Comment Re: How can the situation be improved? (Score 1) 513
(Fiber is dedicated to the building, active ethernet, not GPON, typically delivered on a gigabit port, rate-limited at the CPE according to the requirements of the complex).
(Fiber is dedicated to the building, active ethernet, not GPON, typically delivered on a gigabit port, rate-limited at the CPE according to the requirements of the complex).
Should mention that we overprovision connections by 20%, just because. Connections are typically FTTB + complex-wide wifi.
We charge $10 for 4mbit/s and $20/mo for 10mbit/s symmetrical per subscriber in Southern IL... but that's mostly because the complex itself buys the connection which means we're providing a lot of subscribers in one go, so in fairness we don't have a last mile to worry about.
There are some entities (peering exchanges) which require this already: if your pipe has a CBR of over 80 or 85% they require you to upgrade. In some countries (like India), it's even mandated in one of the various regulations (along with many other things which providers in that country will quite happily flout or ignore).
I understand where you're coming from, but the devices are not black boxes. As I mentioned, the devices operate using a collection of open-source software (Debian MIPS & Vyatta) and presumably can be hacked/upgraded/fiddled with to your hearts content if you so desire.
As I also mentioned, IF ANYTHING probably it's only the GUI that is closed, but having not bothered to check on my own devices I don't know whether it actually is or is not. Considering however that the functionality you're referring to is in the core system (which appears to be OSS), what you're describing seems largely to be a non-issue - you can SSH in to the device and see a good old mostly-standard Linux CLI.
I can and have successfully installed other debian packages on it just by running apt-get, so, as long as the debian repos are up-to-date, I have no reason to suspect that the software on my own ER is not also up-to-date.
I always thought of it like a BBM replacement (which is why I don't get the reason for BBM on Android/iOS) but also
Of course, Skype, Hangouts, and lots of other apps satisfy all those requirements.
As far as taking the account with you when you change phone numbers...
Gosh, yeah, like when you move to a different country or switch to another carrier and can't take your number with you.
As someone who moves countries on a fairly regular basis (7-ish times in 10-ish years), I understand that argument better than most, however, for the majority of people that would be a non issue, hence my expression of desiring the option for a username/password to link the account(s) together and/or keep the history etc.
As for simply moving carrier, number portability is available in many countries - not knowing which country you're from, however, I can't say for certain if it's an option for you.
As for Skype/Hangouts/etc that's all good and well for those with smartphones, but in a country like India for example, most people don't have smartphones yet whatsapp can still run on some fairly basic hardware and fairly reliably on 2G whereas those options require 3G**
(**by "requires", I mean, in order for it to work
I always thought of it like a BBM replacement (which is why I don't get the reason for BBM on Android/iOS) but also
1. Lots of countries don't have unlimited texting
2. Free international texting
3. Texts longer than 160 characters
4. Group texting (we use it to send status updates to subscribers when we have network outages)
5. Easy export of message history
6. Confirmed receipt & reading of messages
As far as taking the account with you when you change phone numbers... I'd call that a feature, not a bug. Presumably if I was changing my phone number, I'd be doing it for a reason, although it would be nice to have the option to register properly and have it tied to a username/password if I wanted to.
They're updating EdgeMax every few months - 1.4.0 came out just a few weeks ago. As far as code is concerned, it seems to be sitting on top of vyatta with a mini install of Debian (MIPS) so it should be fairly straightforward for anyone who has ever used Linux CLI.
As far as any other proprietary stuff is concerned it would seem that, if at all, anything "proprietary" would probably be mostly UI stuff, but even that can be replaced if you really want (there seem to be a couple of projects floating around). You can install anything in the Debian repos (I usually start with nano and for a while I used darkstat but I now run cacti on a separate server and retrieve everything by snmp).
While most of the system is open-source/using FOSS components (that is to say, the entire underlying system is Open Source), is it really *necessary* to have everything 100% FOSS or would you rather have something that is 90-95% FOSS and that which isn't "just works"?
I suspect it's a lot less than that, but good luck getting bandwidth at $0.50/mbit/month (Yes, I'm aware you can get shitty non-premium bandwidth in some major centres for that price but as a carrier 1. you wouldn't want that and 2. most of the US isn't in a major centre and 3. even if you're buying the cheap bandwidth in a major centre, you're also going to have to take in to account all the fiber to get from there to where your customers are (and that may depend on all sorts of factors: buried or overhead, owned or leased,
I've been quoted as much as $55/mbit/month/delivered from Verizon (this would work out to a cost-equivalent of $0.183/GB, or if we were to use simple rule-of-thumb accounting, about $0.57/GB retail price including tax), and an average of $43-49/mbit/month/delivered from AT&T (it's pretty similar pricing in the 4 states I've got pricing from them in).
My alternative would have been to buy in a major centre at $something per mbit (say $1 for the sake of argument) but then I would have had the cost of leasing fiber from Chicago or whereever to the nearest drop plus a buildout cost of between $50k and $100k to reach my premises (and it is only really those upfront costs that prevented us from going ahead with that) - and that's before I even got to the delivery costs.
For that, if I'm deploying my fiber overhead, my direct costs are about $9/pole/year plus the fiber, distribution equipment & CPE plus some nominal costs to the city and state (nothing to the FCC - yet - since we're not doing TV or phone), so using some pretty crude numbers and assuming I could get the customer density I want, with the most expensive prices I've been quoted, I could still **probably** break even on about $20/month for a usage based plan including 30GB. Then, taking in to account that many of the users wouldn't use that much, I could probably offer 100GB for $50 or less.
