It really is the party-pooper solution, as it's so low-tech, but when we priced it out, for most buildings Cat5 wiring is cheaper.
Depending on what kind of walls you're working with, (drywall vs. brick, etc) i've gotten quotes from roughly $30-100 per drop in an apt. Add to that $40/port for a good switch, and you're looking at $140 per room. And good cat5 contractors will give you some ungodly long warranty, on the order of tens of years.
Contrast this with 802.11. You have to pay for multiple APs (500~2k each depending on what you want/need), then you either have to a) pay for the 802.11 card for each pc and have the tenants pay a deposit (which was ~150ish when i priced them out, 100ish if they had a laptop) or b) force the tenants to buy their own. From doing some informal surveys and asking around, the latter wont work. Then you have the line-of-sight problem (the computer has to be kinda near the window for them to pick anything up), the rf interference issue, and other funky stuff rf physics stuff. Not to mention you're on most likely a 1yr warrenty, and have to deal with helping people get their wireless card working, which can be a huge pain in the ass as likely they'll be using one of those pcmcia-pci slot converter things.
Furthermore security-wise, you honestly cannot beat having a plugged vs. not-plugged-in port, thus you can assure people are not stealing your service... A good switch will tell you what mac addresses are coming from what port, so with some good accounting on the side, you can tell exactly which apt has a hub and is sharing with their neighbors, etc. It also makes catching troublemakers (and there will be some, trust me) a lot lot easier, as you can pinpoint it to the room, not just to a mac address.
I more or less planned/ran a campus apartment project like this, and we did at first also seriously consider the 802.11 alternative, but quickly threw it away as we realized that a) it was going to certainly cost more long-run in labor than cat5 would,and b) it most likely wouldnt save us money upfront either.
Yes. Save for the monthly subscription revenue, there was nothing much to the AOL business to begin with, as the first mild downturn in the economy has convincingly shown, with advertising revenues from the service having now collapsed in a heap."
Furthermore, the "save for the monthy subscription revenue"... Dude, the monthly subscription revenue is what makes the company so attractive. If you can count on $21 * (areallylotofusers) / mo. guarenteed, you're already way better off than say, the services industry which is a lot more affected by economic downturns.
The monthly subscription revenue is what separates AOL from your typical ad-market company, it has a large revenue base that i can use to cushion the impact of advertising revenue, which looks to be very cyclical.
Once the ad-market has an upturn, AOL/TW is going to be quite fearsome. They have the ability to do a huge branding effort. Think how MSN teamed up with espn.com. Then MSN teamed with GE for MSNBC. Now multiply that by 20, given the TW media assets. They would have the ability to keep people within AOl/TW owned websites, further increasing ad-revenue. If they so chose, they would have the ability to create 'aol-subscriber' only content, much like they did in the late 90.
Just because things arent working out perfectly doesnt mean a) it was a bad idea, or b) that everyone should abandon ship. While TW was probably charmed by the large amt of stock thrown at them, and soldout for too low, the merger was certainly not as horrible an idea as the article makes it sound.
I connect to UUNet, and they connect to me. In that way we usually use the same amount of each others network, and thus we charge each other the same price for connection )
Sure, that's called bilateral peering, however to do so, it requires that there be roughly the same amt of traffic going in each direction... This is not the case w/ @home/othercableproviders, as they have much more 'download' traffic than they have 'upload', ie they need uunet more than uunet needs them. Although honestly i dont know for sure, i heavily doubt that CableServiceProviderX doesnt pay some fee w/ their peering arrangements.
Why does this perceived "problem" have anything to do with management, or the problems w/ certification? This "problem" the poster perceives is entirely not a problem of technical incompetence, but rather the simple fact that to this isp, supporting these tlds is simply not worth the time/trouble/money.
Webcaching is highly cost effective. It allows providers to charge this $35-50/mo charges w/o going under, or without losing too much money.
As someone with 7 years programming experience, 3 years networking experience, and did months of research on webcaching before implementing it, i can tell you for a fact that I would not think of this as being a problem during implementation. Hell in fact i still don't think it's a problem, quite frankly taking down a router for 30 minutes just to fix one person's obscure problem just isnt worth it. You're paying what? $20/mo? Great if you stay for 6 months then you'll pay for my 1hour of overtime. and that's not counting how many other people we piss off for having this 30min slowness/downtime.
So in response to your post. This was not a bad decision by any means. Done correctly it has the ability to speed up response times (and my data shows that it does) and saving money by using much less bandwidth (btwn 10-40% even while full honoring of if-modified-since headers). It improves service to 99.999% of their customers and pisses off one, because he can't use some esotericthingthattheISPdoesntwanttosupport anyways. Please.
Your points over worthless certification and stuff like that certainly could be valid points if you had applied them a relevant arguement, however this is entirely not, as you have no position/ability to judge that this was indeed a bad decision, or that it had anything to do with (unknowledgable) management decisions.
Cisco wccp/gre tunneling had large amounts of bugs until fairly recently (1-2 years ago), and still require specific ios versions. This bug would cause the gre tunnel (which connected the router to the cache) randomly collapse. That is probably the bug you are seeing, and is entirely unrelated to that the poster's problem is.
If the web proxy can't identify the hostname present in the URL, it should simply pass it through, allowing the client (who already knows the IP), and the Web Server (who also, clearly, already knows it's own IP) to communicate.
Not a horrible idea, i'm fairly doubtful it will work though.
The problem is the caching architectures i've seen have the exit router communicating to the cache through a gre tunnel, so the cache has no way to actually let it "pass through" as the cache isnt the actual box doing the routing. This is vastly superior to "put proxy between person and internet and let it route too" because of loadbalancing and failover reasons, but doesnt allow what you're asking for.
