Slashdot Mirror
Is Linux Used in Production Telephony?
jamesva asks: "The telecommunications industry is rapidly converging on Windows NT/2000 for all telephony and voice-related needs. Most ACD systems, virtual operators, and voicemail are being ported to Windows if they're not already running on it.
In the past, telephony apps have existed most
notably on OS/2, SCO, and even DOS. However,
free Unix (or unix-like) platforms have absolutely no penetration in this area, with seemingly no chance on the horizon.
The Bayonne app server from the GNU folks seems to be the one exception, but even then there doesn't
seem to much built around it or anyone using it. It reached a 1.0 release in September and was met with no fanfare. Even the
LinuxTelephony doesn't seem to have much news. Can someone prove me wrong? Why is this the case? I'm interested in finding out if anyone is using Linux (or any free OS) in a production environment for something like voicemail or ACD. These types of systems require high availability and reliability and Linux just seems like a natural fit."
Linux being the DIY operating system that it is, people tend to write drivers for the hardware that they have. How many linux hackers have dialogic boards in their machines? At >$500, I doubt the number is very high. No drivers, no applications.
The first question I'd ask is: Are the applications there? If not, there it is.
You mentioned one application that uses Linux. There are probably many more that work under Windows, because that's probably what companies are developing for. More to the point, that's probably what companies are asking for -- "Give us something that looks like what we're used to for web surfing already!"
Now granted, within the past few years Linux's desktop has grown leaps and bounds beyond where it was -- but then, it wasn't there when these companies first started developing their apps, and wasn't an option then.
That, ultimately, is the issue.
What runs on my 5Ess then? Or Voicemail system? UNIX has had fantastic penetration in the telecom industry, what with being written in large part by a telco, for telco use. (SYSV)
Linux penetration is a totally different story. Unless I see less than 5 minutes a year of downtime, and more than 20 years of hardware and software support for a platform, I can't see using it any time soon.
-Besh!
We only run unix os's with a few NT machines (mostly admin consoles) in our datacenter. The problem with putting a Linux box in production, we have platinum support from Sun. Every server is a sun box, and its standardized for backup, database and clustering. If we put a Linux box, it has to run it on sparc hardware, and we have to have special procedures just for this one box.
Its much easier to run GPL'ed software ported to Solaris, than to switch the OS. We in fact run many GPL'ed software packages, the cost saving is amazing. The backend software is highly specialized, and will not be ported to linux.
To make sure the software is locked in production, the developers put license strings for everything, and then they lock it down to IP/Domain/Hardware/os version/etc..
Sometimes you want the software to be written in house, but with the features, support, updates to software, its easier to write a check and get everything at once. If you want to know who the main players are, Nokia, Nortel, Software.com and Ericsson are the largest players.
I can't speak to what software runs on phone switches, but I can speak as a user at the "medium sized company" level, and as a user I can say that the industry seems to work primarily with embedded boxes for telecom. When you want a switch you buy a switch, and it does what it does. Whether that switch runs linux or SunOS or VxWorks or some proprietary OS is pretty much irrelevant to you if it functions for you in its capacity as a switch. If linux is being used as the basis for phone switching equipment, people probably wouldn't know, unless they had some contact at the company who developed the switch. This is a traditionally very embedded market, where name recognition of an OS like "Windows" or "linux" or whatever is irrelevant to the function of the device. Telecom can be thought of as the ultimate high availability application. In all my dealings with telephone switches, nothing ever crashed or needed to be rebooted. EVER. Even when installing new hardware. This kind of high availability doesn't readily lend itself to traditionally end-user oriented operating systems. I suspect the reason linux isnt perceived as penetrating the telecom sector is because its not, and if it were, it wouldn't matter because people who set up and managed the switches, by and large, dont give a shit how it works, just that it works, that it works all the time and never stops working. :) If your job is to turn a nut, does it really matter if you use a wrench, pliers or your fingers as long as the nut gets turned?
This too shall pass.
leaders in this area until a few years ago (I haven't been keeping track, so they still may be).
But all the 3rd party vendors recognized that UnixWare was a sinking ship, and started asking the hardware vendors (DLGC, NMS) for Linux support. Why? Because it's cheap and it at least looks like the redheaded stepchild UnixWare never had. So with a little work, the 3rd party vendors have their app running on Linux.
Add into this the fact that Microsoft also recognized UnixWare was a sinking ship, and started marketing. Remember, it's all about developers developers developers. So many 3rd parties were confused and switched their app to Windows (partial rewrites) instead of switching to linux (minimal porting effort).
I think that for some people who ended up on Windows, there may be some displeasure with the reliability / quality. But I think they are probably outnumbered by the people who are having positive results with W2K/NT in these setups. The net effect is that Windows is gaining in this market.
Back when I did telephony programming, it was all on OS/2, largely because OS/2 was a realtime OS, while Windows is not. *nix also generally has realtime kernel support available as well, and the high-end switching software that isn't proprietary generally runs on *nix.
At the time, however, hardware providers were working on Windows APIs, probably because of the prevalence of Windows in the workplace. There was a big push years ago towards PC-managed telephony over old proprietary PBX systems. It gave businesses the ability to have their IT staff do a lot of customization without very much training.
The important thing to note is that the migration to Windows was on the corporate end, not on the provider end. There's no way any seriously critical switching software is going to be running on Windows.
