The FCC is supposed to enforce the Telecom Act, and it is supposed to follow the Administrative Procedures Act in doing so. This means that major decisions have to be done in the open, with a public Comment period. Sure, they can try to ignore the Comments, but the Record they create helps on appeal. The Courts in general do not like this FCC, which tends to be viewed as a bunch of cowboys.
There are some urgent Dockets open now at the Wireline [prevention-of] Competition Bureau (WCB). Two, from BellSouth and Verizon, call for their DSL service (and ATM and Frame Relay, in VZ's case, and possibly T1 and T3 leased line in at least BellSouth's case) to be reclassified as "private", rather than "common carriage". This means that they will be under no obligaton whatsoever to allow competing ISPs, like Speakeasy and Earthlink, to use them! (Their excuse: Cable doesn't haveta so they doan wanna.) Since cable (never a common carrier by law) doesn't provide an alternative, most ISPs will be shut off from their customers and thus put out of business. (Dial-up might remain open, but they're working on it, believe me. I'm in the middle of it.) If they do deign to allow an independent ISP onto their wire, it won't be subject to protections against discrimination or "unreasonable" pricing, which are part of the legal definition of common carriage.
In other words, you'll get two ISPs to choose from, the cable company and the ILEC, and that's it, and they will NOT be required, as unregulated ISPs, to even allow you to connect to the sites you choose. Verizon Online has been blocking tons of legitemate mail lately, for instance, in a profoundly broken anti-spam move. (If it ain't coming from a US server, it's probably spam, they think.)
So if you want a choice of ISPs, COMMENT to the FCC. It's done on line using their Electronic Comment Filing System (ECFS). Go to http://www.fcc.gov/ , then click (top of page) E-filing, then go to ECFS, then read some filed Comments, then add your own. A "brief comment" is easy, just type into the web form, or upload a Word/WP/PDF document. No lawyer necessary.
Docket numbers to plug in to ECFS:
04-405 : BellSouth petition to not be a common carrier. Still open for REPLY COMMENTS, not Comments. (A Reply Comment mentions an already-filed Comment, typically to rebut or add to it. But it's a pretty loose distinction.)
04-440 : Verizon's "me too" petition to not be a common carrier. Open to Comments.
04-416 : Qwest's Petition to be treated as a "non-dominant carrier" in DSL. This doesn't cut off ISPs entirely but treats them as a competitive carrier like Covad, with much looser rules about pricing, for instance. (As dominant providers, they can set their own price, but have to file more paperwork to change it, with more notice.) This docket closes to Comments TOMORROW, January 6, 2005 (a one day extension from the original date).
You're clearly sympathetic to the Bells, but your details on how they do fiber to the home (FTTH) are wrong. They don't use SONET rings for FTTH, and if they did, they wouldn't assign a slot per home, but would do something on demand. SONET technology is advancing rapidly with things like GMPLS. But FTTH typically uses Passive Optical Network (PON) architecture. A PON takes a strand of fiber and optically divides it among up to 32 drops (typically using an 8-way and 4-way splitter). It's a tree, not a ring.
The most standard PON, which I think Verizon is buying, is called APON (A=ATM) or BPON (B=Broadband), depending on who's doing the talking. BPON is an ITU standard so the components are interchangeable between vendors. Tyipcally there's a 622 Mbps (SONET OC-12) downstream on one lambda (wavelength=color), and a 155 Mbps (OC-3) upstream on another lambda. Those carry voice and data. (The upstream transmitters do have to be synchronized, arbitrated, etc.) A third lambda carries analog broadcast video, the cable TV spectrum (tyipcally 54-862 MHz), the same way as cable plant does.
Competing technologies, Ethernet EPON and GPON, are purely packet-based, rather than SONET+analog. These are showing up in a few places but it's largely a religious thing for now. I don't think they are really cheaper, but they're probably better for pure data or mostly-data applications. Packetizing TV channels can get costly, especially for a small system.
>...but if the ILEC were to go under, no one bought them out, and a CLEC was to lay parallel copper to all the ILEC's lines and serve everyone that the ILEC served, they would still be the CLEC and there would be no ILEC.
Well, if you want to be super-picky about it, a CLEC who replaced an ILEC could then become the ILEC, legally, if it wanted to. There's fine print to allow a CLEC to step into ILEC shoes. A Montana rural ILEC set up a CLEC in order to extend service into Qwest territory. (Many rural ILECs, protected against competition in their home territories, compete against neighboring Bells. Think "TDS Metrocom" and "Mid-Maine Tel-Plus" for examples.) They took over more than 90% of the business in one small town. Last month they petitioned the FCC for ILEC status, replacing Qwest.
Why, if ILECs are so over-regulated? Because ILECs are eligible for more rural subsidy money from the Universal Service Fund.
The only reason the Andaman and Nicobar Islands were so familiar to me was because of ham radio. They're well known for being off limits to ham radio operators. I don't know why -- India's had lots of hams for years on the mainland, but they wouldn't let anyone do those islands. So they remained "rare ones" to the main DX award hunters. Hams have been going on "DXpeditions" to rare countries for years, sometimes financed by DXers looking for the contact and QSL card, and it was in the 1960s that I read some travelogues which mentioned trying and failing to get permission to go to "VU4". What's on those islands anyway? (Or what was?)
It's a fortunate coincidence that Charly finally got permission to operate there only a short time before the tsunami!
Surewest is the Incumbent telco in Roseville, but has slipped over the line into PacBell/SBC territory (i.e., the Sacramento area) as a competitor. And since they're not a monopoly there but an upstart, they have to try harder. You're lucky to have a choice. Only a very small percentage of homes have more than one cable or telco. RCN is an alternative cableco in some areas, for instance, but it's bankrupt and such "overbuilds" are not happening on a wide scale, only where the costs are especially low.
The FCC assumes that the RCNs and Surewests of the world will overbuild everywhere, keeping the SBCs on their toes. They also know that that assumption is false, just a convenient lie used as cover for regulatory purposes. The Bells understand that the bulk of their ratepayers won't have alternatives, because it usually isn't profitable to build parallel wires on the poles or in the ground. That's why they were made regulated common carriers in the first place -- before 1912, there was a lot of local telephone competition, but it consisted of lots of competitors hanging their own wire on urban poles, and it got ugly fast.
The DigiMatrix you point to does, indeed, look a lot more appropriate for a home entertainment system.
Alas, only it has a built-in SiS video system, which is rather limited in 3D performance, though apparently adequate for SDTV. And while it says it take a Socket 478 Northwood CPU, I'm bewildered as to how you can fit much of a heat sink into that low profile box! Much as I dislike overly-fat boxes (and I think the base-note's system-cube-with-handle is indeed inappropriate for an entertainment system), I can't help thinking that DigiMatrix could use a little extra headroom, for a slot or two if not just cooling.
But here's the $64 question. Why do you want fiber to the home?
Oh, I know what you expect: Lightning-fast Internet access, right? But you forget that you're dealing with the Bell companies, under the Powell regime at the FCC.