I dont really consider this 'broken proxy' nor do i consider it a misconfiguration, the simple fact is that his ISP (as well as mine) denies the existance of the opentld system. If he really has a problem with it, he should find another provider. Expecting the ISP to bend over backwards to satisfy their.0001% their client base that this concerns is rediculous.
Ok i'm not affiliated w/ the RIAA, in fact i really dont like them, but i also really dont like when/. puts out this biased crap with flawed logic. Here's a response to a few things jamie said:
The SF gate story: who cares, correlation in no way indicates causation. Bad economy coupled with releases that didnt have mass market appeal (a lot of r&b it seems) could easily explain it regardless of napster.
I guess 77 percent are buying more music because they're downloading it for free? Great. If 77% are buying one more cd, while the 23% are buying 8 less cds, then you have a net loss. This figure cited proves nothing.
Most industries, faced with declining sales, try lowering their prices. Not this one. Of course not. Music demand is highly inelastic, they can maximize revenue by raising prices, not lowering. Crack dealers (also in an inelastic industry) do not lower prices to increase revenue.
You know music officially sucks when the labels have to pay someone $28 million not to sing. Or it could be that Mariah has enough personal problems that it would be cheaper for them to pay the 28million than to keep her under her much more expensive contract. Unfortunately Mariah has fallen down from her huge-popularties of the late 80s early 90s, and her sony contract was priced assuming her massive popularity. Sony was merely cutting their losses on her, and is in no way indicitive of the rest of the music released.
Also, every "napster helps sales" argument i have seen is purely anecdotal, or if not, it lacks the numbers needed to prove anything. Furthermore all these surveys have a high bias (people feel more compelled to say they buy more music, than buying less). If you were from say Gallup and flagged me down on the street asking me if i bought more/same/less of music after using napster, of course i would say more, as i wouldnt want (napster|kazaa|morpheus) to go away.
Yes and no. I disagree with your labelling of the action 'knee-jerk'. Here is the dillema:
Bells have no incentive to expand (geographically) their dsl capabilities. If they invest in new ATM upgrades in say, akron ohio, they will do so at a loss, since some/much of their profits are going to be taken away by the CLEC, who do not incur any of the capital cost.
With no incentive to expand, much of the US will never get broadband.
Note that in addition to everything that everyone hates about the bill, the bill also forces the ILECS to provide coverage in 100% of the central offices with in 5 years. Telco is not making bazillions of dollars like everyone thinks they do, and they need some type of incentive to expand, this will give them the incentive/motivation to do so.
So while everyone is saying "monopoly = bad", and "why doesnt the entire nation have broadband yet?" you really get one or the other. If you want nationwide broadband, you have to make it profitable for the ILECs. Girnomous federal subsidies would be the other solution, but would be enormously harder to get through congress.
Honestly, i dont know if i agree with this bill either, but i certainly see what they're trying to solve, and how they're trying to solve it. Please, no more of this "all of congress is evil evil evil corporation bloodsuckers" and "this bill and only benefits the bells". While the email tries to spin it saying "One would think that it would be hard to find people who support this bill", reading the entire bill it makes perfect sense why someone would. Nation-wide broadband.
ISPs have huge fixed costs. city-wide isps are in the tens of thousands of dollars, regional isps in the millions. High fixed cost, while the cost of adding additional subscribers is quite small, you have textbook economies of scale.
Then you have the bulk discounts on bandwidth, which are *huge*. Buy your first T1, it costs 900-1kish. If you are near a NAP and need an oc-3, it only costs. hell if you're big enough you dont even need to pay for commodity bandwidth, you just do bilateral peering w/ the uunets of the world.
Also when you said "thrived", i dont know what you mean by that. If you mean "barely turning a profit thanks to razor thin margins and huge telco tariffs", then you're probably correct. Furthermore, actaully had to maintain modem POPs, pay for PRIs, call forwarding, longhaul, etc, while now for cable|dsl, all you have to do is throw in a dslam, a big router, and buy one piece of fiber to your nap.
Honestly, this whole "The bells letting other providers use their dslam" really sucks. It sucks for the bells, who are losing a ton of money on infrastructure upgrades only to have their customers goto the other LEC. It sucks for the ISPs because after paying the line charge to the main telco company, they have no room to make profit of their own.
As a physics major senior, i can say it's not entirely difficult to get a job as a physics major. While all the ads and job posting aren't going to specifically ask for a physics major, usually we're able to sell ourselves into the engineering-type-but-not-hardcore-engineering jobs w/o a problem. Other quantitative analysis jobs we can also get into easily it's just matter of selling yourself.
Also many ppl i know (and me included) who aren't planning to go into achedemia all have double majors like physics/finance physics/math, etc where it gives us a huge leg up on the plain finance majors.
---
Ok with that aside, in response to the actual question:
Don't be afraid to double/triple major. It's not terribly hard to double major and finish in 4 years if you really put your mind to it. Also if you plan to double major, then hit relativstic electrodynamics (or whatever) and realize you hate that subject, then drop one of them and you're still fine. And have the satisfaction that you tried it and realized you didn't like it. Better than the "what if i had chosen..." thoughts 10 yrs from now.
Imho going into CS/CompE will limit your choices vs. going Physics or EE. It's much much easier to cross over from the engineering side than from the CS side. As a EE you won't be as good as a CS graduate, but if you're smart and very good there certainly will be many opportunities for you in computer-related fields.
I tend to agree. i also think that a lot of the readers are focusing on probably the least important problem with this (MS vs. linux vs. whatever), the socioecnomic impact.
imho this sounds to be a lot more communist than anything else. and why computers? i don't recall anyone giving out free vouchers for cars when they were first made popular. If I cannot afford a tv, the government isn't going to pay for one for me. Hell i don't even have a landline phone, using this logic the government would pay for that too.
