Open Source Communications War

Behold the open source communications wars. In this corner weighing in at millions of downloads is Digium… A Huntsville, Alabama purveyor of all things Asterisk… From training to support to hardware, you name it.

In the other corner with decades of international experience and engineering and a strong alliance with IBM/Microsoft is Nortel/PingTel.

Last week we saw comments from Nortel on why they acquired PingTel and why this solution is better than the “old Asterisk”.

One would imagine this comment was the straw that broke the camel’s back:

By the way, these strategies dictated why sipXecs was chosen over Asterisk, a clearly inferior open source environment.

Today, Digium’s Bill Miller responds…

Blog Response for The Hyperconnected Enterprise Post

 

Presented with the disruptive threat of open source Asterisk, and the recent momentum seen in commercial channels and the enterprise, Nortel acquired sipXecs. Can you buy your way into open source credibility? Nortel’s not the first old-line company to try. Is open source a marketing bullet point for Nortel ? Its certainly not inherent behavior that’s woven into the fabric of the company!

 

Hey, I can’t blame them — if you were Nortel, and saw Digium and Asterisk on one side and Microsoft on the other, threatening to eat into your core business, what would you do?

 

So: Bravo, Nortel!  Welcome to the next generation of telephony. But you’ll need to learn the strengths and limitations of what you just bought. As we have learned from our commercial customers as well as the countless numbers of Open Source Asterisk installations, SIP is not the entirety of UC. True, Asterisk isn’t a SIP proxy — because a SIP proxy alone cannot provide services the world has come to expect from phones.

 

To pick a specific example of the rather misleading comments in your article: It’s incorrect when you claim that all Asterisk calls go through a centralized system. We assume you’ve just been misinformed, but your claim that Asterisk is designed to always handle media streams is just incorrect. You should recognize that the Asterisk rhetoric from the sipXecs and FreeSwitch teams refers to Asterisk over 4 years and several versions ago when they last looked at the feature set. In some cases, the Pingtel/sipXecs team in particular likes to compare against specific SMB packaged systems and not native Asterisk, which would provide a more “same page” comparison. The SCS500 scales to 500 users. It certainly does not need the power already in Asterisk today.

 
 

Let’s talk real numbers. Asterisk community members have quickly created testbed platforms which process  300 SIP calls per second capacity. When RTP is managed by an Asterisk instance, we have demonstrated as many as 1900 concurrent  G.711 channels on inexpensive off-the-shelf hardware. The open source versions of Asterisk were designed to handle hundreds of thousands (yes, hundreds of thousands) of end users in different circumstances, spread across multiple machines with functional role distribution in a way completely unlike a PBX.  Of course, smaller SMB solutions are selling well for our Switchvox line as well as a dozen or more other companies who repackage Asterisk with limited hardware or license caps – that is their business.   But please be a bit more forthcoming in your comparisons – Asterisk is being run in systems spanning huge user PBX and service delivery populations.

 

For sipXecs to compare themselves against the pre-packaged SMB offerings of Asterisk in different flavors is misleading. That’s like saying the Apache web server on my embedded-processor webcam can only handle 2 simultaneous streams, therefore Apache doesn’t scale on any platform. So while Asterisk can serve as a PBX, it can (and for countless customers and projects already does) also serve as an incredibly flexible service-delivery platform for UC services, custom application development, or carrier VoIP integration. It can be used on even the smallest embedded platforms (Linksys WRT54G, AA50) or on the largest voice server farms such as Integrics’ Enswitch which directly supports over 100,000 end points and over 6,000 concurrent calls just for one of their 40+ customers and integrates easily through Asterisk APIs to multiple billing solutions.

 

As the “wildcard” name suggests, Asterisk works with numerous protocols and codecs. When required for communication between disparate endpoints, Asterisk will intelligently negotiate and transcode between them. If the two are compatible, Asterisk will hand off the call and get out of the way. Asterisk has the flexibility of handling media if desired, but RTP between endpoints is the preferred design for larger systems, including video, which Asterisk has handled for several years now.

 

In contrast, the sipXecs architecture enforces the requirement that all endpoints be uniform, which pulls along all sorts of ugly forklift-upgrade requirements for businesses looking to grow into the future, or uses expensive media gateways to do what Asterisk can do in software. We can confidently say that Asterisk does UC and that sipXecs is simply a SIP platform that requires lots of other moving parts to get the job done. Don’t take our word for it–read their comparison posts, which say sipXecs needs other components to complete their system. Asterisk handles the “U” in UC  (as well as the “C”) and has for some time now.

 

The only claim that seems to be correct in Tony Rybczynski’s post is that for large Asterisk installations, there is no comprehensive management interface for all possible aspects of the system.  There are multiple web-based interfaces available for small/medium enterprise PBX-style installations – FreePBX is the most popular open source tool, and Digium’s Switchvox being an excellent representative of a commercial packaging. Both of these examples include automatic phone configuration and provisioning.  However, Digium has found that large enterprise developers who wish to harness the true power of the system typically want to have call control at a much more fundamental level than what a GUI typically offers or what a vendor might consider “simplified.” Therefore, Asterisk is available as a telephony toolkit – a suite of programs and fundamental tools that allows a developer to quickly deploy new voice apps or extend existing legacy platforms if they so choose – it is as flexible as the circumstances require.

