jtodd July 14th, 2009

The fact that Asterisk is being used by huge numbers of people is not a question – it’s clearly the case based on the number of how-to articles, the videos, the books, the classes, the consultants, the twitter posts, the blogs, and the numbers of people on the mailing lists and other forums dedicated to exchanging ideas on both open-source and commercial implementations of Asterisk.  The fact of widespread Asterisk adoption is obvious – but then comes the question: well, it’s a big community… but exactly HOW big is it?  How many implementations are there in education?  Government?  Big Enterprise?  Personal use?  Carriers?  We have what we think is an idea of how large the Asterisk community is, created out of numbers that Digium has conservatively built from our own customer base, our data extracted from our download knowledge, and from surveys performed by other companies.

There is, I believe, a much larger community of Asterisk systems out in the world than anyone may imagine.  There is a larger portion of the community under the surface, hidden from what we can see.  I hear little from non-English-speaking markets (I admit this has to do with my own inability to collect data from those particular communities) and we can see large numbers of Asterisk downloads from many nations which are under-represented or not represented at all in the common forums.  I receive questions frequently from large companies and institutions that are all but invisible in the public forums.  I talk to consultants who have installed astonishingly large and complex Asterisk platforms but are not inclined to try to push upwards into management to get public acknowledgement of those contracts.  I have myself built systems that were quite large and sophisticated, but have not been able to talk about them in detail.

So why is this a problem?  There are a number of reasons that such discrepancy between actual and perceived installations is a problem, but the most troubling aspect of this disparity is related to adoption rates.  Asterisk is getting short shrift.  Despite its widespread use and even decent name recognition, I think that Asterisk is nowhere near as well-known as it should be, and this lack of exposure leaves many unaware of how flexible, stable, and scalable Asterisk is compared to proprietary solutions which live and die by statistical analysis of their installed base.

Linux is an example that I also think is an excellent model to look at as it is similar in many ways to Asterisk.  I believe that Linx “crossed the chasm” because of name recognition, driven by a superior platform and then exemplified in actual installation percentages.   While obviously the growth of Linux is a complex process to describe, it is certainly fair to say that it would not have grown if decent numbers of people had not been advocating their success with the platform in a measurable way.  Or perhaps Linux did not grow as quickly as it could have because this measurement was not particularly clear.

Asterisk is now moving into new territories of both small and large business.  These territories are inhabited by less adventurous sorts, and they want to feel comfortable that a large number of their peers have already moved in this direction before they’ll make a decision themselves.  This cautious outlook may not be the same as many Asterisk adopters currently view the world, but Asterisk is no longer only in markets where the understanding of cutting-edge technology is the actual product.  Brick-and-mortar industries, finance, education, government – these all are less prone to thinking in risky terms when it comes to implementing IT systems because that is not their business.  I believe having a realistic and strong set of data that describes how Asterisk is being used would have an enormous positive influence on uptake in these new adoption areas, and the project would benefit in many other ways as well from collection of usage statistics.

We (Digium) have relied on our own surveys of our customer base over the years to give us an understanding of certain sub-markets – notably, those who use Asterisk for connection to legacy phone systems via PRI or analog lines, or those who use our pre-packaged versions of Asterisk in SwitchVox.  For those markets, we have a very good idea about our customers, but we are limited in what data we collect on open-source Asterisk users.  Several analysts have created very nice profiles of how Asterisk is expanding (Frost & Sullivan report) or how Asterisk-based PBX ports now outnumber all other new installations (Eastern Management Group) but we think that we can do better than reports collected by random survey.  The project is not lacking in possible methods of collection, since we have the most direct and informative method possible of collecting data: we have Asterisk itself.  Asterisk could have (but does not currently) a system for reporting back certain statistics about itself to a central database, either via the build process (makefile) or via a method directly compiled into the software itself.  This set of statistics could be user-configurable to reveal or hide whatever the user felt was reasoanble – even just knowing that Asterisk was installed on a unique hardware platform would be more direct data than we collect today by a large margin, and any other data (kernel type and version, loaded Asterisk modules, etc.) would be tremendously useful for open-source project directions and effort estimates.  It may be overstepping the bounds that most people would choose, but even reporting back on a regular basis with things like core dump events, minutes of uptime, and number of calls would be “gravy” that could be used to track bugs and improve performance in a transparent manner.

But we have shied away from any type of automated collection of data regarding the installed base.  There has been a reluctance to have any sort of information collection routines, and the discussions have stalled not because this information isn’t useful, but because there is a perception that such data collection is not useful enough to deal with the headaches of overcoming potential community objection to opt-in (or opt-out) data collection.  Would you object to optional collection of anonymous data at the point of installation, or during the operational life of your Asterisk system? This post serves to start a discussion on that topic, and I’d like to hear your views either privately or via the comments in this posting.  Should we ask the community to show us their Asterisks?

JT

• Community , Digium , Open Source
• Comments(5)

About the Author