While belgium may not be the epitome of free market capitalism that maybe the us or uk is, this idea scares me to hell because next thing you know some moron senator is going to suggest that we do this here, using the "that's what belgium's doing" idea.
Warranty != technical support. While it may not void the actual equipment warranty, if you call up the TAC with a smartNet-voided router, they have the right to not help you. (Though honestly in my experience they're still more than happy to).
Thank you. All the comments i am reading are confusing that PI-space is required to run BGP. That is not required, all you need are two semi-coooperative isps, one that's willing to punch holes in it's aggregate and the other that'll relay your advertisements.
Again, just as he said:
You can still be multi-homed with netblocks from one ISP to be received by another.
PI-space only makes it a bit easier in transition, but it doesn't make it anywhere near as impossible as the question implys.
I agree with you entirely, but the biggest obstacle to this being globally implemented is @home.
@Home has engrained in people's mind that they should get superfast supergreat superunlimited access at approximately $40/mo, when any real ISP probably at *best* could do maybe a $40/mo base +$x/GB or so. There's just no way to match @home, and obviously, from reading the other comments, a lot of people are unwilling to pay business class prices for good service.
The best thing that could happen to the industry is for a few regional players to sue @home for predatory anti-competitive pricing, and while they probably won't win, it will make the papers and get people thinking about Quality of Service and the like, because it will become obviously apparent that anybody's definition of "good service" cannot be satisfied at $45/mo.
Unfortunately until something like that happens, I can't see much opportunity for change. Though i certainly feel sorry for the ppl w/ @home access...this story about @home possibly having to turn of their service has great long-run implications for the industry; then maybe everyone else can raise their prices... and actaully make a profit one day.
If they have enough phonelines to run a school, there is enough copper in the ground to get a t1 there. If there isn't, I believe it's the telco's legal responsibility to plant (copper|fiber) to get something there. A t1 only uses two pair copper, and if you have a phoneline you have two pair copper. You can even do neat stuff like using X amt of your t1 channels as voice lines, so "i don't have any more copper" isn't a very valid argument because you can replace your pots lines with your t1 as well.
I highly suggest you call your telco reseller and/or look to get a new one if they say it can't be done without a really really really good reason on why it can't. The telcos will *always* create reasons on why things can't be done, but if you (actaully theoretically your reseller) keep on pressing them and/or talk to other people (who's heads aren't in their ass), you'll usually find out that it can.
Also, my figures are based on inter-city copper links of longer distance, they certainly apply here.
Longhaul lines aren't that expensive. I highly doubt the line will be more than $1k/mo, which honestly isn't that bad. Yes wireless can be cheaper, but with that you sacirifice a lot of reliability, and honestly with a budget of $80k, $1k/mo doesn't seem that bad considering the startup costs for a land-based lines is basically nil compared to any wireless solution. (Contrary to common belief, pringles cans are not acceptable antennae).
Also you may want to call a couple Tier1 resellers and see if you can get a decent price from them. Very often the Tier1s are able to undercut local providers in situtations like this because their longhaul lines are cheaper than through the local telco.
Yeah. i tend to agree. Mileage charges or not, either getting a ptp T1 or a framerelay line is probably the way you want to go.
Many people are suggesting exotic ideas like making 802.11 antennae out of pringles cans and such, but I can almost assure you that's going to be a bad idea. When something catastrophic happens, ie, rain, or wind, it's going to be your (now wet from rain) ass going out there trying to adjust your pringles can so that it points the right direction again. And your (this time dry) ass will also be on the line as your administrators ask why their email goes down everytime it rains or gets windy. I'm assuming that at your job you have better things to do than this.
Most LECs (or if you're lucky you can find a longhaul carrier that'll do it) have very decent turnaround times on fixing circuits when they go down (which btw is also very rare with t1s and/or framerelay).
Frankly i'm amazed your ISP isn't helping you more with this. They have a LEC reseller that can easily price this out these land-based options for you, if they won't do this for you it's time to find another ISP. Also you might want to look at buying directly from the Tier1s, often they'll be able to undercut a local/regional ISP just because of their longhaul ties and unusual POPs so they can save large amounts of money on linecharge.
You're missing the point. The start of this thread was puzzlement over why, exactly, mysql is so popular in this community when more complete and more robust alternatives exist (like postgresql).
Speed. For a database with a lot of reads and very little writes, mysql will beat postgres in speed. That's it.
I use both. I use mysql for some of my webprojects precisely for speed, but i use postgres for my non-web projects, where the robustness and particularly transactions and stored procedures, are necessary, and pure speed is not an issue. I (obviously) think that this is really the best to go, use each database where it's best fit.
I don't know if it's a good thing that mysql is adding all these new features, while it is good that they're providing competition for postgres in this non-web market, i hope that they can either keep up the speed (as that is more important for certain web-applications) while adding these new features, or by forking the codebase such that I can choose between speed and robustness.
hmm ok not necessarily a bad idea, though i don't know what advantage it would gain. BGP is a fairly static protocol, it doesn't adapt to changes very well (obviously, since it has to propagate around the world whenever something changes). I guess it could work, though personally i wouldn't try it due to all the potential bad things that could happen doing this and just keep you dampened longer.
Ciscos have the ability to traffic shape, but that does exactly that, traffic shaping. Most of the problems that i'm seeing because of codered/nimda/etc, isn't traffic saturation, but cpu overloads due to excessive arp requests. So what we're probably actaully looking for is some sort of cpu prioritization by process instead of necessarily traffic shaping based on routing/routed protocol.
Very good explanation, but there's one pseudo-misunderstanding that a lot of people didn't pick up on. Routers can normally handle a lot of traffic (well good ones can), but are still susceptible to cpu overload due to the massive ip scanning that these worms do, which overloads the arp subsystem of the router. arp is mainly to blame, not necessarily increased ip traffic.