The Bells have a bad case of cable envy. They want to sell you TV channels, sure, because they see TV as the next big thing. (Not TV over fiber, but TV in general. The Bells are still stuck in a 1950 mindset.) And while it is possible to do TV over ADSL, it's not as good as cable. Fiber optics can be as good as cable -- cable companies, after all, bring it to the neighborhood already, converting to coax for the final run (Hybrid Fiber-Coax). FIOS does the optical conversion on a per-house basis. SBC might do that too, but I'm not sure. BellSouth plans to run fiber "to the curb", and tie in to the old twisted-pair drop wire, up to 500 feet of it, which should be able to deliver 20+ Mbps, enough for switched (tell them what channel you want and they'll connect you to it, keeping track of your viewing like a phone call) TV.
But what about Internet? First off, if you have fiber to the home, an alternative DSL provider like Covad is usually cut off, period. (They might be allowed to salvage the old wire. "Green field" developments are closed to competitors tighter than a drum though.)
Second, BellSouth has petitioned the FCC to "forbear" from enforcing the well-established rules of Common Carriage, as well as Computer II obligations, which require a telco-owned competitive service (ISP) to buy the underlying communications service on the same basis as a competitive provider (independent ISP). In other words, BellSouth wants to be allowed to deny access to its network to any other ISP. It's BellSouth Internet or nothing. If you don't like their backbone speed, their mail blocks, their pr0n filters, their no-server-at-home policies, whatever, tough noogies. And with no competitors save cable (and maybe wireless, in a few places, but that'll usually be slower), how do you think their service quality will evolve? (Remember Lily Tomlin as Geraldine the Operator?)
And while it's BellSouth's petition at the FCC now, if it's granted, it'll be precedent for all of the other telcos. Verizon, SBC, Qwest and even that godwaful CenturyTel will get the same treatment. So your choice of ISP will be the telco-owned ISP or the cable-owned ISP.
The FCC just closed out its Comment period on this abomination, but Reply Comments are being taken until Jan. 28 or so. Go to the FCC web site -> e-filing -> ECFS -> search for filed comments -> enter "04-405" as the docket number.
Be afraid. Be very afraid. You may end up missing your creaky old copper DSL.
Indeed, this has been done. It's just not common any more, because cable and satellite are more effective ways of delivering "subscription TV".
Back in the 1970s, there were plenty of these channels, such as Wometco Home Theater, which was sometimes scrambled (subscription-only), and sometimes ran the Uncle Floyd Show (still a legend in New Jersey, where Floyd carries on on cable). You needed a special box, though there were a fair number of plans around to, uh, homebrew the decoder.
I never subscribed myself, and I don't know if they ran much pr0n or other controversial material, but I suspect the Supreme Court's rule against censoring non-broadcast (subscription) services would apply.
In fact, I suspect we'll see more of this, now that Digital TV is on the air. (Tuners are scarce, to be sure, but they exist and will come down in price.) DTV has a 20 Mbps bitstream, which is typically multiplexed into several separate program feeds. So there might be a 10 Mbps HDTV main feed and a half-dozen 1.5 Mbps secondary channels. Already there are companies renting out these secondary streams for subscription services, and over-the-air DTV decoders for these subscription channels. Since these are carrying the same program feeds as cable and satellite, they're not censored.
You make a good point. In your basic Windows dialog, for instance, there's a cruddy mouse-driven picker, but there's also a line where you can type filenames, and clicking the mouse once just fills in that line with the filename you clicked on. The GIMP box seems to be more in line with the WIMPS orthodoxy that requires the mouse. It's annoying and unnecessary. GUIs should be a tool to empower users, not a strait jacket to dictate to them.
> but on a telephone half a second of delay isn't really noticeable
Hello? On a telephone line, half a second of delay is widely viewed as unacceptable.
I design phone networks, and am rather familiar with the rules. When delay is more than 10 milliseconds, it's traditionally normal to do something about the echo, so it isn't noticeable, though today that's not so commonly done. At 50 milliseconds, active echo cancellation is required. Landline telephone service is considered degraded when the delay is more than about 100 ms., which is quite common (rather short, actually) even on high-quality VoIP.
The delay of a satellite conversation -- about half a second round trip -- is simply not tolerated for telephony unless there is no alternative. During the 1970s, IBM, Aetna and Comsat spent a fortune launching a set of satellites, SBS, the first rooftop VSAT network. It was supposed to handle voice and data, offering the first real end-to-end competition to Ma Bell. It flopped big time in the phone space because of the delay. All subsequent domestic satellite projects met the same fate. The international long distance carriers ended up using satellites for only one leg of an undersea call (back when undersea cable bandwidth was expensive; the common routes are cheaper than dirt nowadays) in order to cut round-trip delay in half. Satellites are great for broadcasting (not interactive) but tolerated for telephony only where there's no alternative, places like islands, ships, reporters on the roof of hotels in Baghdad, and remote Arctic villages.
Add satellite delay to VoIP delay and you end up with a kind of weird conversation, almost half-duplex in nature, not at all natural.
If it were rigged, it would be in the button
on
Adieu to Ken Jennings
·
· Score: 2, Interesting
I do not think that it was rigged. Ken Jennings just had the right stuff for the show, the ability to stay unfazed while his opponents got nervous. Some folks are good at it; he's the top of the curve.
Having seen Nightline last night, which was about Jeopardy, and prepared in advance in secret, to be aired when the show where he lost was aired, I see how secure the questions were. It's highly unlikely that anybody could have been feeding him the questions (not the answers -- it's Jeopardy -- duh!). And given the stakes, it's unlikely that they were cheating at all, even if it were to improve ther ratings and profits. Merv's credibility is on the line, after all.
But Jennings himself explained that much of the trick was in the button. It does not get activated until Alex Trebek is finished giving the "answer", and if you press too soon, you get locked out for a fraction of a second. So there's some logic behind the button mechanism.
I was on a TV quiz show once. I do believe that particular one was rigged. No names -- it was probably before most current Slashdotters were born. It too used a button. And I noticed that one team's buttons -- the designated winners' -- responded differently than my team's. Of course those were the days of mechanical relays; game shows weren't computerized yet. It can be quite subtle, but a fraction of a second in timing can make all the difference.
"Indian" is the preferred term. "Native American" is a Pee Cee insult.
The word "Indian", referring to Americans, comes from Spanish. Columbus, seeing the people who lived on Hispaniola, wrote in his log that they were "una gente in dios", Spanish for "naked people". ("En dios", or "in God", referred to their lack of clothing; in Spanish, "en" becomes "in" when it follows a terminal "e".) Thus they became "indios" in Spanish, and from that, "Indians" in English. India, the country, was known to England at the time as Hindustan, not India.
"Native", while literally meaning anyone born somewhere (including second-generation residents), became a term used by European colonialists to refer to the autochthonous peoples of colonized region. It carried a negative connotation, as mere "natives" were viewed as inferior to "settlers".