But, since I have successfully managed to get my bandwidth costs to a pretty reasonable level, I could probably even look at 500GB - maybe even 1TB "limits" by the time we hit the $99 mark.
ISP Name please?
I must be lucky: testing from home at http://netneutralitytest.com/ my speeds to Linode Atlanta are between 35 & 45mbit/s whereas my speeds to AWS East are between 50 & 60mbit/s. Fairly consistantly (I alternated between each site 3 times).
Of course, I'm not with or carried by Verizon.
I dual-boot Windows & Mint and was using Mint for about 18 months exclusively until I fell back to Windows due to shitty support for Broadcom WiFi in Linux (Internet connectivity circumstances changed), rather than the software (most of what I use is cross-platform anyway: web browsers, Thunderbird, LibreOffice, VLC, Skype and a few other things).
Fortunately, I'm about to upgrade my laptop, and one of the prerequisites is that the WiFi is based on another chipset (such as Intel), so I can go back to an OS that doesn't piss me off as much.
Most of the computers in my business - both desktops and servers - are running Linux in some form or another. Desktops mostly Mint 13 LTS and servers may be either CentOS or Debian-based depending on their function. We did have a BSD-based phone system but I think that's being shuttled on to a Debian-based machine now.
The only Windows machines are for some specific tasks - marketing/design likes to have the Adobe suite available but that's about it. It's due to a strange form of irony that we *don't* have any Apple machines in there for aesthetic reasons (a mostly non-user-serviceable white or brushed aluminium machine in a city like Mumbai? Hah, gross! At least with PC-based machines we can just swap stuff out.)
Yes. Successfully on more than one occasion.
Even those aren't
The point was that the electronics should cost nowhere near $150k per cabinet based on the pricing I've seen from major vendors. They could (should), in fact, double the number of cabinets to reduce the overall copper lengths (thus improving quality of service).
I'm not exactly certain on what else needs to go in a *DSL cabinet other than the ONT and batteries, but with equipment as cheap as it is, the costs for replacing equipment in 16 cabinets (and adding another 16) likely would not cause a loss at the end of the financial year.
I'll use Alcatel as an example because that was the first spreadsheet I opened:
$1k for an Alcatel O-00240v-q ONT (each)
$4k averaged across 32 cabinets for batteries, construction and such is probably not out of the question
=total $5k * 32 = $160k
~$30k for an Alcatel 7302 16 slot shelf with 4x 8 port line cards, 32 SFPs and all the rest
$minimal to change POTS to VOIP
~$100 each for battery backups at each customer location (total $105k)
Some other misc fees the power company will charge to come out and survey the poles and allow things to be changed around (assuming they use the power company poles)
RoI break-even is still only about 2 years with a total cost of maybe $300k all up. Considering they already have the fiber in the ground and copper presumably in the air, most other changes come under labour which one would assume they have a salaried staff member for and thus are already paying.
Then, find a cheaper middle-mile supplier - we all know that AT&T is expensive compared to pretty much everyone except Verizon, but if cost really is an issue, don't use them. Either way, contending bandwidth at 1:30 on a $5500/mo line (assuming 100mbit/s on the highest cost basis I've been quoted in the last 2-3 months for service in rural areas) still works out at a cost per GB equivalent of less than $0.20, and allows 10GB per mbit served (or, an average of 30GB per customer - and considering that they're saying their average customer uses 15GB, they should be golden). And even if that cost is double because they have "plenty" of spare bandwidth for their IPTV services and everything else, their costs aren't anywhere near high enough to justify $5/GB - that's just daylight robbery.
Perhaps, but an ISP is unlikely to buy a 600mbit/s pipe unless it is needed. In practice a small provider buys say 100 mbit/s from y company for $z and 100 mbit/s from b company for $c, configures his routers accordingly (fail-over, least-cost, whatever) and upgrades to the next tier only when he's hitting an average of say 80mbit/s - he doesn't buy 600 mbit/s on day 1 -- unless he has at least say 500 of those mbit/s to fill, of course.
The agreement with the upstream probably supports this with wording along the lines of "we buy x mbit/s for $y per month/year (or possibly but not likely, we buy an IRU for $y per decade) and when our usage hits 75-80% continual flow, you get ready to provision us more and bill accordingly".
Given the timeframes given to me by the likes of AT&T when I've gone to them for quotes (don't usually end up buying from them because they're not very good value over their competition IMO) typically this means there will be a 2-4 week period where the network will seem a bit congested while the upstream provider provisions the necessary bandwidth from point A to point Z in it's network to support the new traffic.**
**If you are able to go with a company more efficient than AT&T, you can get more bandwidth provisioned same-day (depending on exactly how much more you want/need, of course).
And of course, the more you buy, the more you save: at the 100mbit level the ISP pays say $25/mbit/mo but at the gigabit level the ISP pays say $10/mbit/mo - the increase in costs **should** only relate to hardware (new SFPs or even new routers), but on a cost-per-customer basis and with equipment basically only decreasing in price, actual expenditure usually works out to be minimal, so to be honest, how they are spending 900% over what they used to is somewhat baffling: in terms of dollars, perhaps it's true; in terms of dollars per customer, I'd be skeptical -- unless 850% of that comes under "executive renumeration"
After Goliath's defeat, giants ceased to command respect. - Freeman Dyson