Granted, a lot of what you said is subjective, and as you acknowledged, there is no point in arguing over opinion. Furthermore, I feel I understood what you said. In support of this, allow me to summarize your post: You a manager responsible for critical systems, and you absolutely do not modify your production environment without significant justification and complete risk-management.
..., etc.", but as "I do not have enough information about Linux and its tools to implement solutions for that platform". I firmly believe that _with proper procedures_ the appropriate Linux tools can be implemented in a risk-free way, its just a matter of knowing how to do it or knowing someone who does.
I don't think you're alone in that position. In fact, there are likely a number of Slashdot readers that are also responsible for business-critical systems. I even read an occasional story about businesses switching their mission-critical systems over to the Linux platform.
What this means is that even if you aren't comfortable with Linux, there are those in positions similar to yours that are, and they are growing their business in areas they could not otherwise.
It is possible that there does not exist a Linux application that could grow your business in any significant way. If the benefits of a change do not outweigh the effort involved, I admit it is a pointless modification.
However, if there exists a tool that could provably benefit your company, for example, by increasing profitability or decreasing the costs of operation, would you be biased against it because it runs on the Linux platform?
I am planning to start a business myself, and there is no part of it I would trust to the Windows platform. To use your words, I find it "loose, uncontrolled, and random". It does not suit my needs, and Linux does.
In closing, I interpreted your comment on the whole not as "Linux apps are too risky for me because of
Why's that free platforms don't perform well at Telephony servers? The Linux telephony is just not structured enough. Because there's not for Linux one guy who's going to application writers (ACD, pre-dialers etc) and telling them "Hey, we're going to build a soft-PBX platform which is going to enable you to sell applications for more platforms than today. We're going to build a Telephony API for ya, how do you want it to be?"
And then the same one guy gets to telephony H/W manufacturers and says "Hey, we're going to build a soft-PBX platform which is going to enable you to sell more stuff. Only thing we need to do is get the drivers right, and you wouldn't want to see your cards unusable with our platform. So we're counting on your help."
The key there is that when the middle-man (MS) says he's going to build a soft-PBX platform, others know they can rely on him because he's got an interest in the thing (selling more licences), so they actually help him out. Application developers and hardware manufacturers think alike: "Why invest into a relationship with a bunch of noncommitted hobbyists who have no clear interest in making the technology work? What if I don't like what they do, can I say anything? Why don't I just go with the guy whom I know shares an interest in the technology? And in the end, you've got a system running and everyone finding a way to profitability.
It's a partnership between people or companies that works.
I think it's more of a mystery why certain free projects perform so well rather than why many fail.
Fundamentally, because FreeBSD is an Operating System.
As you've guessed by now, I subscribe to the train of thought that Linux is *not* an OS. Red Hat is an OS, Mandrake is an OS, Caldera is an OS, Debian is an OS. Linux is just a kernel.
Why is this an issue for me from a commercial point of view? Well, with FreeBSD, I get to go back to one group for any issues. The whole thing, kernel, binaries, packages, - the works - is managed and documented as one single entity.
If I go Linux, I've made a kernel decision. FWIW, I believe that the Linux nothing short of a bloody brillian kernel. What lets Linux down in my eyes is the operating systems that use it. I have to choose which one, then I need to be convinced that the vendor is committed, that the vendor manages the system well, that I'm not going to be faced with "we didn't design that part" finger pointing wars when I have issues...
I hasten to add that FreeBSD isn't, IMNSHO, the grand solution either. In fact, FreeBSD is annoying the shit out of me right now. I cut my teeth on BSD boxen way-back-when, I'm a BSD boy at heart. Still, as I intimated earlier, my home runs on Linux. I have Mandrake/Intel and Cobalt Qube/MIPS boxes as servers, and Red Hat for a desktop. There are no up/running BSD boxes of any description here.
Why? 'cos the Qube was a freebie \, Mandrake proved itself be be a good server in lieu of a suitable BSD, and Red Hat seemed like a good thing to play with for a desktop. FreeBSD got the arse because the interrupt code in the PCI stuff is broken, and the PCI-PCMCIA bridge stuff I need to make my wireless gear work, doesn't.
On the other hand, I can't afford to be so fickle at work. I need a whole OS with a strong design and project management behind it. That's why, as I said, Solaris, FreeBSD, HP-UX gets the vote. Two of those are commercial products with the incumbent entity-to-point-lawyers-at that senior management loves so much. The other has proven itself to me.
About the only thing I can think of is the fact that Linux distros generally install the kitchen sink, and FreeBSD prefers to make you add ports/packages yourself.
To be honest, that's something that I think most of the Linux based OS's do better than FreeBSD does. FreeBSD invites you to pick a generic build, and, as you say, installs the kitchen sink. Many of the Linux OS's offer much better control
At, I might add, the cost of speed. Scripting is the solution of course, but I can still have a good working FreeBSD system up and running (including custom kernel config and compile) in less time than Windows 98 can be installed (bad comparison, I know, but I raced a guy once, that's how I know!). On the other hand, I've had way too many two and three hour missions installing some Linuxes. Not good.
Is it some sort of "stability" issue? What kind of stability?
In a way, yes. It's the kernel -vs- whole OS thing, and the business-comfort that comes with it.
I find your ideas intriguing and I wish to subscribe to your newsletter.