Assuming that the router has an interface with a larger than/30 subnet, the router has to do an arp request for every ip on that subnet during a scan, and if enough of these ips just don't exist, then it has to wait for a massive amount of timeouts, then rerequests again, etc. Endlessly.
While you suggest that saturated WAN links could be the problem (and it very well could be given enough infected machines and a small enough link), the data i have indicates that most, if not all, of the problems within our organization are because of excessive the excessive arp requests. A router at one of our pops doesn't run bgp and our traffic data shows it had plenty of bandwidth, but it's cpu usage was at 100% for 3 hours during the first nimda attack. We see similar cpu increases on other CPE equipment with no dynamic routing or any significant increase in traffic.
> These same high-end routers often have traffic
> shaping/prioritization features. You'd think >that they could be configured so that the
>routing-protocol packets have a very high
>priority so that they're among the last to be
>dropped even at high load.
Not necessarily. In a lot of cases, mostly with multiple exit routers, it's more desireable for a hosed router to withdraw it's own route, presumably because you have another un-hosed router which can pick up the slack. In most cases, withdrawing a route is a lot better than advertising a route that doesn't work.
Microsoft and Cisco perhaps should be held independently responsible for their failings here, but it certainly does not follow that Qwest ought be absolved of all duty to its customers.
This problem has been known for a few months prior to the CodeRed outbreak. Cisco was fairly responsive in issuing a fix (not as fast as their normal bug fixes... but this isn't an IOS so it's somewhat understandable.)
Qwest should be somewhat held responsible because the fix had been out for a decent period of time, during which Qwest had declared the patch unsupported, leaving people who wanted to patch their routers without much of a choice but to leave it broken, and DOSable.
The configuration webserver on the Cisco 675 had serious DOS problems in the setups that various dsl providers were providing (i think it had to be in bridging mode or something), which were known even prior to the CodeRed problem. Essentially, if you did a getrequest with too much crap in it, it crashed.
It hit bugtraq a few months ago, while cisco was fairly responsive and issued a patch, Qwest at that time declared that patch unsupported.
If you wanted to get into the broadband deployment arena, now would be a really good fscking time to do so.
I shouldn't blast you for being entrepreneurial, but knowing the financials behind an ISP, i can tell you why it's not a good idea.
1) Just because there's a lot of unlit fiber, it doesn't necessarily mean it's entirely dirt cheap. The cost for longhaul lines can easily run into the thousands of dollars. And you can't get fiber to your POP in most places, so you either have to a) plant the fiber yourself, or b) pay your telco a loop fee to get it from your longhauls POP to yours. The latter will also run into the thousands of dollars, the former is obviously a lot larger, but only onetime.
2) While the power company idea is innovative, I personally don't see how/why it would be any cheaper/better than registering as a CLEC, colocating a few DSLAMs in your local CO, and become a DSL provider, ala Covad. And we all know how Covad fared.
3) Supply and demand is a well and cool thing to think about, but it's a bit more complex than that. The first problem is that it's engrained in people's mind that the cost is around $40. Problem there, your average total cost is already over $40/person, you can't make any money.
4) Bandwidth is expensive. Ok not that expensive, but it's a hell of a lot more expensive than $40 for a typical moderate user's use of a cable modem. Owning your own fiber and immediately getting bigenough to be able to peer to the Tier1's is a good way to save money long run, but you'll have to be cash positive in the meantime so you can paydown your debt until you own it yourself. @Home tried this, but didn't do very well as they've had to get bonds from Promethian, whom from what i understand, isn't much more than a legal loan shark.
5) You can get around #3 by having a higher price. Ok say you charge $60/mo. Who is going to pay $60/mo? only the people who understand Quality of Service, and are going to use it the most. You lose the ability to share the costs with the "light users" whom you make a profit in, thus your costs will go up even higher, as a ramification of your prices going up. neat huh?
There is not a lot of money in residential broadband, if there is any money at all. The money is in businesses, who gladly will be much higher prices for speeds that may only be semi-decent to a homeuser. Normal businesses typically do not use a lot of bandwidth, which allows the ISP to oversell, and combined with the higher prices, allows the ISP to make money. We sell our residential DSL for $50-60/mo, and i can say for a fact that we don't make a dime on them, thankfully our residential advertising has been belowpar, and thus we haven't lost our shirts over it.
As an ISP, your best customer is grandma who uses AOL to send email to their grandkids once every other week. Low costs, almost-decent revenue. You don't get those customers with broadband.
Cable companies are doing fine, they'll just need to dump you out onto the web from somewhere else and find you a different email service.
Very good point, but i think you're overlooking something.
What's to assume that other companies would be able to succeed where @home failed? According to a post above, (which I don't know if it's accurate, but knowing how much telco loop costs, it sounds reasonable), the ISP will get roughly $15/mo/person.
$15/mo/person for bandwidth is nothing. We get less than that for dialups, but luckily they can't use more than ~53k/s, and we have lots of them, so we can manage, though barely pull a profit.
Now you can see, $15/mo for a cable modem's worth of bandwidth, it's just not going to fly, unless the ISPs service sucks beyond belief.
Also the cable modem type of service requires a lot more manpower, as unlike a dialup modem, installation time and repair time is put a burden put on the provider rather than the customer. Also dedicated access means a lot more sysadmin related problems (codered, random hacking, etc).
I think the problem with the market is prices are simply too cheap. $45/mo is simply too cheap for broadband service, as witnessed by company after company going out of business (Covad, @Home, etc.)
