Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Technology Stocks : LAST MILE TECHNOLOGIES - Let's Discuss Them Here -- Ignore unavailable to you. Want to Upgrade?

To: ftth who wrote (2667)	12/29/1998 11:36:00 AM
From: Bill Lin	Read Replies (2) \| Respond to of 12823

Sorry to disagree with you and Ken, but that article is the worst crap I've read in a while. xDSL as a standard will probably be a G.Lite implementation. The forces moving G.Lite is too great even for the Telecomm companies to fight (an agreed standard that might work across different silicon implementations). G.Lite is a 1.5Mbps standard, like a T1. The limitation in Internet packet speed is regulated in two general ways: 1) how much oversubscription the ISP has to do to break even or not lose too much money 2) the WAN data routing speed (wrong term, but it's early). 1) The economics of being an ISP are pretty well known in the industry, but badly written about in the press. ISPs have to pay 60-70% of their revenue stream to the Telco in some fashion. With the deals that cable companies offer, that % drops to 40-50%. With Covad and Northpoint, and other DSL sitting at the Central Office, you are pretty much back at 50-90%, until a breakeven point, then it drops on a shallow curve limit function to 35% (large volume - unrealistic). That said, economics force these ISPs to provide data services using the "oversubscription" business model. You guys know this. They assume a) people don't use the internet more than 15% of the time at work or at home, b) the data gathering is bursty, and not constant c)data aggregation will not reduce quality of service based upon user profiles d) Internet speeds are limited today e) bandwidth hogs need to be identified and made to pay for their use. Let's center on d) Internet speeds are limited. Well, as you know from your visits to MP3.com or the hottest sex sites (#2/#1 on the hit lists), these sites are fast. At work, we have a T1, and we've gotten 70+kilobytes/sec on a download from MP3 consistently. True, it takes about 10 seconds to get up to that speed, but it takes a while to download a 3Meg song too, about 40 seconds. So the speed limit is increasing from 300kbps to 600kbps+ from last year to this year. As WAN backbones become smarter with the way they layout their network and limit the "hops" to the Net, then you will see 2x throughput increases every year for the next few years. Let me also go back to the economics argument. a) pipe cost (speed of through-put) is different from b) content cost (Internet access). Pipe is cheap. You can get 4M/1M pipes from US West for the same price as a T1 1.5M/1.5M. But what they don't advertise is the guaranteed content cost. This cost can range from $500/month to $2500/month for a business. (call your ISP, identify yourself as a business, and ask how much a 56K line will cost you. Early this year, it used to be $300/month). RBOCs are not used to cannibalizing their sales, which is just what they have to do, in order to improve their backbones. Right now, lots of companies want in on this lucrative opportunity, which is why they are buying DSLAMs and Central Office space (about $10k/100sq. ft/month?). These costs point to 2 things 1) you gotta have high paying customers who want or need this bandwidth or 2) if you have low paying customers, then you gotta oversubscribe like crazy. The article assumes all ISP biz models are 2), which is absurd. This is not acceptable to the Fortune 1000 companies. He is obviously talking about the consumer market, which operates on 2), which is cable modem territory (or xDSL). Which brings us to the second point of my disagreement, the WAN router speed limit. Backbone speed limits will soon be broken as soon as Savvis, Qwest, and all these other fiber companies start bringing their backbones on line. But then we get to this thread's topic, "last mile" tech. HOW they hook up their fiber will make all the difference in packet speed. Howmany HOPs to the Internet Cloud? Critical Mass also becomes important, because if you are on MCI's biz net, and the source server is also on that net, then you never leave MCI's backbone (oops it WorldCom now, huh?), and your access speed is much faster. Server limitations are also responsible for the slowdown. No matter how we tried, we could not get starwars.com to download faster than 34kbps. But when we opened up 4 downloads on our T1, we got 4 downloads at an average of 28kbps. So, I conclude, 1) the article is 1st grade WAN crap because it does not address or consider the economics of data provisioning or topology of WAN circuits; 2) this is an infant industry with so many issues to fix that it is easy to focus on one issue and hyperbole it to death. Supply/Demand is the equation. If the cost to upgrade the WAN for high speed data access is $250 billion, and the maintenance cost is $50 billion/year, how much are you willing to pay? Sorry I ranted, but that article was shallow drivel. BL

To: ftth who wrote (2667)	12/31/1998 1:08:00 AM
From: George T. Santamaria	Read Replies (1) \| Respond to of 12823

<.... speed" connections have been so successfully brainwashed by the providers' marketing hype that they'll swear up and down they're getting "lightning fast" downloads and that it's 100 times faster than their old telco modem.> I've been brainwashed by the experience of having a Cox@home connection. Anything I get from Yahoo has an imperceptible delay 95% of the time. You ought to get one. Between the noisy phone line and the ISP network congestion, my old dial-up connection was a lot slower than the advertised 56k. "Up to 100X" is no baloney but 56k data rates are baloney. Sometime the dial-up ISP would slow down from a typical 20kBaud to 1k.