Satellite Internet Service for Macs?

Untimely Ripp'd asks: "Satellite broadband has been available to PC users for half a decade, and still is not trivially available to Mac users. It can be done, but it's always an unsupported hack, or it requires buying expensive extra hardware and software. I cannot understand why Hughes and the other providers would refuse to spend the relatively few dollars necessary to develop a couple of device drivers and glue libraries. Time after time, the vendors have said, 'it's coming,' but it never does, and the promise eventually goes away. (Earthlink's FAQ page no longer says that Mac software is being developed, for example). I'm not gung-ho on conspiracy theories, but the only explanation I can figure is that they're either being paid or bullied. Does anyone know of any serious tech hurdle that would make it cost more than $100K or so to develop the necessary software?" this article mentions one-way Mac service coming online from OWC in a future expansion, along with nationwide service. A comment from that story does mention a simple solution, but why is it that Satellite service, even one way satellite service, depends on Windows-only software? What other solutions have Mac users resorted to when they wanted their Macs connected?

Comments

Again?

[Clue4All]

I'm not gung-ho on conspiracy theories, but the only explanation I can figure is that they're either being paid or bullied.

How many times can we go over this same point? It's the same for Linux and Mac, it's just not economically viable to develop software for something used by less than 5% of the computing masses. It doesn't pay, plain and simple, and companies aren't going to waste money developing with little to no returns. I await next week's Ask Slashdot with the same question.

Re:Again?

[NineNine]

Actually, there are more costs than just development... They'd have to spend money on support for the product, money on the marketing, etc. Rolling out a product for a different platform isn't as simple as paying some guy to write the code. There's a lot more that goes into it. So I guess I'm re-enforcing the parent here. It's just not economical, but even more so than saying "Well, it only costs $100K for development".

Re:Again?

[Otter]

Also, while Mac users may be ~5% of the total, they're a vanishingly small percentage of the construction supervisors, petrologists and logging bosses who need a satellite link. This troll is essentially correct, even if he can't spell Shih-Tzu.

Re:Again?

[GlassHeart]

It's the same for Linux and Mac, it's just not economically viable to develop software for something used by less than 5%

One way to make money is to build a product that 95% of the people use. Another way to make money is to build a product that 5% of the people use. Microsoft certainly made a boatload of money, but Apple is not exactly bankrupt. In fact, I'd expect that Mac sales of Adobe Photoshop account for significantly more than 5%.

In practice, a lot of times you'll find that the reason a minority OS is not supported is not because somebody determined that it was not viable, but that nobody ever bothered to see if it was viable or not. Only the former is a good business decision.

Re:Again?

[Anonymous Coward]

This comment is pretty inaccurate. I'm a systems engineer at a major satellite networking provider not unrelated to this story. I'm not working on the current consumer offerings, but on the next-generation system. Of course, I don't speak for said employer.

The DirecWay satellite is pretty much paid for - sure it's a large capital investment up front, but the software side is a constant ongoing expense. The software is constantly under revision by a largeish team of software developers and network engineers. New features are still being added. That's a lot of salaries to pay, and it doesn't stop. Software is never finished. Adding support for another operating system, at least with the current (legacy?) satellite networks, is a large undertaking. All the spoofing for the consumer products is done by software, and it fools with the TCP/IP stack directly. It also needs to be supported. There's at least 10 programmer salaries and about 50 customer service people, just to have a shot at the .1% of the 5% of computer users who have macs and might want satellite internet. If you look at the news, you'll see that the consumer side of the business isn't doing so well anyway - it's a very small take-up rate, so even that .1% estimate is extremely generous.

Are you a shareholder of Hughes? If you were, would you want them spending 3+ million dollars a year on software that _might_ get 5000 subscribers at under $70 a month? It's hard enough making a profit with 100,000 subscribers on the PC platform. Don't forget that there are a whole host of other costs - installer training, advertising (subscriber acquisition costs are huge), etc.