If anyone could pull this off, it would be @home. AFAIK, they own their own backbone, so theoretically once they pay off their bonds, they'd be in much much better shape than anyone else. They have a good solid distribution channel (the cable lines), which Covad and the dsl CLECs have.
Being employed as an ISP, i would like nothing better than @home to die. Their pricing borders on predatory at best, stupid at worst. They have lowered the bar such that everyone assumes broadband is around $40~50ish, when realistically no company (as shown by the slew of bankraupcies/near bankraupcies) can make any money at that price.
Slashdot Mirror
Yes.
It really is the party-pooper solution, as it's so low-tech, but when we priced it out, for most buildings Cat5 wiring is cheaper.
Depending on what kind of walls you're working with, (drywall vs. brick, etc) i've gotten quotes from roughly $30-100 per drop in an apt. Add to that $40/port for a good switch, and you're looking at $140 per room. And good cat5 contractors will give you some ungodly long warranty, on the order of tens of years.
Contrast this with 802.11. You have to pay for multiple APs (500~2k each depending on what you want/need), then you either have to a) pay for the 802.11 card for each pc and have the tenants pay a deposit (which was ~150ish when i priced them out, 100ish if they had a laptop) or b) force the tenants to buy their own. From doing some informal surveys and asking around, the latter wont work.
Then you have the line-of-sight problem (the computer has to be kinda near the window for them to pick anything up), the rf interference issue, and other funky stuff rf physics stuff. Not to mention you're on most likely a 1yr warrenty, and have to deal with helping people get their wireless card working, which can be a huge pain in the ass as likely they'll be using one of those pcmcia-pci slot converter things.
Furthermore security-wise, you honestly cannot beat having a plugged vs. not-plugged-in port, thus you can assure people are not stealing your service... A good switch will tell you what mac addresses are coming from what port, so with some good accounting on the side, you can tell exactly which apt has a hub and is sharing with their neighbors, etc. It also makes catching troublemakers (and there will be some, trust me) a lot lot easier, as you can pinpoint it to the room, not just to a mac address.
I more or less planned/ran a campus apartment project like this, and we did at first also seriously consider the 802.11 alternative, but quickly threw it away as we realized that a) it was going to certainly cost more long-run in labor than cat5 would,and b) it most likely wouldnt save us money upfront either.
Yes.
Save for the monthly subscription revenue, there was nothing much to the AOL business to begin with, as the first mild downturn in the economy has convincingly shown, with advertising revenues from the service having now collapsed in a heap."
Furthermore, the "save for the monthy subscription revenue"... Dude, the monthly subscription revenue is what makes the company so attractive. If you can count on $21 * (areallylotofusers) / mo. guarenteed, you're already way better off than say, the services industry which is a lot more affected by economic downturns.
The monthly subscription revenue is what separates AOL from your typical ad-market company, it has a large revenue base that i can use to cushion the impact of advertising revenue, which looks to be very cyclical.
Once the ad-market has an upturn, AOL/TW is going to be quite fearsome. They have the ability to do a huge branding effort. Think how MSN teamed up with espn.com. Then MSN teamed with GE for MSNBC. Now multiply that by 20, given the TW media assets. They would have the ability to keep people within AOl/TW owned websites, further increasing ad-revenue. If they so chose, they would have the ability to create 'aol-subscriber' only content, much like they did in the late 90.
Just because things arent working out perfectly doesnt mean a) it was a bad idea, or b) that everyone should abandon ship. While TW was probably charmed by the large amt of stock thrown at them, and soldout for too low, the merger was certainly not as horrible an idea as the article makes it sound.
I connect to UUNet, and they connect to me. In that way we usually use the same amount of each others network, and thus we charge each other the same price for connection )
Sure, that's called bilateral peering, however to do so, it requires that there be roughly the same amt of traffic going in each direction... This is not the case w/ @home/othercableproviders, as they have much more 'download' traffic than they have 'upload', ie they need uunet more than uunet needs them. Although honestly i dont know for sure, i heavily doubt that CableServiceProviderX doesnt pay some fee w/ their peering arrangements.
Ok my first response: the fuck?
Why does this perceived "problem" have anything to do with management, or the problems w/ certification? This "problem" the poster perceives is entirely not a problem of technical incompetence, but rather the simple fact that to this isp, supporting these tlds is simply not worth the time/trouble/money.
Webcaching is highly cost effective. It allows providers to charge this $35-50/mo charges w/o going under, or without losing too much money.
As someone with 7 years programming experience, 3 years networking experience, and did months of research on webcaching before implementing it, i can tell you for a fact that I would not think of this as being a problem during implementation. Hell in fact i still don't think it's a problem, quite frankly taking down a router for 30 minutes just to fix one person's obscure problem just isnt worth it. You're paying what? $20/mo? Great if you stay for 6 months then you'll pay for my 1hour of overtime. and that's not counting how many other people we piss off for having this 30min slowness/downtime.
So in response to your post. This was not a bad decision by any means. Done correctly it has the ability to speed up response times (and my data shows that it does) and saving money by using much less bandwidth (btwn 10-40% even while full honoring of if-modified-since headers). It improves service to 99.999% of their customers and pisses off one, because he can't use some esotericthingthattheISPdoesntwanttosupport anyways. Please.
Your points over worthless certification and stuff like that certainly could be valid points if you had applied them a relevant arguement, however this is entirely not, as you have no position/ability to judge that this was indeed a bad decision, or that it had anything to do with (unknowledgable) management decisions.
Cisco wccp/gre tunneling had large amounts of bugs until fairly recently (1-2 years ago), and still require specific ios versions. This bug would cause the gre tunnel (which connected the router to the cache) randomly collapse. That is probably the bug you are seeing, and is entirely unrelated to that the poster's problem is.
