Catch up on stories from the past week (and beyond) at the Slashdot story archive


Forgot your password?

Comment Re:wft ever dude! (Score 1) 168 168

I have a 1Gb link that is rate limited to 100Mb. When I download files, wireshark shows a 1Gb burst of 30 1500 byte packets back-to-back, then my ISP's rate limiting starts to clamp down and the traffic shaping starts to space out the TCP flow and dials down to my 100Mb provisioned rate. This all happens in the first second, not 15-20 seconds. Although I have an 10ms RTT to every major datacenter in Chicago via Level 3 Comm. Low RTTs allow TCP to quickly ramp up. My ISP does have one CDN on their network, akamai. 1.5ms ping.

10ms to Chicago, 30ms to New York City and Atalanta and Washington(AWS), 40ms to Texas and Florida, and 60ms to Cali. Short RTTs help a lot with TCP.

Bandwidth isn't everything, ping, jitter, and loss are also important. Jitter typically indicates congestion and so does loss. I can reach every major datacenter in the world with under a 250ms RTT. That includes Moscow, India, China, Japan, South Korea, New Zealand, and Australia. Also, all under 1ms of jitter.

Comment Re:But 32 bits is enough for anybody (Score 1) 168 168

ARIN already laid out several phases. A few months ago they started to limit how many IPs they handed out, a month ago they started to reject some requests. We're reaching the end game, which includes reclaiming IPs. You will need to prove every year that you still need your blocks more than others, and every year ARIN will get more strict and refuse renewal for some customers so other customers that are more deserving get some. It will start to get painful.

Comment Re:No, it won't be a problem. (Score 2) 168 168

Some CDNs are seeing 18%-40% of web requests from AT&T, Verizon, or Comcast are over IPv6. IPv6 is still growing at an exponential rate for almost a decade now, about 100% per year. At the current about 10% of all USA, given 100% growth that hasn't shown any signs of stopping, we'll be at 40% in two years and 80% in 3 years.

Comment Re:Whats left unsaid... (Score 1) 120 120

Remember this thread started with a conversation about the cost of going to 1gbs. Customers that are having to move from say 30-200mbs to 1gbs+ are going to want more than 1-2mbs dedicated. They are going to want much more than 300gb / month total. Your numbers are way too low for connections that fast. For example residential traffic in the USA is becoming much more steady because of things like Netflix. Netflix needs about 7mbs and that's not including other traffic in the household. And that's all running comfortably on today's 50mbs type connections not for near future when Netflix is using 8x as much and people are pulling down multiple streams.

You're quite right, but even then the increase in usage per subscriber doesn't necessarily correlate with the increase in speed, that is to say, a 50mbps customer going to 100mbps doesn't necessarily double their usage - it may go up by 20-50% or it may even stay near the same as it was before. A much larger jump, say, going from 2 or 4mbps to 100mbps might have that effect though.

We know that home providers right now are underprovisioned. 7pm to 11pm they are only able to deliver 4mbs per home on average, while at times like 4:00 am they can do 18mbs / household and customers express satisfaction. And again that's with many customers buying as little as 10mbs. We know that 2mbs dedicated is less than 1/2 what customers are getting now in the worst markets.

My .2 multiplier I think is fair for good quality peak usage (what the original poster wants). But even if I am to high it is nothing like a .001-.002 as a multiplier. 1000x to one overprovision you just aren't delivering anything like 1gbs of bandwidth.

That's why my numbers are padded to account for 10mbps dedicated which is what I'm working from - that works out to a contention ratio of 100:1 for gigabit connections (a stupid metric, IMO) and comfortably allows 1.5TB per month per subscriber if you're assuming an average of 50% utilization on your trunks with some room for burst/peak.

Individual providers would of course tweak numbers according to their customers habits and availability of capacity to other networks, route costs and so forth, but again, what we're talking about here is large ISPs have cores in several separate areas of the country, so what we're talking about is 130tb/s of core routing and interconnecting bandwidth spread across *all* the major ISPs (let's call it 20tb/s and some change each), divided by region (say 5tb/s per region) which is very doable - if I'm in California, I don't expect my ISP to be trying to deliver say Netflix from their Atlanta core unless said ISP is terrible at routing, I'd expect it to come from San Jose or whereever.