I didn't say it had one. I said Linux didn't, and if Solaris did... I don't know about Solaris drivers. Since it's widely used for embedded systems, I'd expect more stablity than Linux has though. If that's not the case, well, then fooey.
> Well, depending on the licence, drivers from linux could potentially be ported.
Uh, which ones?
That's a big problem with Linux. Drivers! Linus loves to change the driver ABI. So you have to recompile drivers for each version. Or sometimes rewrite them. Linus does that because it forces drivers to be made available in source form, so they can be recompiled, but hardware vendors hate that, because it exposes information about the product they'd rather keep secret. And in the case of wireless cards, may violate legal regulations.
If Solaris has a stable ABI, it may over time attract more driver support. Even if the rate of driver development is slower, all the old drivers will keep working, unlike Linux drivers.
ZDnet is owned by CNET. The same story is on CNET (http://www.news.com/) today too. But the discussion threads beneath the two are separate, so as, I suppose, to make it look as if ZDnet and CNET were separate properties.
Fahrenheit 9/11 had a whole section on the USAP at Riot Act. The gist: The actual final text was not made available to Congress before it was voted on! It was a huge document, written (probably well before 9/11) by Ashcroft and company, and submitted under the heat of the moment to Congress, with a strong push to DO SOMETHING FAST. Stuff got stuck in at the last minute.
So sure, Kerry voted for it, but he has repeastedly said that he wants to make some changes in it too, in places where it infringes upon civil liberties. Bush wants to "strengthen" it. Quite different.
LinuxWorld is certainly not a pro-Linux site, nor is it ZDnet -- it's put out by SysCon media. Well, at least the "con" sounds right! Maureen O'Gara is simply a SCO mouthpiece. I'm perplexed that CowboyNeal ran with it here, giving it credibility that it doesn't deserve. Maybe he's trying to earn those Cowboy Neal jokes.
But what's the lead story on LinuxWorld today?
Fraud in Linuxland? VA Linux Class Action To Go Forward
Yep, it's an attack on Slashdot's owner! If you can't take the message, dig the dirt on the messenger!
Half right. It has nothing to do with the nationality, national origin, or immigration status of employees. It has a lot to do with what is expected of employees, and what processes are involved in writing software.
I worked for some years for one of IBM's competitors. I wasn't a tech writer there, but looked into it before joining. The software development process involved working closely with the writers. The programmer's job, in essence, was to make it work according to the documentation, not the other way around. The relaese cycle was slow, but it was industrial strength code. Something I miss today.
Oddly, it seems to me that most of the tech writers working around here nowadays -- in English -- are not native speakers. Most are Russian. They take care with the language that a native usually misses. But they're not programmers. It's a rare programmer who can write decent text.
The "part 3" article on VoIP sucks as badly as what came before. This time the idea is that VoIP somehow majikally enables k3wl "services" like distributed call centers. And golly gee it sometimes separates calls signaling from the call path.
Well whoop-de-doodle -- that kind of thing was being done over the TDM-based (circuit) telephone network in the 1980s! In the public switched network, call signaling was divorced from the bearer path in the 1970s to 1980s, with CCIS, and in the 1990s with its replacement, Signaling System 7. (ISDN does something similar to the subscriber, but usually just on a separate channel, not a separate physical path. SS7 is a separate network with diverse physical routing.)
What VoIP does is take a step 30 years backwards, putting signaling and the bearer channel on the same path (inband signaling) 99+% of the time! As modern as touch-tone, maybe, but I saw that at the 1964 World's Fair. SIP and H.323 are both based on inband signaling. MGCP usually is.
But more importantly, the creation of an enhanced service, be it a call center, unified messaging, or anything else for that matter, has no dependency on the bearer path's being over eye pee! Since signaling can be out of band and between computers, it's possible to do all of those services with TDM channels, ATM channels, or whatever other channels (bearer paths) you have. Just because a Cisco IP phone has a slick cellular-like display does not mean that you need VoIP to have a slick display -- that's the control circuit, totally independent of the bearer channel.
It's like comparing the economy of the United States and the economy of Mexico, and coming to the conclusion that El Norte's wealth is caused by speaking English. It's a cargo cult.
There are of course a few handy applications for VoIP, but most boil down to clever parasitic applications of bandwidth, and to regulatory arbitrage.
VoIP's main benefit as a PBX is to the vendors, who need something new to sell. I lived through the last really big PBX migration. The first stage was when computerized PBXs came out (mid-1970s). AT&T/Bell didn't sell, they rented, but competitors (called "interconnect companies" in those days) sold. And AT&T had no digital PBX until 1983; its previous flagship, Dimension, was analog. Having a CPU in lieu of relays did provide for good features, but their competitors (Nortel, Rolm, NEC, Mitel, etc.) actually had digital switching much earlier.
Come the Bell divestiture in 1984, all of the once-tariffed PBXs were left with AT&T-IS, who did NOT want to maintain any remaining mechanical stuff. Either they sold you a new one, or somebody else did. So between 1974 and 1984, pretty much all of the PBX installed base turned over. The last analog Dimensions were mostly pulled around 1994, when interchangeable area codes (e.g., 270, 978) came in, and Lucent didn't want to rewrite the software. That replacement cycle was beneficial all around, because the new digital PBXs were a lot better, not to mention a lot cheaper to maintain than their predecessors.
But once you had a digital PBX, why upgrade? Fact is, they still work just fine! A 1995-vintage PBX can take digital trunk interfaces (T1/PRI), digital feature phones with displays and multi-line support, and all sorts of neat features. Administration's a breeze too -- your basic telecom manager wants centralized control, to type in office moves into a terminal, not to have people carry phones around with them!
Not that VoIP enables such features even if you want them. Nortel introduced "Automatic Station Relocation" in 1983! If the manager enabled it (hah!), then the user could dial a feature code, pick up their phone, plug it in the new jack, dial a code, and voila, their line was moved. VoIP advocates think this stuff is new! Okay, so the PBX reads the MAC address out of the phone instead of needing a code to be dialed. But management still probably doesn't want them to move it on their own!
So the PBX vendors are looking for a way to force turnover of otherwise perfectly-good kit. The price of new conventional PBX gear is pretty low already; industrial-strength (not Asterix) VoIP PBXs aren't a great bargain. Telling people that the old PBX is obsolete, and they need VoIP, or they won't be k3wl and 1337, is a marketing technique. But that doesn't make it so.
Unlicensed, 70 Mbps, 30 miles (pick one)
on
WiMax: When, Not If
·
· Score: 1
There's a whole lot of misunderstanding about WiMax floating around! The hype is incredible. You'd think Dick Cheney himself had invented it.;-)
Seriously, the problem is not that WiMax is not interesting technology, but that it's still just a way to use radio waves. It's no miracle. It's designed for longer range than WiFi, but that doesn't overcome licensing issues. So what it boils down to is what my summary said. The good news is that WiMax offers:
1) Unlicensed or licensed operation. Most interest worldwide, btw, is licensed, with the assumption that big rich service providers will pay their government for licenses. Especially the 3.5 GHz fixed-wireless band, which is found in most countries except the United States. Unlicensed, however, is subject to the usual power limits, so the range is necessarily very limited, unless there's a very good antenna, a very clear shot, modest speed, and nobody else causing interference.