If the web proxy can't identify the hostname present in the URL, it should simply pass it through, allowing the client (who already knows the IP), and the Web Server (who also, clearly, already knows it's own IP) to communicate.
.0001% their client base that this concerns is rediculous.
Not a horrible idea, i'm fairly doubtful it will work though.
The problem is the caching architectures i've seen have the exit router communicating to the cache through a gre tunnel, so the cache has no way to actually let it "pass through" as the cache isnt the actual box doing the routing. This is vastly superior to "put proxy between person and internet and let it route too" because of loadbalancing and failover reasons, but doesnt allow what you're asking for.
I dont really consider this 'broken proxy' nor do i consider it a misconfiguration, the simple fact is that his ISP (as well as mine) denies the existance of the opentld system. If he really has a problem with it, he should find another provider. Expecting the ISP to bend over backwards to satisfy their
Great. If 77% are buying one more cd, while the 23% are buying 8 less cds, then you have a net loss. This figure cited proves nothing.
Of course not. Music demand is highly inelastic, they can maximize revenue by raising prices, not lowering. Crack dealers (also in an inelastic industry) do not lower prices to increase revenue.
Or it could be that Mariah has enough personal problems that it would be cheaper for them to pay the 28million than to keep her under her much more expensive contract. Unfortunately Mariah has fallen down from her huge-popularties of the late 80s early 90s, and her sony contract was priced assuming her massive popularity. Sony was merely cutting their losses on her, and is in no way indicitive of the rest of the music released.
Also, every "napster helps sales" argument i have seen is purely anecdotal, or if not, it lacks the numbers needed to prove anything. Furthermore all these surveys have a high bias (people feel more compelled to say they buy more music, than buying less). If you were from say Gallup and flagged me down on the street asking me if i bought more/same/less of music after using napster, of course i would say more, as i wouldnt want (napster|kazaa|morpheus) to go away.
- With no incentive to expand, much of the US will never get broadband.
Note that in addition to everything that everyone hates about the bill, the bill also forces the ILECS to provide coverage in 100% of the central offices with in 5 years. Telco is not making bazillions of dollars like everyone thinks they do, and they need some type of incentive to expand, this will give them the incentive/motivation to do so.So while everyone is saying "monopoly = bad", and "why doesnt the entire nation have broadband yet?" you really get one or the other. If you want nationwide broadband, you have to make it profitable for the ILECs. Girnomous federal subsidies would be the other solution, but would be enormously harder to get through congress.
Honestly, i dont know if i agree with this bill either, but i certainly see what they're trying to solve, and how they're trying to solve it. Please, no more of this "all of congress is evil evil evil corporation bloodsuckers" and "this bill and only benefits the bells". While the email tries to spin it saying "One would think that it would be hard to find people who support this bill", reading the entire bill it makes perfect sense why someone would. Nation-wide broadband.
Ummm... no.
ISPs have huge fixed costs. city-wide isps are in the tens of thousands of dollars, regional isps in the millions. High fixed cost, while the cost of adding additional subscribers is quite small, you have textbook economies of scale.
Then you have the bulk discounts on bandwidth, which are *huge*. Buy your first T1, it costs 900-1kish. If you are near a NAP and need an oc-3, it only costs. hell if you're big enough you dont even need to pay for commodity bandwidth, you just do bilateral peering w/ the uunets of the world.
Also when you said "thrived", i dont know what you mean by that. If you mean "barely turning a profit thanks to razor thin margins and huge telco tariffs", then you're probably correct. Furthermore, actaully had to maintain modem POPs, pay for PRIs, call forwarding, longhaul, etc, while now for cable|dsl, all you have to do is throw in a dslam, a big router, and buy one piece of fiber to your nap.
Honestly, this whole "The bells letting other providers use their dslam" really sucks. It sucks for the bells, who are losing a ton of money on infrastructure upgrades only to have their customers goto the other LEC. It sucks for the ISPs because after paying the line charge to the main telco company, they have no room to make profit of their own.
As a physics major senior, i can say it's not entirely difficult to get a job as a physics major. While all the ads and job posting aren't going to specifically ask for a physics major, usually we're able to sell ourselves into the engineering-type-but-not-hardcore-engineering jobs w/o a problem. Other quantitative analysis jobs we can also get into easily it's just matter of selling yourself.
Also many ppl i know (and me included) who aren't planning to go into achedemia all have double majors like physics/finance physics/math, etc where it gives us a huge leg up on the plain finance majors.
---
Ok with that aside, in response to the actual question:
Don't be afraid to double/triple major. It's not terribly hard to double major and finish in 4 years if you really put your mind to it. Also if you plan to double major, then hit relativstic electrodynamics (or whatever) and realize you hate that subject, then drop one of them and you're still fine. And have the satisfaction that you tried it and realized you didn't like it. Better than the "what if i had chosen..." thoughts 10 yrs from now.
Imho going into CS/CompE will limit your choices vs. going Physics or EE. It's much much easier to cross over from the engineering side than from the CS side. As a EE you won't be as good as a CS graduate, but if you're smart and very good there certainly will be many opportunities for you in computer-related fields.
I tend to agree. i also think that a lot of the readers are focusing on probably the least important problem with this (MS vs. linux vs. whatever), the socioecnomic impact.
imho this sounds to be a lot more communist than anything else. and why computers? i don't recall anyone giving out free vouchers for cars when they were first made popular. If I cannot afford a tv, the government isn't going to pay for one for me. Hell i don't even have a landline phone, using this logic the government would pay for that too.
While belgium may not be the epitome of free market capitalism that maybe the us or uk is, this idea scares me to hell because next thing you know some moron senator is going to suggest that we do this here, using the "that's what belgium's doing" idea.
Warranty != technical support. While it may not void the actual equipment warranty, if you call up the TAC with a smartNet-voided router, they have the right to not help you. (Though honestly in my experience they're still more than happy to).
