Tuesday, September 27, 2005

Even more about asterisk...

A friend pinged me this morning and asked why I was so negative on Asterisk. It's a good question, and let me make one thing clear - I'm not all negative about it. I think that Asterisk is a wonderful thing in many ways. It's been the first and most succesful open source effort in IP telephony. I am a firm believer in the open source movement. It's radically lowered the entry cost for all IP telephony. It's an abassador for VoIP.

More than that, I want to make it PERFECTLY clear that I have nothing against Mark Spencer. In the few times which we've spoken, he's been nothing but gracious and modest. I'm happy for his success. He's a very smart guy. I wish him all the money and Chateau Neuf Du Pape in the world.

So, why the constant diatribe? Simple. I think that open source software is for eternity. I can imagine a time in a thousand years where people use the software we write today. And, if you don't believe me, just look at the COBOL market. It still exists, and there is every reason why it shouldn't. Now, if we are about the business of writing long term software, I think we need to look carefully at what we have, and think deeply about the path it should take. Software engineering teaches us that the choices we make early in the process have very large effects downstream. Take this to the logical consequence... if we are writing software for a 100 years, shouldn't we really deeply think about the architecture today?

Monday, September 26, 2005

My problem with Asterisk

Well, thanks to the History channel, I have my answer about why Asterisk bothers me so much.

I was watching this show on Sunday morning (yes, before church, Mom), and they had a show on about a spy. As part of the show, they mentioned this German airplane that was a "fighter bomber specially modified for reconnaissance". That made my stomach twist up in a knot. The plane was ugly, was probably too slow to be a really good fighter, and couldn't carry enough weight to be a decent bomber, and where are they going to put all that reconnaissance equipment? And these guys are German! They should WAY know better than that. Maybe that was one of the times when all the good German engineers had fled or were bringing the place down from the inside.

And that's my basic issue with Asterisk. It is a project that aims to do everything. And, will someone tell me just how you optimize for everything? In the real world, you either have something that's strong in one area, and weak in the others, or just a very mediocre but rounded effort. Now, I understand that, next to the SER proxy, it's the only strong game in town. I've always been a PingTel fan, but anyone at the VON show would see that Asterisk is ruling the show floor these days.

Well, you know, history might not repeat itself, but it sure as hell rhymes.

Friday, September 23, 2005

Go, maddog, go!

Brazenly stolen from a Network World article... Mad Dog tells us exactly why open source will be succesful:

"You must be telecommunications people who expect your software to work," said speaker John "maddog" Hall, executive director of Linux International. His riposte came after he received a smaller-than-expected response to his request for a show of hands of how many people in the audience had sent bug reports to software companies and then actually had the bugs fixed.

Some 850 million people globally use software and 250 million of them want to change it in some way, Hall said. "I know because they've told me - all 250 million of them have told me." Of that number, 250 million actually have time to make a change, while 65 million have the expertise. In the end, 33 million software users will actually change software code to suit their needs, he said.

Hall also took a swipe at Microsoft. He noted that for a company in China that brings in $1,000 annually in revenue buying Microsoft Office is the equivalent of paying $40,000 in the U.S. "If I had to pay $40,000 for Microsoft Office, I'd put a black patch over my eye and go, 'Arr, arr, arr! I'm a software pirate!' "

Wednesday, September 21, 2005

VON Show... Continued

The real problem with the VON show is the speed of it all. It really moves super-fast, and by the end of the day, you don't even remember the beginning of it. I think it was a pretty good show, but it seems like it is now dominated by the larger companies selling larger products. I remember the time when many of the companies were small - we used to play a game counting the number of new phones from Asia. Now, every company has a big product selling into a big market.

I spoke today about SBCs in a pure sip network... here's the presentation.

Monday, September 19, 2005

Mark Spencer isn't Stupid

I suppose anyone who knows me knows I'm not a big fan of Asterisk. I'm just not a fan of any software, no... make that any THING, that purports to be all things to all people. Maybe that's because I find it hard enough to be something to somebody. That aside, Mr. Spencer taught me some things during his talk today - I wanted to pass it along.

During his talk, he mentioned that open source was right for telecom. First, it's a very large market, and there are plenty of opportunities for players of every size. Second, it's a very technical direct market, where most of the direct consumers of the technology were tech savy. There's a very large incentive for open source software, and can save service providers lots of money. There's an extreme need for customization, since differentiation is critical for carriers, and open source really shines here. And finally, the barriers to entry for open source don't exist as they do on the desktop.

Mark's dead-on with this. I don't think it answers how you can do projects that require very large team efforts to accomplish, but that's OK. It does give a great answer for the smaller projects.

Shoot - I might even start to like Asterisk someday. Until then, I want my money back. Oh yeah - it's open source, so I can't ask for my money back.

First day at the Pulver show

Well, like the change of seasons, the VON show is now upon us, and I spent most of the day at the Open Source summit fishing for new ideas. Basically, I think I confirmed in my mind that the vast majority of success in open source has been in the Enterprise and new carrier market. I asked the panel if they thought that large carriers and open source was necessarily a bad combination. They remarked that they thought large carriers would, and in fact were, deploying open source based systems now. I wonder if they are deploying these solutions at the edge of their networks, or in some sort of trial. I can't imagine that they are replacing core equipment with it. I noticed that even the smaller service providers (that being Telio) had deployed open source solutions that were glued together with their own custom code. In other words, even the smaller carriers weren't taking single open source efforts off the shelf. They did say, and I believe it, that the SER proxy was the best of the available carrier class open source offerings. I do think that it's curious that every time they are asked about carrier class, they speak about scalability, without addressing other carrier concerns like fault tolerant hardware platforms, FCAPs or regulatory concerns such as CALEA. I'm not used to being a naysayer, but I wouldn't be worried about open source taking over Nortel's business, at least this year.