2) 70 Mbps speed. Well, maybe, but remember that bits = power, so if you go faster, you have less range than if you go slower. WiMax lets you have a wide range of speeds, to trade off with range. Sort of like WiFi in that regard.
3) 30 mile range. Well, that's a typical number for high-power (as in licensed) microwave radio systems in the 2-10 GHz bands, if there's a clear path. WiMax can handle some "non-line-of-sight" paths, but that doesn't mean blocked by a hill or horizon, it means that there's multipath, meaning that something is reflecting the signal to the destination, and the different reflections arrive at slightly different times. There still has to be a decent-strength signal. Unlicensed range is more likely to be under a mile, depending on speed and path; in some cases more like a block or two.
So the bad news is "pick one", not pick three. Not even as good as engineering's "faster, better, cheaper, pick two".
Now wrt long-range frequencies in the USA, the FCC is rejiggering some frequencies in the 2.5 GHz range to make it more flexible for licensees who want to offer WiMax (or other data) in lieu of video. And they're talking about a 3.6 GHz unlicensed band, at least for use in rural areas away from the coasts, with more power than the existing WiFi and U-NII (5 GHz) bands. But it ain't there yet.
You're full of crap, and I do know of what I speak, being well connected to the phone industry.
The AT&T domestic backbone was for all intents and purposes entirely circuit-switched (TDM nowadays) until the past couple of years. The Incumbent Local Exchange Carriers are almost entirely circuit-switched. MCI and Sprint are circuit switched. The phone backbones use SONET transmission and TDM switches.
There are narrow examples of packet switching in the phone network. AT&T rolled out a Frame Relay-based system in the late 1980s, to compress the bandwidth on high-cost international links. Others have done similar things, for the same reason, with various vendors' technology. It's called "Digital Circuit Multiplication Equipment" (DCME). But domestic SONET bandwidth is so cheap that DCME isn't worth using, especially since DCME degrades call quality. And it's used below the switch layer, to make more TDM virtual channels.
There's a fair amount of ATM (asynchronous transfer mode) floating around too. Some LD carriers may use it, mostly as a multiplexing technique over SONET pipes. Verizon has started using Nortel's ATM switches in its own local networks, especially in New Jersey. But it's still a minority technology. AT&T (back when it owned Lucent, Bell Labs, etc.) was a major advocate of ATM technology, but hardly invented it in house by itself (more like the work a committee -- and I was on the committee, so I know why the camel's so humpy). AT&T also pushed hard for Frame Relay, largely to mix voice and data on DCME.
In today's LD world, Qwest's backbone is mostly VoIP, using Sonus switches. Ditto Wiltel, a fairly small player. AT&T has started to migrate to VoIP, as have Sprint and MCI, but it's not "over the Internet", it's just using IP headers as a multiplexing technique on fat dedicated circuits. Mainly because the VoIP switches are really cheap, and because Wall Street expects it. (The new switches do circuit switching too; the cost per minute differential is negligible. VoIP actually uses more bandwidth, regardless of what the propagandists claim, but fat TDM switch ports are a bit costlier.)
In the local world, VoIP is coming on strong in the form of PacketCable, again not over the public Internet. And of course Vonage and its imitators, who for all their bluster have a combined market share of far less than one percent. A little VoIP gets used here and there by other carriers. The Bells are experimenting with it, but it's a negligible share of their traffic. They're backward monopolists, but they also have reasonably high service-level standards, and they know how hard it is to do that with VoIP.
The signaling network is packet switched (Signaling System 7), but that's a whole story of its own.
Obviously, if absolute uptime is your only criteria, then Solaris is not a bad choice, but I would take an IBM mainframe over it.
An IBM mainframe to run the Call Agent program in a CO switch? Methinks not -- mainframes are not embedded processors! And modern CO switches are not big iron; they usually take about a third of a rack; half a rack for a really big one. Sometimes with a couple of Netras on the side.
Again, you're thinking about "computers", while I'm talking about critical "processor-controlled systems". Solaris has a head start over Linux in that arena. Years of building Netras and NEBS-compliant systems has given Sun experience that Linux distributors don't have.
And for some reason a large pack of Telecom vendors is busy investing in Carrier Grade Linux.
Umm, funny that you mention wireless drivers and nvidia 3d - because - solaris has NEITHER.
Sure, because it's not a game machine or a laptop system. It's for mainframes, high-reliability embedded systems, and other specialized apps. Games are a nice toy, but Solaris isn't aimed at the toy market. "Toy" in not an insult here; it's a recognition of the fun nature of Linux, something Solaris isn't going for. Hell, I wouldn't want Solaris on my desktop!
"Some day my prince will come." It ain't here yet. That's the point.
And Linux is hot stuff in embedded too. And very much because it's drivers are open source drivers! Need to access some obscure USB devices using a ARM/MIPS/superH system? No prob, just compile the driver. With binary abi drivers, they would only exist for x86.
Again, when building a big machine that happens to have an embedded computer controlling it, I don't care if the machine can be controlled by lots of other people's CPUs. It's not an issue.
Besides, for embedded systems, there is no need to follow mainline anyway. Just freeze at a good enough kernel and merge in the most important fixes back from mainline. even 2.0 kernels get fixes, and I'm pretty sure we will see Linux 2.4 systems still runnning 10y from now.
Sure, but then what happened to that "carrier grade" project? If I freeze at some ancient kernel, sure, I have the source and can hack it myself, but I don't get the new goodies, unless I and *all of my hardware suppliers* give me support for the new ABI.
In other words, Solaris is becoming a mainframe dinosaur for those systems where availability requirements can't be implemented by cluster-style redunancy.
Yes Solaris is a very nice, well engineered System. Unfortunatly, for most tasks Linux is good enough...
Indeed, for most tasks, Linux is good enough. When did I say it wasn't? That's the whole point -- Solaris is better for a specific class of high-cost applications for which its additional reliability features and binary stability are worth the money. Linux is better for desktops, clusterable servers, and a lot of other random applications.
But damn, I hate how Linux often makes me upgrade applications when I upgrade the base system. Sure, the sources are out there, but I really like the way VMS has binary compatibility going back to 1978.
Well, no. SCO was not used for call control in real-time telecom switching. No form of Unix was until recently. Administrative systems, operational support systems, and other back-office telecom systems are 180 degrees different from embedded call processing.
And when you buy a CO switch, you don't ask about plugging in your own hardware! You want the damned thing to work with at least five nines of reliability, and you never take it down -- it has duplex CPUs and you just switch over when you need to work on one. Linux still has this notion that you have to turn it off in order to upgrade the version. Not that Solaris is really mature in that regard, but it's ahead of Linux.
