On March 1, 2015 at 16:13 nick@foobar.org (Nick Hilliard) wrote:
On 01/03/2015 03:41, Barry Shein wrote:
On February 28, 2015 at 23:20 nick@foobar.org (Nick Hilliard) wrote:
there were several reasons for asymmetric services, one of which was commercial. Another was that most users' bandwidth profiles were massively asymmetric to start with so it made sense for consumers to have more bandwidth in one direction than another.
How could they have known this before it was introduced?
because we had modem banks before we had adsl.
And you are asserting that studies were done on user behavior over dial-up modems in order to justify asymmetric service? Well, maybe there was some observation and conclusions from those observations that people tended to download more than they uploaded, it's not inherently hard to believe. I'd've had questions about how well 56kb theoretical max predicted behavior at ~10x higher speeds of *DSL. But whatever you work with what you have. I still think a lot of the motivation was to distinguish residential from commercial products. We are talking about a product sold by regional monopolies, right?
I say that was prescriptive and a best guess that it'd be acceptable and a way to differentiate commercial from residential service. Previously all residential service (e.g., dial-up, ISDN) was symmetrical. Maybe they had some data on that usage but it'd be muddy just due to the low bandwidth they provided.
maybe it was symmetric on your modems; it wasn't on the modems I managed.
Bandwidth or usage? Are you changing the subject? I was talking about bandwidth, bandwidth on dial-up modems was symmetric or roughly symmetric (perhaps 53kbps down and 33kbps up was common, effectively.) Which is why I said residential SERVICE ... was symmetrical.
It was the combination of asymmetric, no or few IPs (and NAT), and bandwidth caps.
let's not rewrite history here: IPv4 address scarcity has been a thing since the very early 1990s. Otherwise why would cidr have been created?
Because Class A/B/C/(D) was obviously wasteful and inflexible compared to CIDR so it caught on. Yes some were projecting an eventual IPv4 runout 20+ years ago, and IPv4 was a cost factor particularly if you were planning on deploying millions of clients tho not a killer. At any rate NAT played well into the hands of any company which wanted to distinguish a residential from commercial IP service, only a tiny per cent could see their way around a non-static address via DDNS etc.
Sure. once it became institutionalized and the market got used to it why not sell tiered bandwidth services at different price points, but that could have been true of symmetrical service also.
my point is simply that there is often more to asymmetric services than extracting more money from the customer.
Ok fine. But don't present it as if it never crossed the minds of telcos and cablecos that asymmetric service, no static ips, etc distinguished residential from commercial service. They do include all that with commercial services, right? Well there are these small business "commercial" services particularly from cablecos which are hybrids, asymmetric bandwidth with static IPs etc. It was a challenge early on, the internet particularly in those days just didn't distinguish such thing as residential vs commercial, bits were bits, other than raw link speed perhaps and even then some were buying 9.6kbps and 56kbps nailed-up leased lines for $1,000+/month while others got that kind of speed over dial-up modems for $20/mo (plus POTS) and faster (128kbps) over ISDN for around $100/mo or less. A very early way to distinguish was idle-out, if you weren't sending traffic you were dropped either from dial-up or your ISDN link shut down or whatever. And someone sending at you didn't (unless you had some exotic set-up) bring the link back up. Some sites would just drop your link if you were logged in more than so many hours straight (trust me on that) to see if anyone was really there to log back in, automating that was way into the few per cent. I had an ethernet switch at home with a built-in 56kbps modem which would keep a dial-up link up, keep redialing if it lost it. In theory it should have worked, in practice it was crap. But that was probably more like 1997 when consumer products catering to this stuff really started hitting the market (other than just modems.) So you couldn't run always available servers from those kinds of services, not even an SMTP incoming server unless you adapted to that, after a few minutes idle you went offline. Some of that was resource conservation but a lot of it was to differentiate residential from commercial service. You want to run a server host it somewhere that sells that or buy an always up link (e.g., leased line.) To some extent this is six vs half a dozen. One reason commercial servers were discouraged was that they used more resources, you weren't just reading your email you were running Amazon! That wasn't sustainable for $20/month and billing for actual metered usage was a PITA for other reasons. So you said ok, it's $XX per month, no metering, but here's how we're going to make it nearly impossible for you to run Amazon off your cheap link: Asymmetrical, NAT, bandwidth caps, port blocking, etc. ********* BUT WHAT WAS THE POINT? ******************* Not to do internet archaeology. It was originally just making a point that asymmetrical service was kind of a hack, a kludge, something driven by limited resources and early attempts at market differentiation, and not inherently desirable as a future. I didn't bring that up I just agreed. It's not even clear what asymmetric service in those terms means in a world where you'll be offered 100/25mbps or 100/50mbps links, that's a far cry from 1m/256kb links. -- -Barry Shein The World | bzs@TheWorld.com | http://www.TheWorld.com Purveyors to the Trade | Voice: 800-THE-WRLD | Dial-Up: US, PR, Canada Software Tool & Die | Public Access Internet | SINCE 1989 *oo*