Then you're looking at distributing bandwidth from those large areas to any given point over trunks which get progressively smaller as you get to smaller towns - n*100gb turns to 100gb which becomes n*10gb which becomes a single 10gb link for a small town because obviously a town with 1,000 households doesn't need 100gb or indeed 13tb/s to it - and that 10gb link would contended by say 1,000 households (100:1) which roughly works out.

As for my house. I have about 8 copper lines running to my house. I have a switch box which connect the outlets to up to 3 of those incoming lines. I also have an older system using lead wires inside glass. 2 of those lines were setup for DSL. 0 of them are supported by the phone company today. All of that is abandoned infrastructure.

Alright, but how long were those lines in service?

I would suppose that *IF* the ROI on that wiring was 5 years, it's because back in the day, Ma Bell charged a fortune for everything because they could - telephones were somewhat of a luxury, long distance was expensive as hell and so on and so they basically had huge margins.

However, margins are a lot thinner these days but that doesn't change the fact that they're installing infrastructure assuming it's going to be there for 2 to 5 decades, not 5 years. As it was with copper in the 70s and 80s so it is with fiber today and it is still such that in order to upgrade a network all you really need to do is change out the equipment on the other end of the line which (on a per subscriber basis) really isn't all that expensive.

As for ROI... Remember where I started. 5 years to pay off the original investment (no profit) then Y years of profit. The goal of infrastructure is not to break even,. not to have an ROI of 0. Your $120 box is a perfect example. You are renting it at $5/mo which means a 24 mo breakeven point not 25 years. A 25 year payoff would be you renting the box for $.40 / mo, a 50 year payoff $.20/mo.

Yes, and my numbers reflect that. The amounts I'm talking about are for the modem and port at the DSLAM *only*, whereas you're kind of bundling the wiring and modem in to a single ROI timeframe which doesn't really work because it doesn't account for the advances in technology which inevitably occur without having to change out any of the wires.

So yes, you have a $120 modem which has an expected lifespan of 5 years = average cost per month of $2. At $5 rental per month it's getting paid off in 24 months and the rest is profit.

I guess what I'm trying to say here is that yes, a 12 year ROI on modems/equipment would be unacceptable, but on wiring it's probably about right (depending on your market and all that) and these things have to be counted separately, which is what my numbers were trying to reflect.

BUT what I'm also trying to explain is that when you have $2,000 worth of wiring which has an expected lifespan of 25 years, the average cost per month of just under $7. Assuming this worst case scenario, the amount of the customer bill attributed to that is probably at least $15, so that would make the break-even time would be 12 years (not counting finance and interest costs but that gets too complicated to go in to here) and anything past that point is profit. However, as we know if we're wiring high-density it's much less because a single feed is reaching multiple households and that $2,000 becomes $500 or $300 or whatever the case may be, and when that happens as it would in the case of the project in question, we're looking at maybe a couple of dollars per month as the cost for the infrastructure, not the $33/month you mentioned.

Comment Re:Memristor? (Score 1) 175 175

Memristor is an a pseudo-object with a set of meta properties. A lot of things could be technically be a memristor. Human skin can technically be one, but making use of it would be rather difficult. XPoint may be similar to a memristor in a layman's sense, but Intel may realize that XPoint only has the property of storage, which is not the only meta property required. It is also not phase-change because nothing physical changes. Not much information to go on.

The only thing for sure is it is a form of ReRAM.

Comment Re:Endurance figures (Score 1) 175 175

Intel said they're unifying memory and storage with this. No more distinction between DRAM and HDs, they're one and the same. Of course other stuff needs to happen first, but that obviously indicates that they expect it to replace DRAM. Actually, the explicitly stated they will make DRAM with it.

FORTRAN is for pipe stress freaks and crystallography weenies.