CAMARA: Open-source API for telecom and 5G networks
gsma.com22 points by teleforce 2 days ago
22 points by teleforce 2 days ago
The telco operators really need to ask themselves how it is that in over 30 years of attempting to expose the telco network to developers, not one single attempt has succeeded.
- IN - the "intelligent network"
- AIN - the advanced intelligent network
- SIP/SDP app servers
- JAIN app servers
- 4G SCEF (service capability exposure)
- 5G NEF (network exposure)
- 5G AF (application function)
- CAMARA
I've probably omitted half a dozen more.
My diagnosis is that the telcos have nothing that developers want and plenty that they don't want. For example, no one is crying out for low level access to the 5G network slicing function and they're definitely not in a hurry to tie their ability to make money to whether they can integrate with some telco's billing system.
Often times a telco will throw up a slide that has, say, a picture of a robot or car using some nifty feature of the 5G network -predictable latency, say- and suggest that developers could use the network for these types of applications. Do you imagine that a FANUC or a Tesla would seriously tie their fortunes to whether some AT&T person would let them get a few bytes across the network? Practically, no way.
There is a demand from developers for telco services, but it is at the level of making API calls to send SMS or make calls. Twilio made good money out of it while the telcos were not looking, or perhaps less charitably, while they were so tied up in whether this SIP header or Diameter PDU should be allowed into their hallowed network that they didnt realize that the world had passed them by.
While I agree with everything you've said, I also wonder if some of it comes down to how hard it is (in my perception, at least) to get access to it? With almost all other developer-facing technologies, you can get a sandbox environment set up, a free tier subscription going, etc., within a couple of days or so if not a few hours.
With the exception of SIP, I have no idea how I'd learn or demo any of the things you mentioned as a random developer. Even if my company asked me to. My immediate assumption would be that you need to be a fellow telco or a multi-million dollar customer to get any kind of access to this stuff. And that setting up a lab of your own to play around with requires pouring though thousands of pages of obtuse standards and documentation to make any sense of it, not to mention the radio hardware if you want to actually interface with modems?
It's a shame, because telecom seems very cool, but I just have no idea where you'd even start.
The funny thing is all of the above APIs? They are actually used, hell, some were used quite extensively in my memory.
It's just that we're in a bubble that does not intersect with that bubble, and a lot of the stuff became less important with always on IP connectivity and big screen smartphones with local applications and a web browser.
Some of the mentioned technologies are also used explicitly to build the actual network, especially combined with the long legacy tail involved
I contracted at one of the Bellcore spinoffs for a while. You're not really going to get anything resembling innovation out of them. They're mature businesses left over from the last millenium, optimized to extract maximum cash from existing products.
They also benefit from regulatory capture - the CTO where I was used to spend a lot of time in DC. Competing on tech capabilities is just not what they do.
Sure, operators upgrade their networks from 3G to 4G to 5G, but they're not the ones driving that innovation, they're just deploying stuff developed at other companies like Qualcomm and Ericsson.
Telco != Techco.
Telcos are cash-flow businesses. They borrow billions with bond sales to fund spectrum acquisition and hardware / network deployment, then optimize payback and profit by taking $80-$200 per account per month from millions of people.
There are smart, forward looking people inside these companies who want to build for 3rd party devs. And they might even get the dispensation from some SVP to start working on an offering, and occasionally one of those makes it out the door to real developers. But ultimately, it isn't a priority for the C-suite, and when that SVP moves off to some other group or company, there's no more sponsorship for the "innovation".
As long as they take in billions reselling Apple/Samsung hardware connecting to Ericsson / Nokia hardware, they're not focusing on anything else.
Telcos are just not set up to produce a usable developer platform. Too many perverse incentives, politics and conflicts of interest, internally.
Alliances like CAMARA usually fail for a few reasons:
1. Telcos want all new features to be purely additive. They're not willing to expose API calls that are even slightly uncomfortable or might cannibalize existing business, for the same reasons as above. As a result you get features no one really asked for.
2. Alliances often fail due to competitive dynamics. The moment a telco group that competes with another telco group or a geopolitically inconvenient telco joins the alliance, work often stops.
3. Inability to find long-term alignment on vision and continuity.
Producing an API would also not solve for all the other operational & regulatory reasons telcos are tough to work with.
The same way the music, movie, and banking industries haven't produces their own digital layer, true innovation in the space will only come from OTT players that create globally scalable infrastructure, exposing network features to engineers, enabling them to spin up in a matter of hours.
That work indeed takes years and millions of USD. At Gigs.com, we're doing just that. Building Stripe for telecom. We believe that by giving the best technology companies in the world access to connectivity, we will finally start seeing a lot of innovation in the space, after 3 decades of stagnation.