AT&T Open Sources Its SDN Framework To The Linux Foundation (fiercetelecom.com)

An anonymous reader writes "It's no secret that AT&T has been planning to move to a software-defined network for quite a while. Now, they've decided to open-source the whole thing." From Fierce Telecom: AT&T today announced it will release its Enhanced Control, Orchestration, Management and Policy (ECOMP) platform to the wider telecom industry as an open source offering managed by the Linux Foundation. The goal, the company said, is to make ECOMP the telecom industry's standard automation platform for managing virtual network functions and other software-centric network capabilities.
AT&T delivered 8.5 million lines of code to the Linux Foundation on Wednesday, saying "We want to build a community -- where people contribute to the code base and advance the platform..." AT&T said Wednesday they've already received interest from other major telecoms, and "we want this to help align the global industry." While their ultimate goal is to virtualize 75% of their own network by 2020, at least one analyst sees a larger trend where the whole telecom industry collectively bypasses equipment vendors and begins "taking network innovation into its own hands."

Comments

AT&T delivered 8.5 million lines of code

[fustakrakich]

Oh, that's right! Drown 'em in paperwork.

Re:AT&T delivered 8.5 million lines of code

[fustakrakich]

What does it all do that it needs that many lines of code, anyway?

Well, I hope it can at least print "hello world" or display the time on the screen. Whatever it does, the kludge makes it look fat. I'll bet ya that if it were written in Assembly it would fit on a floppy

License?

[willoughby]

What license is it released under?

Re:License?

[fustakrakich]

Your guess is as good as mine.. Can it be more ambiguous?

AT&T is currently working with the Linux Foundation on the structure for this open source initiative.

Re:License?

[HiThere]

An excellent question. If AT&T has any sense it's the GPL, as it retains ownership of the copyright, but the article didn't even use the word license.

That said, I believe that the "Linux Foundation" is an association of companies with no community representatives. They often makes statements that try to convince people that the Linux community feels in some particular way, but it's frequently a way that only benefits the corporate interests.

Symbolicly significant

[inode_buddha]

This is very significant in its symbolism actually -- remember, it was ATT/Bell Labs that invented C and UNIX and used it internally for their systems management. So for them to open this to the Linux Foundation could be seen as a strong endorsement.

Re:Symbolicly significant

[Anonymous Coward]

> remember, it was ATT/Bell Labs that invented C and UNIX

No, it was Dennis Ritchie who invented C, and also Unix along with Ken Thompson. I would say they invented it *despite* ATT.
Big corporations never invent works of genius, individual employees do. All works of genius are creative acts by individuals. What corporations do at most is to try *kill* employee creativity by planning incremental results every quarter.

Re:Symbolicly significant

[evilviper]

The AT&T of today is just an empty brand name that SBC was able to purchase for a pittance, as the withered husk of the old AT&T was practically gone.

There's no significance in buying a Polaroid DVD player, either.

Re:Symbolicly significant

[frankenheinz]

*This* is not *that* AT&T. (Two different companies entirely.)

Re: Why do companies do this?

[The Other White Meat]

That hasn't been true since the 1970s.

I would like to build a community too

[Anonymous Coward]

where we can all watch AT&T and Comcast and the rest of them go fuck themselves

The kernel -includes- SDN in 98% fewer lines

[raymorris]

Kernel programmer here (though I barely qualify for the title).

The stock kernel includes all the basics of software-defined networking - bridging, firewalls, etc. Obviously that's just a small percentage of the kernel, less than 1% of the kernel's 15 million lines. Which means the kernel does basic SDN in no more than about 150,000 lines. Why AT&T needed to add 50 times as much is a legitimate question. I wouldn't be at all surprised if, like many software projects, they used ten times as much code as needed for the job, resulting in ten times as many bugs.

Re:The kernel -includes- SDN in 98% fewer lines

[HiThere]

That's possible, but my guess would be that they implemented a large number of features that the kernel didn't even contemplate (and would insist on relegating to user space). Feature creep is a constant problem.

Another thing is, AT&T probably didn't HAVE user space separate from the kernel, as the entire application was intended as a software defined network.

Re: The kernel -includes- SDN in 98% fewer lines

[Anonymous Coward]

Too much latency in the stock kernel, that is why we have Netmap, DPDK, PF-Ring, etc.

Yes, and so what you use becomes the standard

[raymorris]

> Is it to get people to write code for them for free?

Yes. AT&T and BT both need much of the same code. It's inefficient for them to both write separate, incompatible implementations. Better for AT&T to share the initial code base, then when BT makes improvements or adds things, AT&T can use BT's work too.

Also, it's entirely possible that if AT&T kept their code proprietary, the industry as a whole would largely standardize on some OTHER code base. Then AT&T would be stuck with a huge mass of outdated, incompatible code. By open sourcing a reasonably good system, it encourages others in the industry to use it, so that the industry tends to standardize on what AT&T is already invested in.

Seems like communism

[dhaen]

All these people working for the common good ;-)

Re:Why do companies do this?

[Tailhook]

It makes the industry more liquid. The investors that hold these companies don't just hold AT&T. They also invest in Verizon and Deutsche Telekom and a plethora of smaller telecoms. The more standardized the systems are among these companies the easier it is to merge and consolidate them. That has real value for major share holders. The end state for a successful telecom company is getting bought by some politically connected monster.

Why ATT really wants this

[knorthern knight]

Today's situation... ATT has to pay through the nose to Cisco/Juniper etal for network switches, and beg and plead for the specs ATT wants.

ATT's goal... ATT gets to slap the code with the specs they want, into a glorified Raspberry Pi or Nvidia GPU, maybe as firmware. And Cisco/Juniper/etal stocks are going to collapse. ATT is not in the network switch/router business. But it is a major capital cost for them. If they can roll-their-own, or get Foxconn to build to their specs, ATT stands to save a bundle of money. Imagine you're a taxi company. You're not in the car manufacturing business. But if you could get Foxconn to custom-build taxis for you at a fraction of the cost of what Ford or GM charges, you'd jump on the deal right away.

Has an email client and lisp interpreter too...

[mattr]

Considering the code size it undoubtedly has many auxiliary features including the ability to read email. Considering the provenance it probably can read everyone else's email too.