So my position isn't one where I would know what anyone's future plans are - hell, they could be releasing Mac software tomorrow and I wouldn't know. But you're seriously underestimating the costs of software development and support for adding an OS.

Re:Again?

[swdunlop]

Having done a stint or two of technical support, a tour in quality assurance, I can point out several costs of supporting less-used OS's and hardware platforms.

First, it's not quite so easy to provide technical support for more than one environment. Like it or not, it isn't just a case of hitting Command-C instead of Control-C; Windows users expect one set of UI conventions, Mac users expect another, and Unix users want it three different ways, and it should also work from the command line. Each OS requires different tactics to work around existing issues. For technical support departments supporting diverse platforms is a nightmare, since it means they have to either spend most of their time in training, or provide multiple specialized departments, which can easily eat up manpower.

Setting rigid support boundaries is a partial solution, but I have yet to have a customer who knew of the support boundaries and had a problem outside those nicely defined lines who didn't try to cajole and/or threaten me into helping him "just this once." It wastes my time, and his.

For developers, this is similarly a nightmare. I know of one decent cross-platform GUI toolkit that works on X11, Win32 and Mac OS, and on the Mac, it doesn't conform strictly to the Apple User Interface guidelines. Compound that with the highly specialized skillset required to write drivers for a particular operating system, with some exceptions, and to go with your bloated support department, you'll have an increased software development budget, and QA budget.

The only recourse we fringe users have is open standards; encourage your vendors to use published protocols, open standards and to document their API's. Take those documents and write a solution for your environment of choice.

Re:Again?

[Graymalkin]

Your arguement makes perfect sense and I have to say I agree. However I think the problem on the developer side doesn't lie with the fact supporting Macs would only get you .1% of 5% of the market but instead with the fact Hughes/EchoStar/everyone else screwed up originally in their implementation. The soon to be released DW4020 (maybe it is released already) is how the sat providers SHOULD have originally rolled out their connection equipment.

Using USB based equipment in my ever so humble opinion was a major screwup on their part. Besides the obscene price tag of the equipment and installation (far more than my DirecTV system cost me) the USB satellite modem was a huge turn off. It is far more inefficient to fsck with the Windows networking set up than to do it in your own little box in a self contained manner. A satellite modem with an ethernet port is a much better idea, yet again in my opinion. For starters support is rather trivial, instead of needing to rely on Windows to work properly which is a lot to ask, they only need to really maintain their own software. All the network stack customizations and proxying tricks to let the network run on a high latency connection would be relatively simple to maintain on something like VxWorks or some other embedded system. All the end user would need is an ethernet port which in available on a huge percentage of systems, including every Mac made since the iMac.

At the time I was looking at DirectWay about when I was looking for boradband and was picking up a DirecTV system anyways the only satellite modem options were USB. Had I been able to plug it into my Ethernet hub I probably would have bought the service. For a long time I lived out of reach of both cable and DSL and my telephone line choked data down at a staggering 24kbit/s. Now I have a cable modem plugged into my router which is plugged into my hub. I think there's plenty of Linux/Mac/Whatever users who also would have signed up for their service a long time ago and thus been locked into satellite instead of opting for cable or DSL. I think Hughes dropped the ball with DirectWay, it had a major opening even in metropolitan areas before the massive cable internet rollouts of the past two years. Not only could they have likely increased their customer base but they could have also lowered their costs by not relying on Windows hacks in order to get their systems to work right.

Why would Mac users need fast internet anyway?

[realmolo]

I thought they were all too busy making Quicktime movies of their trip to the Volvo dealer, and ripping Yanni CDs to play on their iPod while they wait at the coffee shop for their Shitsu to get it's nails manicured, sipping latte's. You know, Thinking Differently.

It's not about the cost to *develop* the software

[mbessey]

The main barrier to providing a service like this for the Mac, or for Linux, Solaris, BeOS, or whatever, isn't the cost to develop the software, it's the cost of supporting users on another platform.

