NFV (Network functions virtualization)

Standards Lose Steam as Software Dominates

SANTA CLARA, Calif. — Open Daylight Summit — As software ascends in importance throughout telecom, the emphasis on standards is going to wane, according to much of the discussion here this week.

Multiple speakers have actually emphasized the dangers of rushing to harden standards too early, saying service providers and enterprises are better served by using open-source software and systems and maintaining their flexibility to support whatever applications come down the pipe. The traditional standards process involving consensus and compromise among competing interests can do a disservice to innovation, or so the thinking here goes.

In particular, the much-discussed northbound interface of software-defined networking (SDN) is something no one seems to be in a rush to establish.

Dan Pitt, executive director of the Open Networking Foundation , told a standing-room only keynote crowd that his organization was being pushed to create an NBI standard two-and-a-half years ago, and if they'd given into that pressure, "we'd have gotten it wrong."

An entire panel devoted to the northbound interface spent a lively 50 minutes disagreeing on just about everything except the need to keep the interface from being set in stone.

In general, the networking industry needs to let go of what Guru Parulkar of Stanford University and the Open Networking Lab called its "obsession" with standards, and allow the ultimate choices to be determined by what works in the real world.

"We need to standardize as little as possible," he said. "You can't keep the same standards process in place once you become more software-based."

In his keynote, Ericsson AB (Nasdaq: ERIC)'s Erik Ekudden, VP and head of technology strategies, offered a somewhat less strident position, saying that traditional standards and de facto standards each serve their purpose and can be complementary.

What will become important, he stressed, is for the telecom industry to be able to offer other industries an easy way to access the network resources they need, and simply offering an applications programing interface (API) may not be enough, especially for businesses that aren't already working in the cloud and don't have on-board IT expertise.

Pitt reminded the crowd that his board is comprised of major telecom operators, and what they are focusing on is the art of the possible, not elegant approaches to building plug and play networks.

"We have to match what is needed with what is possible," he said. "We are working at solutions that succeed because the ultimate goal is commercial success."

The work of organizations such as OpenDaylight becomes more important to help match open-source solutions with what enterprises are looking for, added Nick Lippis of the Open Networking User Group.

And even on the service provider side, the open-source approach means everyone has the ability to reach out for help from the open-source community, said Christos Kolias, senior research scientist at Orange (NYSE: FTE) Silicon Valley.

"We can find out who has a solution, and pick the best one, without waiting for a committee to make up its mind," allowing for greater flexibility and more nimble networks, two of the primary goals of SDN and network functions virtualization, he said.

— Carol Wilson, Editor-at-Large, Light Reading

<<   <   Page 2 / 3   >   >>
Carol Wilson 2/5/2014 | 6:44:39 PM
Re: The Age of Software isn't the Age of Standards Penguins? Really? That's adorable.

I thought it was clear I was being sarcastic when I said everything magically works together in open source, but apparently not. Sorry.

To be fair to the folks here at the OpenDaylight Summit, they aren't disparaging standards so much as the standards process and particularly the standards process in the telecom network industry. The feeling is that process moves so slowly that by the time something gets standardized, the IT world has moved on.

And, as more than one person said yesterday, if you standardize something at the wrong time, you may actually stifle innovation and screw things up. 

But yes, open source builds on standards - Hydrogen, the OpenDaylight SDN controller, uses a whole bunch of standard stuff from IETF and others. 

The challenge seems to be to find a way to blend what's best about open source such as speed, flexibility and avoiding vendor lock-in, with what network operators need from standards, such as reliability and backward compatibility, where that makes sense. 

And, oh yeah, what Tom said. 
TomNolle 2/5/2014 | 6:40:14 PM
Re: The Age of Software isn't the Age of Standards You make a bunch of good points, Seven, as usual, but I have to make a few(!) gentle counterpoints.

First, what's at issue here is the carrier interface standards, not standards overall.  I don't disagree that there are standards in the software space, but I do disagree that the existence of software standards implies that standards for carrier interconnection are necessary.  What's necessary is interoperability, and that can be achieved more easily using APIs that can be transmuted, IMHO.

