In 2006, I came across a Network World article, which espoused the fact that Sam Houston State University (SHSU) had switched from the Cisco CallManager IP-PBX to open source Asterisk. I wrote about this news since 6,000 students and faculty were moved off Cisco to the open source Asterisk IP-PBX, which was great news for the open source Asterisk community. This deployment demonstrated that Asterisk could scale and put to rest one of the main complaints against Asterisk.
Well, 3 years have passed, and according to this thread written by Jason Fuermann, who is responsible for SHSU's IP phone system, SHSU has switched back to Cisco from Asterisk. Say what?
Let's go back to what I wrote in 2006...
So if the main driver to switching to Asterisk was lower costs and lower TCO, why would they switch back? Who wants to pay more for something? Is it because they thought Cisco had superior features? Is it because they thought it was more reliable? Is it because Cisco begged or bribed them to come back? Is it because of turnover in the computer/telecom department with someone that insisted on Cisco only?
A 'hint' can be found in the 2006 Network World article:
Jason writes:
Apparently, people with Cisco certifications are still more ubiquitous. Although, there are many Cisco certifications in networking that are not voice over IP related. In this case, you'd have to be a Cisco Certified Voice Professional (CCVP) to be a certified CallManager expert. Asterisk isn't that hard to learn, so I'm surprised this is the reason. Also, if someone is a CCVP, I guarantee they can pick up & learn Asterisk very quickly.
What say you?
The main reason for this migration was cost, according to Aaron Daniel, senior voice analyst at Sam Houston State University. "We thought that it will be more cost effective in the long run to go with an open source solution, because of the massive amounts of licensing fees required to keep the Cisco CallManager network up and running," says Daniel. According to the article, each phone attached to the CallManager required a separate annual licensing fee to operate. I'm not sure that's entirely accurate. I could have sworn you weren't required to pay annual licensing fees for the phones, but you did have to pay an optional support cost based on the number of phones. But assuming this is true, this could become yet another strong driver to cause Cisco CallManager shops to jump ship to save on TCO (Total Cost of Ownership). In SHSU's Asterisk/Cisco setup, they will keep their existing Cisco phones but attach them to Asterisk servers on the back end, thus eliminating the phone licensing costs.
So if the main driver to switching to Asterisk was lower costs and lower TCO, why would they switch back? Who wants to pay more for something? Is it because they thought Cisco had superior features? Is it because they thought it was more reliable? Is it because Cisco begged or bribed them to come back? Is it because of turnover in the computer/telecom department with someone that insisted on Cisco only?
A 'hint' can be found in the 2006 Network World article:
The answer is revealed in the thread posted by Jason Fuermann, who works at SHSU and is responsible for their IP phone system.In another potential issue with open-source VoIP, SHSU loses the technical support from Cisco with its Asterisk migration. But Daniel says he has so far been able to keep up with support issues through mailing lists and the online community that develops and supports Asterisk.
Jason writes:
So the reason was not more features in Cisco. It wasn't lower TCO and it wasn't Cisco bribery. Nope, the reason was the university lost personnel that were familiar with Asterisk and they want to be sure they can hire anyone off the street to maintain their IP-PBX.Well...
If you can't tell we've moved to call manger. There were a lot of factors for this. Most significantly we were essentially maintaining our own branch of the code since we had to develop some of the features that were required and were in the process of trying to figure out how to program shared line appearances. In the beginning that wasn't a problem (we were used to it since we were a majority open source shop), but after a significant loss of personnel these types of systems became much harder to maintain. It was decided that we needed something we could bring joe shmoe off the street to administer and here we are now. We successfully rolled our entire campus to voip and haven't looked back.
Apparently, people with Cisco certifications are still more ubiquitous. Although, there are many Cisco certifications in networking that are not voice over IP related. In this case, you'd have to be a Cisco Certified Voice Professional (CCVP) to be a certified CallManager expert. Asterisk isn't that hard to learn, so I'm surprised this is the reason. Also, if someone is a CCVP, I guarantee they can pick up & learn Asterisk very quickly.
What say you?
Technorati
Del.icio.us
Slashdot
Digg
twitter
Learning Asterisk is a lot different than learning a whole branch of code stemming from Asterisk. I can walk into any CCM shop and have a good idea of what is going on. Walking into a customer programmed Asterisk shop would take significantly more time to acclimate.
Picking up asterisk and being able to maintain a custom branch of asterisk are two different things. If they rolled it out three years ago they probably started coding four years ago. The would have been using 1.2.X and probably still there.
If a critical feature was missing they would have been better to have a 10K bounty and have it submitted back tot he community for inclusion into the main branch.
I think they leaped before they looked.
-Matt
It also might have something to do with the fact that virtually all of their custom 1.2 code would have been broken by Asterisk 1.4. Doesn't take long to eat up your savings when every release of Asterisk requires a major rewrite of all your application code. Some of us have tried to encourage Digium to adopt a different design strategy, but you can guess what their answer was and continues to be.
I think it is also got lot to do with planning. Perhaps SHSU did not foresee the future requirements completely. May be Aron Daniel left the scene the rest did not know how to hold all the ropes. It is different in academic world, some like to hold the ace cards!
Perhaps it is time to promote Asterisk Certification!
I agree that maintaining and improving your own branch of Asterisk will require a ton of resources. However, this is not how open source works!
I understand that they needed new features, so they added them. That is well and good. The question is why didn't they contribute those changes back to Asterisk? If they did, the entire community would benefit and start maintaining the code as part of Asterisk.
Any news on why the open source process was not followed?
Thanks,
Tristan Rhodes
I would like to clear up the "they should have committed upstream" comments:
I'm currently employed as a GNU/Linux systems administrator for Sam Houston State University, and while I wasn't personally on staff here when the Asterisk system was being worked on it is my understanding that the personnel that was involved with the development did actually commit the developments made upstream and that ended up getting them a job offer from Digium (causing part of the mentioned lacking personnel from the article) which then left SHSU in a situation because the former employee would no longer be focused on the feature requests of SHSU, but of their new employer. At that point it became a management decision to make a switch to something that already had all the features needed or to hire/pay/contract/whatever someone new to continue to develop them, and as much as I'm an open source advocate I can't fault their decision from a business perspective.
Thanks,
Adam Miller
At my place of employment, we're heading down this road as well. Not necessarily trading out all of our Call Managers for Asterisk boxes. But Asterisk does a bang-up job in place of Unity for Voicemail Services. Integrated with our Cisco Call Manager Cluster (SIP Trunks), we're able to offer all of the same services for less than 1/3rd the up front and recurring cost as Unity.
Cisco will bleed you dry with licensing. It's true, they have a great product, but when it all boils down to it...Asterisk can give you most of what any Cisco product can provide for a much more manageable TCO.
That said, it is an Open Source Product and having appropriately trained staff is key. Fortunately, if your staff includes Linux Admins, some developers, and engineers familiar with voice services, then you should be well prepared to support the product.
To reduce our exposure and any potential risks, we went with Digium's "Asterisk Business Edition" product with a Digium Support Package, so we're well prepared for anything that might arise that cannot be solved in house.
In addition, the community support around Asterisk is nothing short of amazing.
Seems to me the folks at SHSU probably could have made it work. But my guess is the pressure was on, and they didn't want to assume the risk.