Friday, September 16, 2005

Oh - I'm public now

I suppose the cat's out of the bag now... I'm officially resigned from Versatel Networks. Here's the letter I've sent to family and friends...

Dear friends, family and co-workers,

It is with great regret that I've handed in my letter of resignation as CTO of Versatel Networks effective at the end of September. A recent family event has made extensive travel impossible for me, and because of this, I can no longer serve Versatel in this capacity. As I am sure you know, I make my home on Cape Cod, and never thought I would retire in Canada - but I honestly hoped my stay would be longer. Luckily for me, and with the very generous support of the Versatel executive team, I will continue my association with Versatel as a consultant for some time to come. I have personally worked with over fifty companies in VoIP in the last ten years, and all of them have taught me something. Versatel, bar none, taught me how to treat customers right. They are truly world class in this regard, and I remain a student of their excellence.

I am going back to consulting, and will probably focus on open source development and consulting, writing and teaching. As this is a fairly challenging personal time for me, I would ask for your good thoughts and wishes, because I feel like I'm going need all of them.


Open source directions

I think I've come to a personal conclusion on where I'm going to spend my open source efforts : enterprises and smaller service providers. Coming from Versatel, I can see how even a company as excellent as they are have an uphill climb against the much larger Tier 1 vendors who supply to the carriers. Traditionally, the enterprise was all about channel, and if you didn't own the channel, you couldn't win the war. My bet is that the channel is now the Internet itself, and the days of the stupid phone installer are numbered. Instead, we are coming to the days of the stupid system administrator, and he goes to the Internet first.

In the very long term, the large service providers may simply be the ones that supply the best pipes, as sustainable differentiation becomes harder to maintain. It will be up to the smaller providers to come up with the niche applications that are compelling and add value. Adding to that, I know that the PBX is not a function that's going away anytime soon, and I wonder if any PBX vendor can win a long term battle against Chinese phones and asterisk.

That said, I think I'm staying with the smaller service providers and enterprises.

Tuesday, September 13, 2005


Sounds like a start on an open source applications platform for VoIP... from their site.

Shtoom is a open-source, cross-platform VoIP softphone, implemented in Python. As well as the basic phone, the package also includes a number of other applications -

shtoomphone - the end-user phone
doug - a framework for writing server-side voip applications.
There's a number of applications implemented using doug that ship with shtoom.
shtam - a simple answering machine/voicemail application
shmessage - an announcement server
shtoomcu - a simple conferencing server
Shtoom works on Windows, Linux/Unix and Mac OS X. The phone has a number of user interfaces - Qt/KDE, Gtk/GNOME, wxWidgets, Cocoa (OS X), Tk and a command line.

The phone has audio support on Linux using OSS or ALSA, on OS X using native CoreAudio, and on all other platforms using the PortAudio library. A native sound drivers for Windows has been implemented as part of Pythonwin, and needs to be hooked up.

Shtoom requires Python 2.3 or later, and Twisted 1.3 or later.

Monday, September 12, 2005

And maybe there's....

monitoring software for VoIP. Measure QOS, traffic patterns, unauthorized intrusion? No real excellent CALEA story out there, never mind an open source effort around that. It'll be important for nearly eveyone, and could really improve the quality of VoIP for every service provider, and ensure security for enterprises....
I know that there are some good hacking tools, like SNORT, SNIFF and Ethereal that might be used as a start for a project.
Think, think, think....

Thursday, September 08, 2005

An interesting thing happened today... the comcast installer came to my house to install a new service (I just moved). And I realized that this whole VoIP thing was really happening, if there was a guy in my small town who's job was VoIP. I mean, someone else other than me.

One thing that's really hard for me, and probably anyone, is to time what's going to be next. I am thinking that we've got the whole world of IP services availble now for the picking. Any service that involves voice can now be pretty easily deployed into the network, and anyone can subscribe to it. The problem is that it still costs some pretty good money to deploy these services. Even at Versatel Networks (which makes a very excellent product, by the way), most customers are looking at tens of thousands to purchase the platform, and off they go.

I wonder what sort of impact would be made by a generalized service platform for VoIP, based on software and developed through open source. It would have an API that developers could write to, and handle prompt playing, recording, some mixing. It would be a SIP only device - no hardware. You write your application to the API, host it on the computer. Maybe make it really scalable, maybe make it fault tolerant....

Thursday, September 01, 2005

The search for an open source project...

Well, I'm now deep in the search for an open source project. Here's what I'm thinking about - VoIP Peering. A VoIP Peering and Interconnect BOF was held last month at the IETF meeting, and from what I hear, there was real involvement from the carriers. It's pretty clear (I think) that, in the long term, carriers MUST use IP peering when available because of cost reasons. Never mind what this means for the large amount of IP services that will follow afterwards.

This sort of goes with something else I'm thinking about - the principal of end to end transparency in the Internet. One knock against session border controllers is that they break the transparency rule in the Internet.

So, given that peering will be done over IP, but today's SBCs break transparency rules, what I'm thinking about is leading an open source effort to make an Open SBC. This Open SBC will follow a route more friendly to the IETF, and concentrate on executing policy between networks, establish authentication and encryption, maybe help with media routing and try to find some way of doing all of this without breaking the transparency rules. Or at least, as few as possible.

I feel like Pooh bear walking around, scratching his head saying "think, think".