Slashdot Mirror
The FCC is supposed to enforce the Telecom Act, and it is supposed to follow the Administrative Procedures Act in doing so. This means that major decisions have to be done in the open, with a public Comment period. Sure, they can try to ignore the Comments, but the Record they create helps on appeal. The Courts in general do not like this FCC, which tends to be viewed as a bunch of cowboys.
There are some urgent Dockets open now at the Wireline [prevention-of] Competition Bureau (WCB). Two, from BellSouth and Verizon, call for their DSL service (and ATM and Frame Relay, in VZ's case, and possibly T1 and T3 leased line in at least BellSouth's case) to be reclassified as "private", rather than "common carriage". This means that they will be under no obligaton whatsoever to allow competing ISPs, like Speakeasy and Earthlink, to use them! (Their excuse: Cable doesn't haveta so they doan wanna.) Since cable (never a common carrier by law) doesn't provide an alternative, most ISPs will be shut off from their customers and thus put out of business. (Dial-up might remain open, but they're working on it, believe me. I'm in the middle of it.) If they do deign to allow an independent ISP onto their wire, it won't be subject to protections against discrimination or "unreasonable" pricing, which are part of the legal definition of common carriage.
In other words, you'll get two ISPs to choose from, the cable company and the ILEC, and that's it, and they will NOT be required, as unregulated ISPs, to even allow you to connect to the sites you choose. Verizon Online has been blocking tons of legitemate mail lately, for instance, in a profoundly broken anti-spam move. (If it ain't coming from a US server, it's probably spam, they think.)
So if you want a choice of ISPs, COMMENT to the FCC. It's done on line using their Electronic Comment Filing System (ECFS). Go to http://www.fcc.gov/ , then click (top of page) E-filing, then go to ECFS, then read some filed Comments, then add your own. A "brief comment" is easy, just type into the web form, or upload a Word/WP/PDF document. No lawyer necessary.
Docket numbers to plug in to ECFS:
04-405 : BellSouth petition to not be a common carrier. Still open for REPLY COMMENTS, not Comments. (A Reply Comment mentions an already-filed Comment, typically to rebut or add to it. But it's a pretty loose distinction.)
04-440 : Verizon's "me too" petition to not be a common carrier. Open to Comments.
04-416 : Qwest's Petition to be treated as a "non-dominant carrier" in DSL. This doesn't cut off ISPs entirely but treats them as a competitive carrier like Covad, with much looser rules about pricing, for instance. (As dominant providers, they can set their own price, but have to file more paperwork to change it, with more notice.) This docket closes to Comments TOMORROW, January 6, 2005 (a one day extension from the original date).
Will Team Slashdot come through?
You're clearly sympathetic to the Bells, but your details on how they do fiber to the home (FTTH) are wrong. They don't use SONET rings for FTTH, and if they did, they wouldn't assign a slot per home, but would do something on demand. SONET technology is advancing rapidly with things like GMPLS. But FTTH typically uses Passive Optical Network (PON) architecture. A PON takes a strand of fiber and optically divides it among up to 32 drops (typically using an 8-way and 4-way splitter). It's a tree, not a ring.
The most standard PON, which I think Verizon is buying, is called APON (A=ATM) or BPON (B=Broadband), depending on who's doing the talking. BPON is an ITU standard so the components are interchangeable between vendors. Tyipcally there's a 622 Mbps (SONET OC-12) downstream on one lambda (wavelength=color), and a 155 Mbps (OC-3) upstream on another lambda. Those carry voice and data. (The upstream transmitters do have to be synchronized, arbitrated, etc.) A third lambda carries analog broadcast video, the cable TV spectrum (tyipcally 54-862 MHz), the same way as cable plant does.
Competing technologies, Ethernet EPON and GPON, are purely packet-based, rather than SONET+analog. These are showing up in a few places but it's largely a religious thing for now. I don't think they are really cheaper, but they're probably better for pure data or mostly-data applications. Packetizing TV channels can get costly, especially for a small system.
> ...but if the ILEC were to go under, no one bought them out, and a CLEC was to lay parallel copper to all the ILEC's lines and serve everyone that the ILEC served, they would still be the CLEC and there would be no ILEC.
Well, if you want to be super-picky about it, a CLEC who replaced an ILEC could then become the ILEC, legally, if it wanted to. There's fine print to allow a CLEC to step into ILEC shoes. A Montana rural ILEC set up a CLEC in order to extend service into Qwest territory. (Many rural ILECs, protected against competition in their home territories, compete against neighboring Bells. Think "TDS Metrocom" and "Mid-Maine Tel-Plus" for examples.) They took over more than 90% of the business in one small town. Last month they petitioned the FCC for ILEC status, replacing Qwest.
Why, if ILECs are so over-regulated? Because ILECs are eligible for more rural subsidy money from the Universal Service Fund.
Strange but true.
The only reason the Andaman and Nicobar Islands were so familiar to me was because of ham radio. They're well known for being off limits to ham radio operators. I don't know why -- India's had lots of hams for years on the mainland, but they wouldn't let anyone do those islands. So they remained "rare ones" to the main DX award hunters. Hams have been going on "DXpeditions" to rare countries for years, sometimes financed by DXers looking for the contact and QSL card, and it was in the 1960s that I read some travelogues which mentioned trying and failing to get permission to go to "VU4". What's on those islands anyway? (Or what was?)
It's a fortunate coincidence that Charly finally got permission to operate there only a short time before the tsunami!
Surewest is the Incumbent telco in Roseville, but has slipped over the line into PacBell/SBC territory (i.e., the Sacramento area) as a competitor. And since they're not a monopoly there but an upstart, they have to try harder. You're lucky to have a choice. Only a very small percentage of homes have more than one cable or telco. RCN is an alternative cableco in some areas, for instance, but it's bankrupt and such "overbuilds" are not happening on a wide scale, only where the costs are especially low.
The FCC assumes that the RCNs and Surewests of the world will overbuild everywhere, keeping the SBCs on their toes. They also know that that assumption is false, just a convenient lie used as cover for regulatory purposes. The Bells understand that the bulk of their ratepayers won't have alternatives, because it usually isn't profitable to build parallel wires on the poles or in the ground. That's why they were made regulated common carriers in the first place -- before 1912, there was a lot of local telephone competition, but it consisted of lots of competitors hanging their own wire on urban poles, and it got ugly fast.
The DigiMatrix you point to does, indeed, look a lot more appropriate for a home entertainment system.
Alas, only it has a built-in SiS video system, which is rather limited in 3D performance, though apparently adequate for SDTV. And while it says it take a Socket 478 Northwood CPU, I'm bewildered as to how you can fit much of a heat sink into that low profile box! Much as I dislike overly-fat boxes (and I think the base-note's system-cube-with-handle is indeed inappropriate for an entertainment system), I can't help thinking that DigiMatrix could use a little extra headroom, for a slot or two if not just cooling.
