© 2022 Black Swan Telecom Journal | • | protecting and growing a robust communications business | • a service of |
Any market analyst worth his salt cherishes his role as an industry cynic. We just love to scorn a hyped technology or blast “conventional wisdom.” It’s especially fun if our bold predictions come true. And if not, well, at least people’s memories are short.
Perhaps the greatest of cynic of all time was the Greek philosopher Diogenes the Cynic, born around the year 412 B.C. Diogenes was the bad boy of Athens. For sport, he liked to intellectually humble Plato and sabotage his lectures. He would sometimes carry a lamp in the daytime, claiming to be looking for an honest man. And he was one of the few men who ever publicly mocked Alexander the Great and lived.
Well, at a recent Telcordia analyst conference, I got a chance to play Diogenes for a day.
Telcordia has a great ritual of periodically flying in a couple dozen analysts and press for a day and a half of briefings. The discussions at these events are free-wheeling and a chance for analysts to pepper Telcordia execs with some tough questions. So I popped this challenge to Telcordia about its Service Delivery Platform: “Your competitors are always telling me that service delivery platforms born in the IN technology era are inflexible and expensive to maintain.”
And Grant Lenahan responded, “Dan, that’s just not true. And you really owe it to yourself to pay a visit to our customer briefing center to see our SDP in action and hear the full story about our technology.”
Grant should require no introduction. He’s a frequent B/OSS columnist and conference speaker. At Telcordia, he’s vice president of service delivery and strategy, but he also fills the role of what I’ll call Chief R&D Translator. The history of Telcordia is long and deep and there are so many R&D projects going on in Piscataway that you really need a business-oriented interpreter like Grant to make sense of the stuff the scientists and Ph.D.s are cooking up in the lab.
Well, I took Grant up on his suggestion and made a special visit to see the SDP platform in action and interview Grant. And my original cynicism about IN’s relevance has dissolved. As you’ll read, plenty of under-the-covers innovation is happening in the IN platform space these days. Here’s the edited transcript of our discussion.
Dan Baker: Grant, let’s start with my original question. Are the IN (intelligent network) prepaid platforms of the world as inflexible and expensive to maintain as your charging competitors say? |
Grant Lenahan: It’s a mixed bag, Dan. And it’s also hard for me to generalize because I’m not privy to the nuances of my IN competitors‘ products.
One thing to understand about IN is the difference between where it started and where it is today. In some cases, there is a direct link from the legacy platforms. In other cases, there’s much less of a link. At Telcordia, for example, we have considerably enhanced our platform over what we had 15 years ago.
Some of today’s IN platforms are hardware based; some are a combination. Almost everything these days uses lots of software, but still, many of the IN platforms are quite proprietary. So if you are buying into a hardware-based architecture, you are still buying something at a fairly high cost.
Back in 1993 when I worked for VDC Research, I authored a research study on the IN platform market and I remember at the time Tandem Computers (now HP) was the market-share leader. |
Tandem had its Non-Stop platform so their IN implementation was based on custom high availability hardware. That solution had two of everything on the hardware side for redundancy.
Telcordia earned a nice IN reputation because we were the first company to build true high-availability, network-grade stuff on open Unix hardware. If I recall, our first Unix system came out in 1992 and was running on the IBM RS/6000. And the price point was very low. You’re talking $100,000 per box and maybe you needed seven of them. That was a fantastic price at the time.
Well, that was a long time ago. And over the years we’ve pretty much re-engineered that system to the point where there’s almost nothing left of the original. Today we run on blades, which happen to be IBM pSeries blades because they are in a telco stack.
How much of the IN platform market would you say is still being used for enhanced services like 800 numbers and CLASS services like Call Blocking and Caller ID? |
I’d guess that two-thirds of the market is being used for prepaid and charging applications these days. Prepaid is definitely where the money is. And prepaid is what forced us to do such a rapid re-architecture of our product. If we could cruise along with only enhanced services or AIN functionality, it wouldn‘t have stressed our systems very much.
Prepaid, particularly in places like Oi in Brazil and Tata Teleservices in India, is taking all the performance we can throw at the platform. And the price points we need to reach are unbelievable. One of our customers now bills SMS based on the number of characters transmitted. Think about that: The cost of our charging transaction has to be a tiny fraction of the cost of sending a short SMS like that.
So what improvements have you made on the high-availability side? |
Well, I’ll start with high availability middleware. Now why is that important? Because we have built something that simply doesn‘t go down.
Tata in India is between 90 percent and 95 percent prepaid. Almost nobody is postpaid. So, this isn‘t about gold-plated reliability or your reputation as a reliable carrier. If our system goes down, nobody is making calls and Tata is not making money!
So how do we achieve that? Number one, we architect the system with shared loads across multiple sites, so if a site goes down you’re automatically working at another location.
We always use at least two systems separated by at least 25 kilometers. And we make sure they are not located on the same earthquake fault, river, or fiber leader.