 

One final observation counterpoint to a comment you made: As the progenitors of the venerable DMS-100, Nortel should know by now that the age of code — or lines of source — mean nothing when compared against other software. Do more lines of code indicate more features or quality? Do fewer lines evidence efficient, bug-free code? Lines of code are typically irrelevant in doing anything other than measuring platforms against themselves over time or measuring individual coding productivity.

 

It’s going to take more than this acquisition to, as Tony says, solidify Nortel’s “leadership in the global open source ecosystem.” We hope that this purchase creates a more viable and useful application that can
be used by the open source community – we hope this isn’t a repeat of the Vovida(Vocal)/Cisco purchase and subsequent smothering. But f
or the sake of the open source telephony movement that Digium started, Nortel, we welcome you to the open source revolution.

  • Anthony Minessale
    August 18, 2008 at 6:30 pm

    Pointing to anyone else claiming what they say is rhetoric is somewhat amusing considering the content of the posting which is what many call “The pot calling the kettle black”. 😉

  • Frank
    August 19, 2008 at 10:57 pm

    Actually, I am glad to see this kind of dialog going on. I would be very interested to know if anyone (preferably a 3rd party with no affiliation to any particular vendor/product) has done any benchmarking of the various open source telephony packages such as Asterisk and its derivatives (e.g., Trixbox, Elastix, etc.) vs. sipXecs vs. FreeSWITCH.
    I must admit I, too, was caught up in all the rhetoric. As someone who started playing with Asterisk via Trixbox more than a year ago, bought the O’Reilly book when the 2nd edition came out, and made it past that first weekend of toying, when I read about FreeSWITCH’s progress after it’s recent release at v1.0, I was floored by what some folks were saying regarding its performance vs. Asterisk on the same hardware. (The numbers that stuck out for me were 250 simultaneous calls with Asterisk vs. ~3000 with FreeSWITCH on the same hardware.)
    And no, I did not confirm this by running my own benchmarks (though that would be a great thing if it existed… an automated testing/benchmarking tool so folks could compare the various packages themselves, preferably doing more than just setting up calls but generating pseudo-realistic voice traffic on all calls). So I began reading up on FreeSWITCH out of curiousity, and finding that I did like some of the design decisions made (more portable, using XML for config files, using SQLite for various internal storage, and the modularity).
    But if what Bill Miller of Digium says is, in fact, true–that “You should recognize that the Asterisk rhetoric from the sipXecs and FreeSwitch teams refers to Asterisk over 4 years and several versions ago when they last looked at the feature set”–I would be very curious to see a more accurate comparison done, say between the current stable releases of Asterisk v1.4.21.2 release (since v1.6.x is still classified beta) vs. FreeSWITCH v1.0.1 vs. sipXecs v3.10.
    Though benchmarks fall into that area that reminds me of the expression “There’s lies, damn lies, and then there’s statistics”, at least it would provide some measure of comparison. Sure, one set of benchmarks could be done on stock installs, vs. another where each community could tweak their configuration to the fullest. But such comparisons can only help bring more folks into the whole open-source fold, especially if the same benchmarks could be performed against other, closed systems.
    Anyway, thanks Bill Miller for the response. And thanks to all the projects for giving us the freedom to choose.

  • Anthony Minessale
    August 21, 2008 at 5:05 pm

    The sad thing is this article is posted officially here:
    http://blogs.digium.com/2008/08/19/asterisk-the-global-leader-in-open-source-ip-pbx-and-telephony/
    And my comment was censored by the moderator. (nice)
    Here it is:
    ——————————————————–
    I didn’t know open source projects had to compete. They are all free aren’t they? I think that’s the whole problem here. Calling my list of valid issues with Asterisk rhetoric, won’t make them go away. Pretending Asterisk does not suffer from any problems and only pointing out it’s strengths is not the way to make it better.
    Please admit that I have done more than most people are willing to do for completely FREE to try to make Asterisk better for several years. I only know enough to itemize the issues from that *long* experience as a an Asterisk developer.
    I, in fact, invented the whole idea of the “function variables” that now are rampant in Asterisk 1.6 and there are *plenty* more things I could list if I wanted to. I also see plenty of ideas we have already implemented in FreeSWITCH starting to crop up in 1.6 as well. This is the nature of open source. If Digium chooses to actually cooperate with the open source telephony community there is much to be gained for all.

  • Jason Sjöbeck
    September 29, 2008 at 6:24 pm

    As the leading installer & integrator of Asterisk, TrixBox, PBX-In-A-Flash, Finality, SwitchVox, freeSwitch & sipX and all the others here in Oregon, we say: let the best functionalities spill forth.
    We can not wait.
    Thanks very much.

Leave Your Comment