Thank you. All the comments i am reading are confusing that PI-space is required to run BGP. That is not required, all you need are two semi-coooperative isps, one that's willing to punch holes in it's aggregate and the other that'll relay your advertisements.
Again, just as he said:
You can still be multi-homed with netblocks from one ISP to be received by another.
PI-space only makes it a bit easier in transition, but it doesn't make it anywhere near as impossible as the question implys.
I agree with you entirely, but the biggest obstacle to this being globally implemented is @home.
@Home has engrained in people's mind that they should get superfast supergreat superunlimited access at approximately $40/mo, when any real ISP probably at *best* could do maybe a $40/mo base +$x/GB or so. There's just no way to match @home, and obviously, from reading the other comments, a lot of people are unwilling to pay business class prices for good service.
The best thing that could happen to the industry is for a few regional players to sue @home for predatory anti-competitive pricing, and while they probably won't win, it will make the papers and get people thinking about Quality of Service and the like, because it will become obviously apparent that anybody's definition of "good service" cannot be satisfied at $45/mo.
Unfortunately until something like that happens, I can't see much opportunity for change. Though i certainly feel sorry for the ppl w/ @home access...this story about @home possibly having to turn of their service has great long-run implications for the industry; then maybe everyone else can raise their prices... and actaully make a profit one day.
If they have enough phonelines to run a school, there is enough copper in the ground to get a t1 there. If there isn't, I believe it's the telco's legal responsibility to plant (copper|fiber) to get something there. A t1 only uses two pair copper, and if you have a phoneline you have two pair copper. You can even do neat stuff like using X amt of your t1 channels as voice lines, so "i don't have any more copper" isn't a very valid argument because you can replace your pots lines with your t1 as well.
I highly suggest you call your telco reseller and/or look to get a new one if they say it can't be done without a really really really good reason on why it can't. The telcos will *always* create reasons on why things can't be done, but if you (actaully theoretically your reseller) keep on pressing them and/or talk to other people (who's heads aren't in their ass), you'll usually find out that it can.
Also, my figures are based on inter-city copper links of longer distance, they certainly apply here.
Longhaul lines aren't that expensive. I highly doubt the line will be more than $1k/mo, which honestly isn't that bad. Yes wireless can be cheaper, but with that you sacirifice a lot of reliability, and honestly with a budget of $80k, $1k/mo doesn't seem that bad considering the startup costs for a land-based lines is basically nil compared to any wireless solution. (Contrary to common belief, pringles cans are not acceptable antennae).
Also you may want to call a couple Tier1 resellers and see if you can get a decent price from them. Very often the Tier1s are able to undercut local providers in situtations like this because their longhaul lines are cheaper than through the local telco.
Yeah. i tend to agree. Mileage charges or not, either getting a ptp T1 or a framerelay line is probably the way you want to go.
Many people are suggesting exotic ideas like making 802.11 antennae out of pringles cans and such, but I can almost assure you that's going to be a bad idea. When something catastrophic happens, ie, rain, or wind, it's going to be your (now wet from rain) ass going out there trying to adjust your pringles can so that it points the right direction again. And your (this time dry) ass will also be on the line as your administrators ask why their email goes down everytime it rains or gets windy. I'm assuming that at your job you have better things to do than this.
Most LECs (or if you're lucky you can find a longhaul carrier that'll do it) have very decent turnaround times on fixing circuits when they go down (which btw is also very rare with t1s and/or framerelay).
Frankly i'm amazed your ISP isn't helping you more with this. They have a LEC reseller that can easily price this out these land-based options for you, if they won't do this for you it's time to find another ISP. Also you might want to look at buying directly from the Tier1s, often they'll be able to undercut a local/regional ISP just because of their longhaul ties and unusual POPs so they can save large amounts of money on linecharge.
You're missing the point. The start of this thread was puzzlement over why, exactly, mysql is so popular in this community when more complete and more robust alternatives exist (like postgresql).
Speed. For a database with a lot of reads and very little writes, mysql will beat postgres in speed. That's it.
I use both. I use mysql for some of my webprojects precisely for speed, but i use postgres for my non-web projects, where the robustness and particularly transactions and stored procedures, are necessary, and pure speed is not an issue. I (obviously) think that this is really the best to go, use each database where it's best fit.
I don't know if it's a good thing that mysql is adding all these new features, while it is good that they're providing competition for postgres in this non-web market, i hope that they can either keep up the speed (as that is more important for certain web-applications) while adding these new features, or by forking the codebase such that I can choose between speed and robustness.
hmm ok not necessarily a bad idea, though i don't know what advantage it would gain. BGP is a fairly static protocol, it doesn't adapt to changes very well (obviously, since it has to propagate around the world whenever something changes). I guess it could work, though personally i wouldn't try it due to all the potential bad things that could happen doing this and just keep you dampened longer.
Ciscos have the ability to traffic shape, but that does exactly that, traffic shaping. Most of the problems that i'm seeing because of codered/nimda/etc, isn't traffic saturation, but cpu overloads due to excessive arp requests. So what we're probably actaully looking for is some sort of cpu prioritization by process instead of necessarily traffic shaping based on routing/routed protocol.
Very good explanation, but there's one pseudo-misunderstanding that a lot of people didn't pick up on. Routers can normally handle a lot of traffic (well good ones can), but are still susceptible to cpu overload due to the massive ip scanning that these worms do, which overloads the arp subsystem of the router. arp is mainly to blame, not necessarily increased ip traffic.
/30 subnet, the router has to do an arp request for every ip on that subnet during a scan, and if enough of these ips just don't exist, then it has to wait for a massive amount of timeouts, then rerequests again, etc. Endlessly.
