Wednesday, October 26, 2005

Ooohh... it's a growing mob!

MORE people are pissed. Vive le revolution!

The best things I saw at the IT show...

First, I need to take one company out of the running - Iotum. I have interests there that conflict with a completely impartial analysis. Let's just say that Iotum is really rocking, and I completely agree with the always intelligent Jon Arnold.

So, Iotum aside, here's my top three places:

1) Xelor Software : Xelor makes software that works directly with your existing Cisco infrastucture to automatically configure it to live up to VoIP standards. An excellent management team, with Mark Reid from PictureTel, Steven Guthrie from PingTel and Rob Scott from a host of excellent plces.

From their site : Xelor Software provides the first proactive solution to address the three major challenges facing enterprises with IP telephony: the reliability of phone service, the performance of the converged network, and the quality of each call.

The company’s software has been in development for more than three years. The core research and development was undertaken at the Australian Telecommunications Cooperative Research Centre (ATcrc) in Perth, Western Australia. Leveraging years of experience focusing on QoS issues for ATM networks, the initial engineering team applied the maturing standards for Class of Service for Layer 2 networks and Differentiated Service for Layer 3 networks to the fast-growing convergence of voice and data on a single network.

2) Inveneo.org : Only an altruistic geek could come up with this one... it's a wonderful thing to see. Inveneo is a non-profit company that creates products to be placed in ultra-remote places that enable communications infrastructure. I saw their solution that takes a bicycle, attaches a generator to it, then uses it to power a VoIP system. Now, who does that remind you of?


From their site: A low cost, ruggedized, pedal and solar-powered PC and communications system that provides remote villages access to simple computing, voice calling, e-mail and the Internet.

3) SiVox : The company I mentioned that did the call agent training software. It was really, very cool.

I'm not the only one grumbling at the IT show

Can you believe that it costs me $5.00 an hour to do this? I'm at the Internet Telephony show in LA, and they charge us $5.00 an hour for WiFi access, and then they limit the bandwidth to 64kbps. They are so generous. And I'm not the only pissed off one, either. I say we hack'em.

At the IT show...

Well, there's a marked difference between the Pulver show and the IT show. At the last Pulver show, I noticed a considerable lack of small companies, leading me to believe that it was growing harder to have a small company in the VoIP space. Now that I've been to the IT show, I've found them again. I'll hunt the show floor today and give you a list of the companies I think are exciting, but not well known.

What should be fixed in Asterisk #3

I'm at the Internet Telephony show, and a friend commented that I really didn't like Asterisk. I said that wasn't quite right, because I liked it for being an open source iPBX for Linux. I just didn't like that everyone was morphing it into these very un-PBX like applications. I understand that their options are quite small, and Asterisk will get the basic job done, but it is very square peggish. Which brings me to #3....

Stop spending time working on anything that's not PBX-ish. For instance, I hear about Asterisk being used for LCR and predictive dialing all the time. ACD functions as well. These functions are not really in the realm of PBXs, but of call centers and class 4 switchees. Imagine spending that time making a really, really cool soft client (or management dashboad) for Asterisk. From the looks of it, iaxcomm hasn't been updated in 10 months. How about we all goose that for a while?
Imagine that it will tell you all the states of the people on the phone, give you a visual call directory, click to dial, integrated with presence, Outlook, etc. That would be so hot. We had that back at Netphone in 1996, and it made my life as a manager SO much easier. Want an example of what I'm talking about? Check out the Gizmo project. So very cool....

One really cool application

One of the things I'm afraid of is that I'm full of sh*t. I mean, I'm not pulling a fast one on anybody intentionally, but predicting the future is a tricky business. As a former CTO, and as a whatever-I-am now, I make a lot of statements about how I think the world is, and where it's going, and there's always this voice in the back of my head that says "Are you really sure about that?"

My biggest gut wrenching comes when I talk about next generation voice applications. I've been saying, as have many others, that next generation applications are going to be important, and will drive service providers. The voice inside my head screams "Prove it."

So, here's some proof, and maybe the voices will stop. (Maybe I should just up the dosage on my medication, too.) I met with a company at the Internet Telephony show yesterday called SIVOX (www.sivox.com) that makes a call agent performance suite. It's an application that trains call agents using real time simulations. Their customer (I think it was SBC) has hundreds of scenarios designed to train the people that answer their phones. The one I heard was an irate customer calling into a health care company. The application comes with an agent screen to tell you what to say, like, "I'm sorry to hear that, sir. Could you please give me you social security number, and I'll look it up?" If you don't say the right thing, an integrated speech recognition engine will detect it, and a talking head (called a coach) will tell you what to do. At the end, you get scored.

Now, isn't that a cool application? It's a perfect example, because it relies so heavily on computer/telephone/human interaction. This is the second cool application I've found around the training space. So, I know that, at least for this vertical, next generation applications will be important. So, shut up voices. I said shut up!!!!

Monday, October 24, 2005

What should be fixed in Asterisk #4

How about making SIP the backbone of the application?

Let's talk about the future of telephony, especially in the enterprise. Telephony in the enterprise is all about SIP. Sorry, let's fix that... ALL ABOUT SIP. At this point, it's really not even up for serious discussion. Any enterprise infrastructure will probably live and die on it's ability to play well in a SIP architecture. A quick check on the Net shows some hard clients that support IAX, which I suppose is inevitable. But, the number of SIP endpoints far outweighs the number of IAX endpoints. The functionality in SIP far outweighs IAX. That's not to say that IAX doesn't do firewalls better, because it does. But IAX and SIP are fundamentally different, and SIP is fundamentally better. Any iPBX that isn't very, very SIPPY is going to lose to one that is.

