Cloud // Infrastructure as a Service
Commentary
1/31/2014
09:26 AM
Charles Babcock
Charles Babcock
Commentary
Connect Directly
Twitter
RSS
E-Mail
50%
50%

Open Hardware Is Like Linux: True Or False?

Examine this analogy closely. Open-source hardware and open-source software involve different processes with different levels of user participation.

The analogy of how Open Compute is similar to Linux was made repeatedly at the Open Compute Summit V this week. The analogy is fair -- open-source hardware shares many underlying values with open-source software -- but I found myself curiously disagreeing with the statement each time I heard it.

The Open Compute Project (OCP) is a bold initiative to put hardware designs into the public sphere and let many parties use them. Collaborative groups have formed to specify what they want in an OCP-certified server, storage device, or datacenter switch, giving hardware manufacturers the option to choose to produce it or not. The goal is to reduce vendor lock-in, put more power into the hands of end users, and standardize key pieces of hardware in the datacenter to create more interchangeable parts.

These are worthy goals, ones that potentially overturn many of the established ways of producing data center hardware. So, why, after spending two days in San Jose at Open Compute Summit V, am I thinking "but... but..." as Facebook's Frank Frankovsky and Nick Corddry assert that Open Compute is just like Linux?

References to Linux come up naturally because it is one of the most successful, sustained, and adopted open-source software projects. New releases of the Linux kernel now appear every 70 days. Each contains up to 10,000 updates and patches, a rate of change that equals 7.14 an hour. Linux's fame rests not on the fact that it's frequently modified. Rather, it's frequently modified and also respected as having a long-term future in the enterprise datacenter. The way things are shaping up, it also very likely has a permanent place in cloud architectures.

[Want to learn more about complex Linux kernel development? See Linux Kernel Development Gets An Early Bug Fix Stage.]

The most comparable effort might be the thousands of programmers that Microsoft has working on Windows Server, which also appears poised for a sustained run as a datacenter operating system. But even it can't compare to Linux, which periodically integrates thousands of new contributors and accepts code from hundreds of new and casually appearing, then disappearing, contributors. From the start of 2012 until the end of 2013, Linux incorporated changes from 1,100 contributors who worked at 225 different companies. And of course the results of all this work is freely given away on faith that working on Linux will be reward enough for continuing legions of programmers. That's never been true for Windows.

And in a way, it's also not true of the Open Compute Project.

The Open Compute designs are freely available, but someone still has to produce the hardware resulting from the design. There has to be a price tag attached to that hardware, even if it's an OCP design. And once a design is set, it would unwise for the end user to fiddle with it too much. An end user who wanted to substitute one component for another or perhaps give a component a different type of connector would have to convince the manufacturer to change a production line. The price tag would have to be adjusted accordingly -- and significantly.

This is another way of saying hardware isn't software. The free-wheeling nature of Linux in its early days attracted developers from around the world. It always seemed to me the Scandanavian countries had a share of contributors out of proportion to their population, although all their work on MySQL may also have also colored my view.

Also, a talented high school student can't download the bits of Open Compute design and do anything with it. In the world today, high school students are routinely learning to program with the code they've downloaded from a favorite open-source site.

By its nature, Open Compute is an organization of hardware interests and their auxiliary participants: datacenter builders, component suppliers, integrators, and custom builders. There were users at Open Compute as well, but they needed to be large, well-heeled users with an ability to order thousands of copies of a design they've influenced. And they were. Goldman Sachs, Bank of America, and Fidelity Investments are giving Open Compute a breath of life that it needs and the user influence it must try to expand.

To compare this process to the frenetic, churning, but somehow managed process of building the Linux operating system is still a stretch. It's unfair to compare the one to the other, when Open Compute at three years old is still in its infancy. But hardware is hardware and software is different than hardware. Until we can see better how open-source hardware will work, best be wary of comparisons to Linux.

Find out how a government program is putting cloud computing on the fast track to better security. Also in the Cloud Security issue of InformationWeek Government: Defense CIO Teri Takai on why FedRAMP helps everyone.

Charles Babcock is an editor-at-large for InformationWeek, having joined the publication in 2003. He is the former editor-in-chief of Digital News, former software editor of Computerworld and former technology editor of Interactive Week. He is a graduate of Syracuse ... View Full Bio

Comment  | 
Print  | 
More Insights
Comments
Newest First  |  Oldest First  |  Threaded View
Page 1 / 2   >   >>
Lorna Garey
50%
50%
Lorna Garey,
User Rank: Author
2/3/2014 | 9:45:27 AM
Re: WIld card
@jgherbert - IT at some point has to become like more mature engineering disciplines, where a standard is the default and proprietary and specialized options have to be justified by offering an amount of benefit in line with not only the upfront but ongoing cost. You don't see an architectural engineering team deciding that 1,500 interior doors in an office high-rise won't be a standard opening. They might, however, decide that it's worthwhile to make the front door of the building custom.