But here's the $64 question. Why do you want fiber to the home?
Oh, I know what you expect: Lightning-fast Internet access, right? But you forget that you're dealing with the Bell companies, under the Powell regime at the FCC.
The Bells have a bad case of cable envy. They want to sell you TV channels, sure, because they see TV as the next big thing. (Not TV over fiber, but TV in general. The Bells are still stuck in a 1950 mindset.) And while it is possible to do TV over ADSL, it's not as good as cable. Fiber optics can be as good as cable -- cable companies, after all, bring it to the neighborhood already, converting to coax for the final run (Hybrid Fiber-Coax). FIOS does the optical conversion on a per-house basis. SBC might do that too, but I'm not sure. BellSouth plans to run fiber "to the curb", and tie in to the old twisted-pair drop wire, up to 500 feet of it, which should be able to deliver 20+ Mbps, enough for switched (tell them what channel you want and they'll connect you to it, keeping track of your viewing like a phone call) TV.
But what about Internet? First off, if you have fiber to the home, an alternative DSL provider like Covad is usually cut off, period. (They might be allowed to salvage the old wire. "Green field" developments are closed to competitors tighter than a drum though.)
Second, BellSouth has petitioned the FCC to "forbear" from enforcing the well-established rules of Common Carriage, as well as Computer II obligations, which require a telco-owned competitive service (ISP) to buy the underlying communications service on the same basis as a competitive provider (independent ISP). In other words, BellSouth wants to be allowed to deny access to its network to any other ISP. It's BellSouth Internet or nothing. If you don't like their backbone speed, their mail blocks, their pr0n filters, their no-server-at-home policies, whatever, tough noogies. And with no competitors save cable (and maybe wireless, in a few places, but that'll usually be slower), how do you think their service quality will evolve? (Remember Lily Tomlin as Geraldine the Operator?)
And while it's BellSouth's petition at the FCC now, if it's granted, it'll be precedent for all of the other telcos. Verizon, SBC, Qwest and even that godwaful CenturyTel will get the same treatment. So your choice of ISP will be the telco-owned ISP or the cable-owned ISP.
The FCC just closed out its Comment period on this abomination, but Reply Comments are being taken until Jan. 28 or so. Go to the FCC web site -> e-filing -> ECFS -> search for filed comments -> enter "04-405" as the docket number.
Be afraid. Be very afraid. You may end up missing your creaky old copper DSL.
Indeed, this has been done. It's just not common any more, because cable and satellite are more effective ways of delivering "subscription TV".
Back in the 1970s, there were plenty of these channels, such as Wometco Home Theater, which was sometimes scrambled (subscription-only), and sometimes ran the Uncle Floyd Show (still a legend in New Jersey, where Floyd carries on on cable). You needed a special box, though there were a fair number of plans around to, uh, homebrew the decoder.
I never subscribed myself, and I don't know if they ran much pr0n or other controversial material, but I suspect the Supreme Court's rule against censoring non-broadcast (subscription) services would apply.
In fact, I suspect we'll see more of this, now that Digital TV is on the air. (Tuners are scarce, to be sure, but they exist and will come down in price.) DTV has a 20 Mbps bitstream, which is typically multiplexed into several separate program feeds. So there might be a 10 Mbps HDTV main feed and a half-dozen 1.5 Mbps secondary channels. Already there are companies renting out these secondary streams for subscription services, and over-the-air DTV decoders for these subscription channels. Since these are carrying the same program feeds as cable and satellite, they're not censored.
You make a good point. In your basic Windows dialog, for instance, there's a cruddy mouse-driven picker, but there's also a line where you can type filenames, and clicking the mouse once just fills in that line with the filename you clicked on. The GIMP box seems to be more in line with the WIMPS orthodoxy that requires the mouse. It's annoying and unnecessary. GUIs should be a tool to empower users, not a strait jacket to dictate to them.
> but on a telephone half a second of delay isn't really noticeable
Hello? On a telephone line, half a second of delay is widely viewed as unacceptable.
I design phone networks, and am rather familiar with the rules. When delay is more than 10 milliseconds, it's traditionally normal to do something about the echo, so it isn't noticeable, though today that's not so commonly done. At 50 milliseconds, active echo cancellation is required. Landline telephone service is considered degraded when the delay is more than about 100 ms., which is quite common (rather short, actually) even on high-quality VoIP.
The delay of a satellite conversation -- about half a second round trip -- is simply not tolerated for telephony unless there is no alternative. During the 1970s, IBM, Aetna and Comsat spent a fortune launching a set of satellites, SBS, the first rooftop VSAT network. It was supposed to handle voice and data, offering the first real end-to-end competition to Ma Bell. It flopped big time in the phone space because of the delay. All subsequent domestic satellite projects met the same fate. The international long distance carriers ended up using satellites for only one leg of an undersea call (back when undersea cable bandwidth was expensive; the common routes are cheaper than dirt nowadays) in order to cut round-trip delay in half. Satellites are great for broadcasting (not interactive) but tolerated for telephony only where there's no alternative, places like islands, ships, reporters on the roof of hotels in Baghdad, and remote Arctic villages.
Add satellite delay to VoIP delay and you end up with a kind of weird conversation, almost half-duplex in nature, not at all natural.
I do not think that it was rigged. Ken Jennings just had the right stuff for the show, the ability to stay unfazed while his opponents got nervous. Some folks are good at it; he's the top of the curve.
Having seen Nightline last night, which was about Jeopardy, and prepared in advance in secret, to be aired when the show where he lost was aired, I see how secure the questions were. It's highly unlikely that anybody could have been feeding him the questions (not the answers -- it's Jeopardy -- duh!). And given the stakes, it's unlikely that they were cheating at all, even if it were to improve ther ratings and profits. Merv's credibility is on the line, after all.
But Jennings himself explained that much of the trick was in the button. It does not get activated until Alex Trebek is finished giving the "answer", and if you press too soon, you get locked out for a fraction of a second. So there's some logic behind the button mechanism.
I was on a TV quiz show once. I do believe that particular one was rigged. No names -- it was probably before most current Slashdotters were born. It too used a button. And I noticed that one team's buttons -- the designated winners' -- responded differently than my team's. Of course those were the days of mechanical relays; game shows weren't computerized yet. It can be quite subtle, but a fraction of a second in timing can make all the difference.
I just found Ken's observation interesting.
"Indian" is the preferred term. "Native American" is a Pee Cee insult.
The word "Indian", referring to Americans, comes from Spanish. Columbus, seeing the people who lived on Hispaniola, wrote in his log that they were "una gente in dios", Spanish for "naked people". ("En dios", or "in God", referred to their lack of clothing; in Spanish, "en" becomes "in" when it follows a terminal "e".) Thus they became "indios" in Spanish, and from that, "Indians" in English. India, the country, was known to England at the time as Hindustan, not India.
"Native", while literally meaning anyone born somewhere (including second-generation residents), became a term used by European colonialists to refer to the autochthonous peoples of colonized region. It carried a negative connotation, as mere "natives" were viewed as inferior to "settlers".