Every time someone calls with a question, it costs the company money. The quicker you can answer their question and get them off the phone, the better. This means minimizing the number of different systems your support folks have to be trained for.

-Mark

I tried to like get the satellite to work.....like

[Typingsux]

on my Mac.... Then all of a sudden beep beep beep beep beep beep
Unhhhhh?

Why Satellite?

[MidnightBrewer]

I'm living in Osaka, Japan right now, and the biggest hurdle the technology-savvy Japanese have to face in the telecommunications field is geography: Japan is 70% mountains.

Their solution? Wireless internet. Give your user a wireless internet card, then connect the receiver to a fiber-optic network offering 100Mbps. Works with Windows as well as OS 9 and OS X.

Currently, ADSL alone in Japan offers 12Mbps, for a slightly cheaper price than in the states.

Re:Earthlink Satellite is USB only: why?

[chill]

Most computers come with USB ports. Most home computers do NOT come with ethernet jacks as standard.

Most satellite Internet providers use a form of header-rewrite on the packets, routing everything back to your PC through their NOC. It makes for difficulties in setting up direct-to-router connections.

I think the answer is pretty obvious

[strictnein]

It is market share plain and simple.

First off, you have the very small market share that Mac users represent. About 1-3% (right?)

Then you have to factor in what percentage of the Mac users live in or near major metropolitan areas. I would argue that number is probably near 90% of the total Mac users (a number I am pulling out of my ass, but I just don't think there are a ton of Mac users in the rural US, which is where the Sat. companies are focusing).

So, you have a possible market that makes up maybe .1 - .3% of the total computer users in the US.
Hell, even if my numbers are a little bit off, the total market share for Mac users in rural areas can't be more than 1% of all internet users.

So, if the software costs $100k to write, and then another $5k - $10k / year (/month?) to support, plus retraining all (or many) of your support/install personel to use the Macs, is it really worth it?

I personally don't think I would do it if I ran the company.

But whatever... hmmm... the linux router seems to having problems... wonder if throwing it out the window will solve it

Technical issues

[rochlin]

I think it's worth mentioning a couple technical points that have been overlooked by people considering using a PC as a bridge between the satellite receiver and the Mac.

I set up a PC system (with win2k) purely as a bridge. The satellite reciever uses USB (this is from Pegasus-DirecPC-Hughes - now acquired by Earthlink). Pegasus and DirecPC provide proprietary Windows only drivers to deal with the USB network-satellite connection. That's issue #1. That means using Linux or (as suggested in the "comment" link in the oringal post) a simple router won't fly.

Issue 2: Optimizing the Window size for the ethernet connections --
The fact is, the TCP/IP conneciton to the satellite (high bandwidth - extremely high latency) needs different rwin settings to optimize the connection than the simple pc->mac LAN connection. So far as I can figure, Windows lets you choose one setting for all NICs (in this case the USB satellite connection is a NIC).

Issue 3 - you need some kind of 3rd party NAT/Bridge software like Sygate to share the connection with the Mac. The built in (to Win2k, 98) Internet connection software won't work because it can't bridge different subnets. The USB conneciton is on a different subnet vs. a regular NIC. I don't think it can be configured otherwise. WinXP might fix that.

Bottom line: You need a PC with Windows to share the satellite with the Mac and even then the Mac will have inferior service vs. the directly connected PC. So a satellite service supporting Macs would be nice :)

As a junk hardware collector...

[NoMoreNicksLeft]

I have both the DirecPC pci card, and the usb modem version. I'm not capable of reverse engineering these myself, but anyone that is, is welcome to mine. I could probably even spare an 18" dish+LNB.

I mean, every time we wait for these fuckers, we end up losing. Maybe you need to decide to write it yourself? It's the only way to be sure it's done right.

PS Anyone that knows the pinout for the power on the DirecPC usb modem (mini-din 8), could you send it to me? I know it's gotta have 14v for the lnb power, in addition to 5v, but last time I tried to deduce this from looking at the pcb, it took me a day and I still fried the device.