Second, I agree that open-source can be contaminated by whimsey or by malice.  So can commercial code.  Operators tell me that they WANT an open-source solution wherever possible and they're trying to figure out how to make it possible.  But I don't think open source is the perfect answer (even though I ran an open-source project in the carrier service layer).  I think that open interfaces are the answer, not standard ones.  However, right now we don't have any forum in which to advance "open" interfaces other than standards bodies (too slow, too controlled by vendors) and open source.  I'd love to have another choice.

chechaco 2/5/2014 | 6:26:50 PM
Re: The Age of Software isn't the Age of Standards Excellent! Just excellent.
chechaco 2/5/2014 | 6:24:32 PM
Re: The Age of Software isn't the Age of Standards Not. Using Open Source standard would only guarantee that products are inter-changeable, not that they interwork on their layer.
chechaco 2/5/2014 | 6:22:32 PM
Re: The Age of Software isn't the Age of Standards I've heard these comments time and time again. Mostly from people who never did any standards, nor tried to deploy or run heterogeneous netowrk.
chechaco 2/5/2014 | 6:20:53 PM
Re: The Age of Software isn't the Age of Standards The Open Source model is a standard, de-facto standard of itsown. Building a service layer on top of it certainly may lead to non-interoperable solutions that would be inter-changeable but not cooperative.

Standards, particularly at IETF, been concentrating on informational models for quite some time. And getting these models right, ensuring interoperability is, IMHO, in the interests of providers first and foremost. Vendors would more than happy to lock their customers into proprietary solutions. Why wouldn't they?
brookseven 2/5/2014 | 5:27:27 PM
Re: The Age of Software isn't the Age of Standards Okay wait - wait - wait.

There are plenty of standards in the OpenSource Software world.  XML, HTML5, etc. and that is the standard of interoperability.

Beyond that OpenSource creates a handful of eco-systems.  It creates LOTS of them and most of them die ugly deaths.  The rest thrive in their own little world and declare the others bad.  

Also, there is a real issue with the speed of API changing.  Imagine the API being rewritten on every major revision.  It HAPPENS.  Code breaks because of this.  So, you have to make a project to adopt the next version of the OpenSource you have.  In fact, you often have to make a choice to take the OpenSource and live with your current version forever or not.  And it is BAD if you happen to choose the wrong Opensource project.

What this creates is this huge amount of churn and foam.  You can get whole chunks of development, but don't assume that the next version will work for you (Yeah real funny the guys at OpenSuse who made it mandatory on a random boot up to have penguins to walk across the screen).

It is why we actually created our own Linux distribution and standard set of packages.  Did our own package signing (and we used RPMs not DEBs - but DEBs are fine).  Every time we updated a piece of Open Source we treated it like a product release and did a regression test.

So first, don't the Open Source crowd snow you into believing they don't adopt standards - They Do...mostly IETF standards.  Second, don't make an assumption that they think backward compatability like telecom folks do - They Don't.  Third, don't make an assumption that all Open Source succeeds - It Doesn't.  Fourth, don't make an assumption that if an Open Source element is key to your product that you won't contribute to it - You WILL.

Just so we are clear....suppose an Open Source developer though it was funny to put in an interface in an SDN controller that could say send out a joke of the day randomly.  What would you do?  If you think I was kidding above about the penguins http://forums.opensuse.org/showthread.php/492868-Skating-Penguins-A-December-1-Install-Goodie

Carol Wilson 2/5/2014 | 4:23:12 PM
Re: The Age of Software isn't the Age of Standards Well if every vendor uses open source, everything will just magically work together without standards, right?

Actually what I am hearing here is that, for example, instead of creating a standard for a northbound interface from an SDN controller into business apps and management systems, for example, there may be multiple interfaces, depending on what's needed.

Mitch Wagner 2/5/2014 | 4:02:34 PM
Re: The Age of Software isn't the Age of Standards Standards organizations provide grounds through which vendors can fight wars for control of customer premises. 

If there are no standards, how to users ensure that multiple vendors' products interoperate and can talk with each other?
Carol Wilson 2/5/2014 | 2:11:42 PM
Re: The Age of Software isn't the Age of Standards Tom, 

That's exactly what I'm hearing here  in Santa Clara. Standards mean less and time spent on them - especially early in the process - is being viewed as time wasted getting the "real" work done.

But at the OpenDaylight Summit, that's a point of view to be expected. I wonder if there are dissenters out there?
<<   <   Page 2 / 3   >   >>
Sign In