I didn't say it had one. I said Linux didn't, and if Solaris did... I don't know about Solaris drivers. Since it's widely used for embedded systems, I'd expect more stablity than Linux has though. If that's not the case, well, then fooey.
> Well, depending on the licence, drivers from linux could potentially be ported.
Uh, which ones?
That's a big problem with Linux. Drivers! Linus loves to change the driver ABI. So you have to recompile drivers for each version. Or sometimes rewrite them. Linus does that because it forces drivers to be made available in source form, so they can be recompiled, but hardware vendors hate that, because it exposes information about the product they'd rather keep secret. And in the case of wireless cards, may violate legal regulations.
If Solaris has a stable ABI, it may over time attract more driver support. Even if the rate of driver development is slower, all the old drivers will keep working, unlike Linux drivers.
ZDnet is owned by CNET. The same story is on CNET (http://www.news.com/) today too. But the discussion threads beneath the two are separate, so as, I suppose, to make it look as if ZDnet and CNET were separate properties.
Fahrenheit 9/11 had a whole section on the USAP at Riot Act. The gist: The actual final text was not made available to Congress before it was voted on! It was a huge document, written (probably well before 9/11) by Ashcroft and company, and submitted under the heat of the moment to Congress, with a strong push to DO SOMETHING FAST. Stuff got stuck in at the last minute.
So sure, Kerry voted for it, but he has repeastedly said that he wants to make some changes in it too, in places where it infringes upon civil liberties. Bush wants to "strengthen" it. Quite different.
LinuxWorld is certainly not a pro-Linux site, nor is it ZDnet -- it's put out by SysCon media. Well, at least the "con" sounds right! Maureen O'Gara is simply a SCO mouthpiece. I'm perplexed that CowboyNeal ran with it here, giving it credibility that it doesn't deserve. Maybe he's trying to earn those Cowboy Neal jokes.
But what's the lead story on LinuxWorld today?
Fraud in Linuxland? VA Linux Class Action To Go Forward
Yep, it's an attack on Slashdot's owner! If you can't take the message, dig the dirt on the messenger!
Half right. It has nothing to do with the nationality, national origin, or immigration status of employees. It has a lot to do with what is expected of employees, and what processes are involved in writing software.
I worked for some years for one of IBM's competitors. I wasn't a tech writer there, but looked into it before joining. The software development process involved working closely with the writers. The programmer's job, in essence, was to make it work according to the documentation, not the other way around. The relaese cycle was slow, but it was industrial strength code. Something I miss today.
Oddly, it seems to me that most of the tech writers working around here nowadays -- in English -- are not native speakers. Most are Russian. They take care with the language that a native usually misses. But they're not programmers. It's a rare programmer who can write decent text.
The "part 3" article on VoIP sucks as badly as what came before. This time the idea is that VoIP somehow majikally enables k3wl "services" like distributed call centers. And golly gee it sometimes separates calls signaling from the call path.
Well whoop-de-doodle -- that kind of thing was being done over the TDM-based (circuit) telephone network in the 1980s! In the public switched network, call signaling was divorced from the bearer path in the 1970s to 1980s, with CCIS, and in the 1990s with its replacement, Signaling System 7. (ISDN does something similar to the subscriber, but usually just on a separate channel, not a separate physical path. SS7 is a separate network with diverse physical routing.)
What VoIP does is take a step 30 years backwards, putting signaling and the bearer channel on the same path (inband signaling) 99+% of the time! As modern as touch-tone, maybe, but I saw that at the 1964 World's Fair. SIP and H.323 are both based on inband signaling. MGCP usually is.
But more importantly, the creation of an enhanced service, be it a call center, unified messaging, or anything else for that matter, has no dependency on the bearer path's being over eye pee! Since signaling can be out of band and between computers, it's possible to do all of those services with TDM channels, ATM channels, or whatever other channels (bearer paths) you have. Just because a Cisco IP phone has a slick cellular-like display does not mean that you need VoIP to have a slick display -- that's the control circuit, totally independent of the bearer channel.
It's like comparing the economy of the United States and the economy of Mexico, and coming to the conclusion that El Norte's wealth is caused by speaking English. It's a cargo cult.
There are of course a few handy applications for VoIP, but most boil down to clever parasitic applications of bandwidth, and to regulatory arbitrage.
VoIP's main benefit as a PBX is to the vendors, who need something new to sell. I lived through the last really big PBX migration. The first stage was when computerized PBXs came out (mid-1970s). AT&T/Bell didn't sell, they rented, but competitors (called "interconnect companies" in those days) sold. And AT&T had no digital PBX until 1983; its previous flagship, Dimension, was analog. Having a CPU in lieu of relays did provide for good features, but their competitors (Nortel, Rolm, NEC, Mitel, etc.) actually had digital switching much earlier.
Come the Bell divestiture in 1984, all of the once-tariffed PBXs were left with AT&T-IS, who did NOT want to maintain any remaining mechanical stuff. Either they sold you a new one, or somebody else did. So between 1974 and 1984, pretty much all of the PBX installed base turned over. The last analog Dimensions were mostly pulled around 1994, when interchangeable area codes (e.g., 270, 978) came in, and Lucent didn't want to rewrite the software. That replacement cycle was beneficial all around, because the new digital PBXs were a lot better, not to mention a lot cheaper to maintain than their predecessors.
But once you had a digital PBX, why upgrade? Fact is, they still work just fine! A 1995-vintage PBX can take digital trunk interfaces (T1/PRI), digital feature phones with displays and multi-line support, and all sorts of neat features. Administration's a breeze too -- your basic telecom manager wants centralized control, to type in office moves into a terminal, not to have people carry phones around with them!
Not that VoIP enables such features even if you want them. Nortel introduced "Automatic Station Relocation" in 1983! If the manager enabled it (hah!), then the user could dial a feature code, pick up their phone, plug it in the new jack, dial a code, and voila, their line was moved. VoIP advocates think this stuff is new! Okay, so the PBX reads the MAC address out of the phone instead of needing a code to be dialed. But management still probably doesn't want them to move it on their own!
So the PBX vendors are looking for a way to force turnover of otherwise perfectly-good kit. The price of new conventional PBX gear is pretty low already; industrial-strength (not Asterix) VoIP PBXs aren't a great bargain. Telling people that the old PBX is obsolete, and they need VoIP, or they won't be k3wl and 1337, is a marketing technique. But that doesn't make it so.
There's a whole lot of misunderstanding about WiMax floating around! The hype is incredible. You'd think Dick Cheney himself had invented it. ;-)
Seriously, the problem is not that WiMax is not interesting technology, but that it's still just a way to use radio waves. It's no miracle. It's designed for longer range than WiFi, but that doesn't overcome licensing issues. So what it boils down to is what my summary said. The good news is that WiMax offers:
1) Unlicensed or licensed operation. Most interest worldwide, btw, is licensed, with the assumption that big rich service providers will pay their government for licenses. Especially the 3.5 GHz fixed-wireless band, which is found in most countries except the United States. Unlicensed, however, is subject to the usual power limits, so the range is necessarily very limited, unless there's a very good antenna, a very clear shot, modest speed, and nobody else causing interference.