I'm not sure I agree...

[tkrotchko]

2) I think Window size is settable by device. Otherwise, there's no way you could route between say, a token ring card and an ethernet card (something that I'm certain can be done).

3) You're probably right except that home networks don't have different subnets. Or I should say, there's no good reason to have multiple subnets.

I think the primary problem with 2-way satellite service is that latency is so high that for the common things home users do (open up their home page of http://www.msn.com) its likely to be no faster than a dial-up connection.

Your answer: Real soon.

[Guppy06]

According to DirecWay's FAQ bot thingie, they will be selling the DW4020 to consumers "fall of 2002" (read: "any day now").

The DW4020 is pretty much the standard DW4000 satellite modem boxen they currently sell, except it includes a third boxen that eliminates the need for a USB connection and presents 4 Fast Ethernet ports. Supposedly you'll also be able to buy this box separately to upgrade your existing DW4000.

Now the only question is when EarthLink will lower their monthly satellite service fees to match DirecWay through DirecTV. I just dropped BellSouth in favor of EarthLink this past June and I'm not interested in changing ISPs yet again so soon.

Re:well maybe

[interstellar_donkey]

You'll probably be modded down for saying that, but your right. I long ago conceded the fact that Macs were at the very least as good as PCs (in some areas), but Mac users are a whole different story.

I worked the phones in tech support for quite some time for a major ISP. I hated having to get mac calls. Not because they were difficult to troubleshoot; macs are surprisingly easy to fix when there are network issues, and I always liked that.

But the users, as a general rule, were much much worse then PC users. The worst were the imac users. I have dozens of theories about why this is, but the only one I can come up with is mac users simply don't invest the time needed to really understand their own computers, or at least the time needed to properly opperate a PC, but instead just want everything to work right. When it doesn't it's the fault of whoever is on the other end of the phone.

This is not a bash of macs. Macs are so easy to troubleshoot, the computer literate mac user rarely needs to call tech support, so the support folks just get the worst of the worst.

But it IS economically viable to ...

[Ungrounded Lightning]

How many times can we go over this same point? It's the same for Linux and Mac, it's just not economically viable to develop software for something used by less than 5% of the computing masses.

Let's phrase this another way:

How much would a company pay for ADVERTISING to get a 5% increase in sales? (And thus a MUCH greater than 5% improvement in profits, since the development is already amortized.)

Now if that same amount bought you the development of an incremental feature (i.e. a Linux or Mac driver) that enables another 5% of the market to use your product, it's the same case. (Actually, if you're currently addressing 90% of the potential market and the new segment is an incremental 5% you're adding 5%/90% or about 5.6%).

But wait, it's better ...

Suppose that you're currently splitting the market evenly with one other competitor. If YOU do it and HE doesn't, that 5.55% about doubles to 11.1%. With an even split among three competitors the first mover gets about a 16.7% bump in potential sales (and more in profit), and so on.

With something like networking you have a small number of competitors but MAJOR lock-in. First mover gets the prize and KEEPS it. With something (like a device) with more competitors and less lock-in you may not keep it, but you get a BIG boost until your competition wises up.

But WAIT! You don't HAVE to develop it yourself! Publish enough of the interoperability specs and - at least for Linux - SOMEONE ELSE will do it FOR you! You get the benefits and do only a tiny fraction of the work.

Your work consists mostly editing your internal documents into an externally-releasable one that will enable a developer without giving away your trade-secret farm. But don't get too paranoid: Your competitors are ALREADY reverse-engineering you. You should have your critrical IP already locked up in patent-pending, which will keep your competition at bay if you publish more than you intended. Meanwhile, better specs mean better and sooner community software to enable your sales.