Another example of voice form

I have a friend who is sitting in the office with me who runs a VAR that caters to the HR departments for governments, health care facilities, etc. We were sitting and speaking about what sorts of applications could be done when you integrated next generation applications and HR. Here's one that we came up with:

Imagine voice forms used to do job applications. Instead of filling out a form, you would have your applicants call on the phone. They could speak their name, type in a phone number and answer some yes or no questions. For instance, have you ever worked at a fast food restaurant before? How many years? For Burger King, press 1. For McDonalds, press 2.

Now, that's not bad. But, here's where it goes nutty - add an online quiz. You want to be a waiter in my restaurant? Here's the quiz I'd give...

"A customer asks if a Pinot Noir is a good wine to have with lamb. Is this a good pairing? Press 1 for yes, 2 for no."
"When you are waiting for your customers, do you wait in the kitchen? Press 1 for yes, 2 for no."
"Please press the number of times you've been arrested for DUI. "
"You pressed 0. I'm sorry, but every good waiter has had at least one DUI. I'm sorry, you are unsuitable." (Just joking here... well, not really. It's one of the biggest reasons I was never an excellent waiter. No big drinkers in my family.)

Now, lots of value here for everyone. Quality control on hiring. Value added for the person who installs the system. Indepenent experts who start to create their own tests to determine skill levels.

Friday, October 14, 2005

Real Life Example of Voice Form

Here's an auto-attendantee example of how you could write a form based application:

Imagine you own a heating and air conditioning company, and you want your phones to work like this: When someone calls into your shop, you want the phone to be answered with your company greeting. It says "For faster service, please enter your home telephone number. " If the phone number matches a current customer, find the owner right away on his cell phone, home phone, whatever. If the phone number doesn't, bring it to an operator or to voice mail. With a form based application, that's maybe a few hundred lines of VxML and PHP, a $500.00 computer, a SIP infrastructure.... and that's it.

Here's to irony... I signed up for google ads, and it places adds up there that relate to what I'm writing about. But, everytime I criticize Asterisk, it puts Asterisk ads up there!

Wednesday, October 12, 2005

Voice Forms

So, I feel like I'm gravitating to an open source project. I'm thinking about taking the SIP forum toolkits, and their VXML effort (whereever that's at), and making an open source tool to host "voice forms" and other voice based applications. A voice form is a VXML script that interacts with a caller, and asks for certain information and stores it into database. It's possible to take the results of the voice form and send the caller to another place. The main functionality would be voice forms, but things like Auto Attendants and ACD applications would be supported as well. It would be software only, but would support interfaces to standards based hardware for scalability reasons.

Writing voice forms is pretty easy - very much like writing HTML. One goal of the project would be to allow anybody to deploy a SIP based voip form from scratch in about an hour, only using an off the shelf PC. A number of Voice Form examples would be shipped with the application, and the documentation and tutorials would be as complete as possible. From a business perspective, this would enable any company to quickly integrate voice into existing web infrastructures without requiring custom hardware.

I'll try to put this into better words after work...

Friday, October 07, 2005

OpenPBX.org

Not that I'm at all taking credit for this, but OpenPBX.org has forked the Asterisk code. According to them, the main reason was that they weren't able to get the changes they needed into the mainline of Asterisk. I think there's another reason too - the dual source licensing issue. Either way, asterisk now has an evil twin. I mean OpenPBX.org has an evil twin. Whatever.

Oh the humanity!!!!

I confessed to a friend this week that everytime I started to write about Asterisk, I felt like I was bashing it. He confessed to me that he was afraid sometimes to criticize it as well, because of the overwhelming popularity of the software, and the large number of smart people who liked it. His problem is that he's smart too - and obviously I'm not so smart. Hey, if you don't like it - ask for your money back.

So, I'm starting a top five list of things to fix in Asterisk... Number 5 : Replace that SIP stack! Last time I looked (which is about two minutes ago), there's a file called chan_sip.c, and that file holds the entire SIP stack. Oh my goodness, where do you start with that one? One file? The entire stack? Hello? I hate to be sarcastic here, but I DID hear that it will run faster if you put it all in a file.

If the IETF is producing a draft a week, exactly how does the asterisk community expect to keep up with an isolated implementation? Note to any asterisk developer out there : there's a thing called the SIP Foundry. Look into it - it's a good thing. Now, all of you check me on this... ever have any problems interoperating SIP devices with Asterisk? OK - so do you ever wonder why?

Ok, so if I'm Mr. Smarty Pants, why don't I go and fix this? Well, I just might. But I hope to be fixing a more fundamental thing. Open source is forever. I mean, come back in a long time, and you might just find it still hanging around. If this stuff is going to be with us for a long, long time... shouldn't we working on the architecture issues? Setting this stuff up so that we don't build a pile of code? I think we should, and my contribution for today is to look at the big picture and question it.

God d*mn - I do bash it every time I write. Mr. Spencer - it's nothing personal. I swear.

Monday, October 03, 2005

Saturday, October 01, 2005

Session Border Controllers

Lately, I've been thinking and writing (for myself, mostly, but I'm going to send it off to the SIP Forum as well) about session border controllers in a pure SIP network. Here's what I've been able to come up with, and I wanted to share it with you:

1) As long as there are private interests that use the Internet, they will have their own networks they attach to the Internet, and will therefore create a boundary. These boundaries will most likely contain sessions, and they will require control. Therefore, somewhere, there will be seesion border controllers.

2) If pressed to boil them down to their essence, SBCs would perform authentication, encryption, media traffic control and execute policy decisions.

3) Since policy decisions will most likely be complicated, and might tightly integrate with many third party technologies and programs, I think it will be essentially software based. That is, I don't expect this function of SBCs to live in firewalls any time soon.

As I look at Asterisk and SER today, I don't really see either one being excellent at SBC functionality. Here's one hole in the open source world.