|
This is mostly a review, but also a "HOLY CRAP" statement.
Last month I was offered a deal on a new speed of broadband service being offered by Cox Communications, and I took it. I had been operating at 4 Mb/s before, and as of last night, I was bumped up to 9 Mb/s. I didn't expect much because over speeds of about 4-5 Mb/s, one starts running into barriers other than one's personal connection that limit the effective speed of the connection. I was fairly certain I would notice no difference in casual web browsing and expected a slight speed increase when downloading large files. I expected the real performance boost to become apparent when downloading two or more large files and when networking two or more computers that would share the bandwidth. For the most part, my expectations were met, but with some exceptions.
Web browsing many sites is actually a bit faster. Using DU's home page as a testing page, I had checked download/rendering times for loading the full page several times in the past and had calculated a most recent average of 1.8 seconds over 10 separate loads of the page using no cached images. With the new connection speed, that average dropped to 1.4 seconds. This of course is hardly noticeable, but it is an improvement. To get a practical picture of this improvement, I used a testing scenario that loads single, similarly sized graphic images from several sites at once. Prior to the speed upgrade, a 1 minute test completed 10 times, loading 50 images simultaneously, I averaged 482 KB/s or 3.8 Mb/s. At the same time of day, using the same test, I averaged 4.9 Mb/s with the new speed. Over a session of browsing, especially when loading large pages with lots of graphics, I suppose one would notice this. (BTW, I used the home page because it remains fairly consistent in size and doesn't invoke the same the same server code that slows things down when accessing the forums.)
Next, I tried downloading single, large files. For testing, I used a ~20MB .mpg file so that file compression wouldn't be much of a factor. Again doing this ten times, the speed of the download with the new connection averaged 6.5 Mb/s. Prior to the upgrade, I had averaged 3.7 Mb/s for the same file from the same server, close to double the speed, but still not quite utilizing the 9 Mb/s fully. An interesting note here that shows how bottlenecks in the network affect effective speed became apparent during this test. I had a monitor running that updated the actual download speed every second. That speed ranged anywhere from 1.2 Mb/s at the start of the download up to as high as 11.52 Mb/s at some points. (As advertised if you read the fine print, any high speed connection's speed is considered an "average top speed," and this test shows part of what they mean. If I'd done this in the middle of the night, the speeds likely would have fluctuated even higher, which is another part of the averaging equation.)
As expected, the biggest improvement in speed was realized when downloading several large files at the same time. Using four separate FTP connections, I downloaded four separate ~100 MB files from four different servers. My download speed quickly normalized across the four connections so that it remained fairly constant during the entire download, giving a final average of 8.1 Mb/s. I had never intentionally tried doing this with my prior connection under circumstances in which I was taking notes, so I have no idea what speed I was getting before, but I know it wasn't that high.
And on that last note, HOLY CRAP. I've gone from a 300 baud modem i nthe beginning of my "online" days that displayed text from a BBS so slowly that I could actually read it as it appeared to this. It's something of a shock to my system, but I like it.
|