Not only that, using multiple copies and some unique technology, we can actually upgrade you to a new version while the system is running. We think we may be the only ones who do that and have geographic redundancy at the same time.
The way it works, you go through stages of installing and testing, then building a new stack up on the one that you take down. The process needs to be pretty fast and automated. It is not like you can build it up, flip a switch and find it didn‘t work and try again.
Grant, one of the difficulties of IN is that it’s a protocol developed in an older era. What have you done to support the latest telecom protocols? |
To be honest, though its origins are in IN, our Service Delivery Platform is a generic, protocol-agnostic app server that does things like charging and advertising applications through a library of protocols.
We mediate all sorts of protocols in the platform — Diameter, IMS, GSM and more. And granted, some of those interfaces are more mature than others because you can‘t build everything at once. But all these protocols come in. So, our Diameter access to the system, our IMS online charging system, our Web services access to it, traditional IN access, GSM access, and our SIP access to it — these are all coming into the same logic with the same customer data.
So we built is a network control application server that runs on Unix — on blade servers and on open stuff. But be careful: If you want high availability, you can‘t just go to Dell and get whatever you want.
What I like about that is if you want to upgrade to the next-generation technology, you’re ready because it’s already on the same platform. |
Exactly. And in fact, even though our customers dominantly use GSM protocols, they are very, very interested in making sure they can build their content charging and their IMS charging on the same platform as their GSM.
Another advantage here is if someone runs out of money, you don‘t need to drop the call, you redirect them to a website where they can add money to their account. Oi is real clever here. They redirect you to a website that interacts with you and can even create a short-term, high-interest micro loan so you can keep using your handset.
We haven‘t talked about the Service Creation Environment yet, but I imagine that’s come a long way in the last 15 years as well. |
Service creation is probably where we’ve focused most of our development energy. And that’s the functionality that really separates the IN platforms of today from those of yesteryear.
Years ago, Lucent’s IN architecture was considered advanced because instead of a proprietary-based system, Lucent allowed you could to program it in C++. Well, the good news is it was C++, and the bad news is was C++.
Early on, we built the first GUI called SPACE that allowed you to parse SS7 and tie it back to the many things you want to do with it. It allows you to drag and drop things to create a service.
But we also have a modern integrated development platform based on ECLIPSE, which is controlled by IBM and is a fairly open environment that allows you to put libraries of objects into it. And frankly that’s where more and more development is going these days, so we give you both options side-by-side.
Now to further impress the techies at an IBM or Oracle, we also have a Java execution container that works with HTTP or SIP.
And so all of this taken together is why we say our platform is quite a big more than IN alone. We actually have a place where you can run the programs you write or that we write for you.
Even more importantly, in the rating area, our editor is designed around rules or policies.
The earliest rating engines worked off of hard-coded prices and equations and that was real clunky to change. Then someone came up with the idea of table-driven rating, so if you could enter any number of rates — an 8 o‘clock peak rate, a 6 o’clock in the evening off-peak, and so forth. But even tables became a nightmare to maintain.
I think Telcordia was the first major vendor on the prepaid market with a rules-based approach to rating. And this came out of our research labs.
OK, Grant, we’ve covered high availability, flexibility and service creation. But can it scale? |
Scalability is critical to on-line charging platform performance yet we’ve merely scratched the surface.
In the future, its not just prepaid guys who will roll with real-time, the big postpaid world will need real-time for things like advise of charge, parental controls, and subscriber preferences. Just as you log onto Yahoo! and get your personalized stocks, news, weather, and sports games, telecom networks will personalize your handset like that. And the transactions numbers are going to jump like crazy.
Today we support over 50 million online users at one of our clients. And by the way, that’s in one logical database. That’s something many of our competitors won‘t tell you about — particularly the little companies because they just don’t have the resources to build a large-scale system. They say, “You want more capacity, fine, buy more pizza boxes from us.”
Ours doesn‘t work that way. Yes, we spread the load across multiple machines, but it’s one logical system. That is very important to customers who are trying to project capacity and availability. Otherwise, you get stranded capacity.
Stranded capacity adds up to big money. But it’s more than the hardware costs. You have SS7 links coming in, IP links, and people need to working on these systems to preserve the integration. So, all of these things add up to real money.
Grant, thanks for this insightful discussion. I’ve got a new respect for IN platforms and some questions to ask of some competitors of yours who paint with a broad brush. |
Much of what I’ve talked about is unique to us. But you’ll hear a similar story from some of the IT-based guys. But push them on their IN capabilities, push them on GSM, CDMA, U.S. IN, and European IN.
And there’s an analogy here. GSM is a pretty old technology, but it’s still the dominant communications protocol in the world. Likewise, IN is an SS7-based that’s even older than GSM, but that doesn‘t make it any less smart or less relevant to the future.
This article first appeared in Billing and OSS World.
Copyright 2011 Black Swan Telecom Journal