and 23 Kbps upstream with 1xRTT. However, at that particular point in time, the news was not all good for EV-DO; I was seeing 400+ msec response latencies when pinging Yahoo's server, versus ~200 msec with 1xRTT. And, although I don't have a screenshot to prove it, I did experience one test where EV-DO's measured downstream bandwidth was well below 100 Kbps.
The next day, I put the Verizon software on my PowerBook running OS 10.3.9, and did the testing twice (with less than a minute delay between the two tests). The resultant screenshots reveal 315 Kbps downstream/77 Kbps upstream and 419 Kbps downstream/73 Kbps upstream performance. And the EV-DO 'ping' times were better this time too; 200+ msec. However, as you'll soon see, I don't think this day-to-day improvement reflects any fundamental Mac platform strength versus Windows with respect to Kyocera's and/or Verizon's EV-DO implementation.
Continued with 'Verizon's EV-DO: On The Road'….