Network operators might have some issues with security - but that's already been addressed elsewhere. (Bottom line is that the black hats will get you anyhow if you're already BADLY broken, regardless of whether you publish, while if you're reasonably secure (i.e. only a little flakey) the exposure will get the white hats on your side and you'll probably increase your lead in the arms race.)

Don't forget path packet size discovery...

[raehl]

Or whatever the hell you really call it.

The satelite drivers are written to keep the high-bandwidth pipe full (i.e. you can put a lot of data in the air before it gets to the satalite and back.) Since various networks that carry TCP/IP (Etehrnet, ATM, etc) are based on different optimal packet sizes, so oyu generally probe your connection to figure out the what link is going to split your packets into the smallest size and then just send packets of that size.

I'd suspect that's why the Mac on a PC performs a lot worse than the PC in general - the ethernet packets that get sent to the PC probably get passed on as-is instead of reassembled into larger packets for the satalite link.

One huge packet with one header is obviously more efficient than one huge packet made up of lotsa smaller packets each with their own header.

Need custom stack, do it in hardware/firmware!

[aquarian]

A couple of people have posted saying a custom TCP/IP stack is needed. Well, then, do it in the sattelite receiver, in the hardware and firmware! Proprietary secrets would be safe within the box, and the damned thing would actually work! Plug and play- no drivers, no tech support calls. Give us our ethernet, dammit!

Betcha this would be cheaper than creating and supporting software, too. They have to make the receiver/modem box anyway- so stick a router chip in there, and be done with it.

We can provide Mac support *if* demand is there.

[schnell]

The company I work for, Spacenet, is the second-largest business satellite ISP out there. We serve *business* customers who have large and small multiple (5-5000+) locations (retailers, food service, energy, financial, services, etc.).

If you have a *business* meeting these criteria and are looking for satellite connectivity that supports Macs, send e-mail to me and I can push for Mac compatibility if there is significant demand.

Don't just say there is no corporate satellite ISP support for Macs and do nothing about it ... if you can genuinely justify large multi-site Mac satellite network support, I can help make it happen.

As a BSD guru-turned-Mac-guru myself, I would love to help this but I do need the numbers to prove it. Right now, we have almost zero requests for this, but an influx of REAL potential customers asking for this could make it happen. I would really, really like to make this available, but I can't do it by telling our MS-oriented development guys to do it without visible justification.

This isn't some random spam for business, this is a real request from a company's senior marketing staff to help build demand and make this happen. In your e-mail, please describe your multi-site business and its needs, and I can use this info to get Mac support for Spacenet's services.

Simple

[Perdo]

Buy an old x86 box, install windows 98 SE or better and use internet connection shareing. Shouldn't cost more than $50 for a crap box with windows still on it.

Power consumption will become a cost issue so you might try a mini-itx box, which will only suck about 25 watts, but up front cost will be higher, on the order of $200 for a complete system.

You paid $2500 for your mac and around $500 for the satellite install what is another 200 bucks?

If you want the service, you have to solve the problem yourself. The bonus is, you can have as many computers using the connection as you want without paying the satellite company's per seat fee.

The only problem with satellite besides the cost is latency worse than a phone modem. your signal has to travel at least 46,000 miles round trip to a geosynch bird over the equator from the southernmost parts of the US. That's a 500ms ping time minimum.

So, running a mac on satellite is no problem technically if you consider an extra 200 bucks for installation fees independent of what your provider is charging.

The only time you will really notice the 500ms lag is in a game and, well, you are using a mac so that shouldn't be a problem either.

That exactly what my company does

[Srin Tuar]

It more of a commercial service than a residential one, but we do exactly what you describe: the custom TCP acceleration is in the embedded box itself. (Although the real heavy lifting is done by a linux box on the ISP side)

We test with mostly Linux and Win2k, but apples should work fine over regular ethernet. Ne special software is needed to run a client site. Just plug and play (It also does DHCP and DNS-caching)

Look at http://idirect.net/

Its a pretty good programming gig, I get to work with gcc/cvs/all my favorites.

jmaiorana at idirect.net