2) 70 Mbps speed. Well, maybe, but remember that bits = power, so if you go faster, you have less range than if you go slower. WiMax lets you have a wide range of speeds, to trade off with range. Sort of like WiFi in that regard.
3) 30 mile range. Well, that's a typical number for high-power (as in licensed) microwave radio systems in the 2-10 GHz bands, if there's a clear path. WiMax can handle some "non-line-of-sight" paths, but that doesn't mean blocked by a hill or horizon, it means that there's multipath, meaning that something is reflecting the signal to the destination, and the different reflections arrive at slightly different times. There still has to be a decent-strength signal. Unlicensed range is more likely to be under a mile, depending on speed and path; in some cases more like a block or two.
So the bad news is "pick one", not pick three. Not even as good as engineering's "faster, better, cheaper, pick two".
Now wrt long-range frequencies in the USA, the FCC is rejiggering some frequencies in the 2.5 GHz range to make it more flexible for licensees who want to offer WiMax (or other data) in lieu of video. And they're talking about a 3.6 GHz unlicensed band, at least for use in rural areas away from the coasts, with more power than the existing WiFi and U-NII (5 GHz) bands. But it ain't there yet.
You're full of crap, and I do know of what I speak, being well connected to the phone industry.
The AT&T domestic backbone was for all intents and purposes entirely circuit-switched (TDM nowadays) until the past couple of years. The Incumbent Local Exchange Carriers are almost entirely circuit-switched. MCI and Sprint are circuit switched. The phone backbones use SONET transmission and TDM switches.
There are narrow examples of packet switching in the phone network. AT&T rolled out a Frame Relay-based system in the late 1980s, to compress the bandwidth on high-cost international links. Others have done similar things, for the same reason, with various vendors' technology. It's called "Digital Circuit Multiplication Equipment" (DCME). But domestic SONET bandwidth is so cheap that DCME isn't worth using, especially since DCME degrades call quality. And it's used below the switch layer, to make more TDM virtual channels.
There's a fair amount of ATM (asynchronous transfer mode) floating around too. Some LD carriers may use it, mostly as a multiplexing technique over SONET pipes. Verizon has started using Nortel's ATM switches in its own local networks, especially in New Jersey. But it's still a minority technology. AT&T (back when it owned Lucent, Bell Labs, etc.) was a major advocate of ATM technology, but hardly invented it in house by itself (more like the work a committee -- and I was on the committee, so I know why the camel's so humpy). AT&T also pushed hard for Frame Relay, largely to mix voice and data on DCME.
In today's LD world, Qwest's backbone is mostly VoIP, using Sonus switches. Ditto Wiltel, a fairly small player. AT&T has started to migrate to VoIP, as have Sprint and MCI, but it's not "over the Internet", it's just using IP headers as a multiplexing technique on fat dedicated circuits. Mainly because the VoIP switches are really cheap, and because Wall Street expects it. (The new switches do circuit switching too; the cost per minute differential is negligible. VoIP actually uses more bandwidth, regardless of what the propagandists claim, but fat TDM switch ports are a bit costlier.)
In the local world, VoIP is coming on strong in the form of PacketCable, again not over the public Internet. And of course Vonage and its imitators, who for all their bluster have a combined market share of far less than one percent. A little VoIP gets used here and there by other carriers. The Bells are experimenting with it, but it's a negligible share of their traffic. They're backward monopolists, but they also have reasonably high service-level standards, and they know how hard it is to do that with VoIP.
The signaling network is packet switched (Signaling System 7), but that's a whole story of its own.
Obviously, if absolute uptime is your only criteria, then Solaris is not a bad choice, but I would take an IBM mainframe over it.
An IBM mainframe to run the Call Agent program in a CO switch? Methinks not -- mainframes are not embedded processors! And modern CO switches are not big iron; they usually take about a third of a rack; half a rack for a really big one. Sometimes with a couple of Netras on the side.
Again, you're thinking about "computers", while I'm talking about critical "processor-controlled systems". Solaris has a head start over Linux in that arena. Years of building Netras and NEBS-compliant systems has given Sun experience that Linux distributors don't have.
And for some reason a large pack of Telecom vendors is busy investing in Carrier Grade Linux.
Umm, funny that you mention wireless drivers and nvidia 3d - because - solaris has NEITHER.
Sure, because it's not a game machine or a laptop system. It's for mainframes, high-reliability embedded systems, and other specialized apps. Games are a nice toy, but Solaris isn't aimed at the toy market. "Toy" in not an insult here; it's a recognition of the fun nature of Linux, something Solaris isn't going for. Hell, I wouldn't want Solaris on my desktop!
"Some day my prince will come." It ain't here yet. That's the point.
And Linux is hot stuff in embedded too. And very much because it's drivers are open source drivers! Need to access some obscure USB devices using a ARM/MIPS/superH system? No prob, just compile the driver. With binary abi drivers, they would only exist for x86.
Again, when building a big machine that happens to have an embedded computer controlling it, I don't care if the machine can be controlled by lots of other people's CPUs. It's not an issue.
Besides, for embedded systems, there is no need to follow mainline anyway. Just freeze at a good enough kernel and merge in the most important fixes back from mainline. even 2.0 kernels get fixes, and I'm pretty sure we will see Linux 2.4 systems still runnning 10y from now.
Sure, but then what happened to that "carrier grade" project? If I freeze at some ancient kernel, sure, I have the source and can hack it myself, but I don't get the new goodies, unless I and *all of my hardware suppliers* give me support for the new ABI.
In other words, Solaris is becoming a mainframe dinosaur for those systems where availability requirements can't be implemented by cluster-style redunancy.
Yes Solaris is a very nice, well engineered System. Unfortunatly, for most tasks Linux is good enough...
Indeed, for most tasks, Linux is good enough. When did I say it wasn't? That's the whole point -- Solaris is better for a specific class of high-cost applications for which its additional reliability features and binary stability are worth the money. Linux is better for desktops, clusterable servers, and a lot of other random applications.
But damn, I hate how Linux often makes me upgrade applications when I upgrade the base system. Sure, the sources are out there, but I really like the way VMS has binary compatibility going back to 1978.
Well, no. SCO was not used for call control in real-time telecom switching. No form of Unix was until recently. Administrative systems, operational support systems, and other back-office telecom systems are 180 degrees different from embedded call processing.
And when you buy a CO switch, you don't ask about plugging in your own hardware! You want the damned thing to work with at least five nines of reliability, and you never take it down -- it has duplex CPUs and you just switch over when you need to work on one. Linux still has this notion that you have to turn it off in order to upgrade the version. Not that Solaris is really mature in that regard, but it's ahead of Linux.