;)
Assuming that the router has an interface with a larger than
While you suggest that saturated WAN links could be the problem (and it very well could be given enough infected machines and a small enough link), the data i have indicates that most, if not all, of the problems within our organization are because of excessive the excessive arp requests. A router at one of our pops doesn't run bgp and our traffic data shows it had plenty of bandwidth, but it's cpu usage was at 100% for 3 hours during the first nimda attack. We see similar cpu increases on other CPE equipment with no dynamic routing or any significant increase in traffic.
(ccie in progress
> These same high-end routers often have traffic
> shaping/prioritization features. You'd think >that they could be configured so that the
>routing-protocol packets have a very high
>priority so that they're among the last to be
>dropped even at high load.
Not necessarily. In a lot of cases, mostly with multiple exit routers, it's more desireable for a hosed router to withdraw it's own route, presumably because you have another un-hosed router which can pick up the slack. In most cases, withdrawing a route is a lot better than advertising a route that doesn't work.
Microsoft and Cisco perhaps should be held independently responsible for their failings here, but it certainly does not follow that Qwest ought be absolved of all duty to its customers.
This problem has been known for a few months prior to the CodeRed outbreak. Cisco was fairly responsive in issuing a fix (not as fast as their normal bug fixes... but this isn't an IOS so it's somewhat understandable.)
Qwest should be somewhat held responsible because the fix had been out for a decent period of time, during which Qwest had declared the patch unsupported, leaving people who wanted to patch their routers without much of a choice but to leave it broken, and DOSable.
The configuration webserver on the Cisco 675 had serious DOS problems in the setups that various dsl providers were providing (i think it had to be in bridging mode or something), which were known even prior to the CodeRed problem. Essentially, if you did a getrequest with too much crap in it, it crashed.
It hit bugtraq a few months ago, while cisco was fairly responsive and issued a patch, Qwest at that time declared that patch unsupported.
If you wanted to get into the broadband deployment arena, now would be a really good fscking time to do so.
I shouldn't blast you for being entrepreneurial, but knowing the financials behind an ISP, i can tell you why it's not a good idea.
1) Just because there's a lot of unlit fiber, it doesn't necessarily mean it's entirely dirt cheap. The cost for longhaul lines can easily run into the thousands of dollars. And you can't get fiber to your POP in most places, so you either have to a) plant the fiber yourself, or b) pay your telco a loop fee to get it from your longhauls POP to yours. The latter will also run into the thousands of dollars, the former is obviously a lot larger, but only onetime.
2) While the power company idea is innovative, I personally don't see how/why it would be any cheaper/better than registering as a CLEC, colocating a few DSLAMs in your local CO, and become a DSL provider, ala Covad. And we all know how Covad fared.
3) Supply and demand is a well and cool thing to think about, but it's a bit more complex than that. The first problem is that it's engrained in people's mind that the cost is around $40. Problem there, your average total cost is already over $40/person, you can't make any money.
4) Bandwidth is expensive. Ok not that expensive, but it's a hell of a lot more expensive than $40 for a typical moderate user's use of a cable modem. Owning your own fiber and immediately getting bigenough to be able to peer to the Tier1's is a good way to save money long run, but you'll have to be cash positive in the meantime so you can paydown your debt until you own it yourself. @Home tried this, but didn't do very well as they've had to get bonds from Promethian, whom from what i understand, isn't much more than a legal loan shark.
5) You can get around #3 by having a higher price. Ok say you charge $60/mo. Who is going to pay $60/mo? only the people who understand Quality of Service, and are going to use it the most. You lose the ability to share the costs with the "light users" whom you make a profit in, thus your costs will go up even higher, as a ramification of your prices going up. neat huh?
There is not a lot of money in residential broadband, if there is any money at all. The money is in businesses, who gladly will be much higher prices for speeds that may only be semi-decent to a homeuser. Normal businesses typically do not use a lot of bandwidth, which allows the ISP to oversell, and combined with the higher prices, allows the ISP to make money. We sell our residential DSL for $50-60/mo, and i can say for a fact that we don't make a dime on them, thankfully our residential advertising has been belowpar, and thus we haven't lost our shirts over it.
As an ISP, your best customer is grandma who uses AOL to send email to their grandkids once every other week. Low costs, almost-decent revenue. You don't get those customers with broadband.
Very good point, but i think you're overlooking something.
What's to assume that other companies would be able to succeed where @home failed? According to a post above, (which I don't know if it's accurate, but knowing how much telco loop costs, it sounds reasonable), the ISP will get roughly $15/mo/person.
$15/mo/person for bandwidth is nothing. We get less than that for dialups, but luckily they can't use more than ~53k/s, and we have lots of them, so we can manage, though barely pull a profit.
Now you can see, $15/mo for a cable modem's worth of bandwidth, it's just not going to fly, unless the ISPs service sucks beyond belief.
Also the cable modem type of service requires a lot more manpower, as unlike a dialup modem, installation time and repair time is put a burden put on the provider rather than the customer. Also dedicated access means a lot more sysadmin related problems (codered, random hacking, etc).
I think the problem with the market is prices are simply too cheap. $45/mo is simply too cheap for broadband service, as witnessed by company after company going out of business (Covad, @Home, etc.)
If anyone could pull this off, it would be @home. AFAIK, they own their own backbone, so theoretically once they pay off their bonds, they'd be in much much better shape than anyone else. They have a good solid distribution channel (the cable lines), which Covad and the dsl CLECs have.
Being employed as an ISP, i would like nothing better than @home to die. Their pricing borders on predatory at best, stupid at worst. They have lowered the bar such that everyone assumes broadband is around $40~50ish, when realistically no company (as shown by the slew of bankraupcies/near bankraupcies) can make any money at that price.