Why should a datacenter be different? The basic building blocks - servers, storage, networking - should default to being cookie cutter. IT should think carefully about where it's worth adding proprietary or in-house-developed elements. 

We recently did a survey on exactly this topic: http://reports.informationweek.com/abstract/6/10096/Data-Center/Research:-Data-Center-Debate:-Standardization-vs.-Specialization.html

 
Li Tan
50%
50%
Li Tan,
User Rank: Ninja
2/3/2014 | 1:48:29 AM
Re: WIld card
I agree - the hardware will take longer time to morph than hardware. You cannot simply ask your customer to abolish the old and legacy hardware in one night. There is ROI and payback period issue for hardware investment. The connector is especially a difficult thing to handle - the interface of various hardware pieces will not get fit to each other without step-by-step evolution. The Open Hardware is a good idea but it won't prevail in the short term. 
MarkHaysHarris
100%
0%
MarkHaysHarris,
User Rank: Apprentice
2/1/2014 | 10:07:18 PM
Open Hardware is Similar to . . .
. . . the FOSS movement which produced Gnu/Linux.

 

No, of course not, Open Hardware is not like Linux.  Seriously?

 

The Open Hardware movement is similar in many respects to the free and open source software movement that produced the Linux kernel and the GNU system. 

The point regarding vendor lock-in is paramount. The community at large is sick-n-tired of vendors like NVidia, Broadcom, and others, who without regard for the free development community to "lock-out" software or hardware driver developers from certain platforms (Gnu/Linux, &c) forcing "lock-in" to other platforms... the 21st century has no patience with this working into the future.

The second point that needs to be explored is how much should "we the users" have as input to the hardware development process?  At a bare minimum we need to know how the device works, and have clear and concise APIs for application and kernel development.  ---no more proprietary firmware (we want the source) and no more propietary kernel blobs... the frimware does not belong in the kernel, period. If its not in the device itself, put it in a file, and give us the source.

And the thrid point is that we need hardware standards.  Everyone is sick-n-tired of fourteen different wifi card|standards and several different video standards.  We need industry standard open wifi and open vidieo (audio). 

Open hardware is important, but its not like the Linux kernel.  Gnu/Linux is open,  and the hardware|firmware needs to be open too.  If that is what is meant by "like Linux," well, ok...

 

Cheers
anon6063359013
50%
50%
anon6063359013,
User Rank: Apprentice
2/1/2014 | 3:03:28 PM
Re: Hey, I'm a tech enthusiastic
Just wanted to comment that i would not use the phrase " The goal is to reduce vendor lock-in" but  insted at  would say "  The goal is to offer an alternative ... my point is that openess should not look like a war of hate against close sources , i believe that both open and close source can coexist and everyone is to make thier choices, im just saying on my broken english, bear with me please, im spanish speaking  
jagibbons
50%
50%
jagibbons,
User Rank: Ninja
2/1/2014 | 1:17:47 PM
Re: WIld card
Once you're down a road where all of your widgets have to fit one kind of connector, whether it was "open source" or not, there's now a cost to change the connector to adapt to a different design of a widget. Perhaps simplistic, but physical hardware connectivity doesn't seem as easy to modify and morph over time like software.
jgherbert
50%
50%
jgherbert,
User Rank: Ninja
2/1/2014 | 12:48:40 PM
Re: WIld card
And then there are standards created by Reference Designs, say. Build this and it will work for sure, but that doesn't make it the only way to build the system. Will you still see huge divergence for cost or performance gains, and thus move away from the very standardized architecture you wanted to create?
jagibbons
50%
50%
jagibbons,
User Rank: Ninja
2/1/2014 | 9:10:29 AM
Re: WIld card
Standards will always emerge, and hardware standards create lock-in in the absense of multiple competitors. Scale is also a challenge with hardware. Even if you had options to tinker with the parts, there are fixed costs in setting up manufacturing. Those components don't get cheaper without large production runs.
jgherbert
100%
0%
jgherbert,
User Rank: Ninja
1/31/2014 | 11:53:52 PM
Re: WIld card
@Lorna Garey: "It's never wise to underestimate how much engineers like standards."

 

Ooh, interesting. Standards or features? Cisco for example has for years made a habit of taking a standard and enhancing it for (as they see it) a better end result for their users. The standards often catch up later. Is there a danger that the same could end up happening here?

 
Lorna Garey
50%
50%
Lorna Garey,
User Rank: Author
1/31/2014 | 4:30:21 PM
Re: WIld card
It's never wise to underestimate how much engineers like standards.
Laurianne
50%
50%
Laurianne,
User Rank: Author
1/31/2014 | 3:54:38 PM
The download factor
Charlie is right, the ease with which students can donwload code and tinker with Linux is another ballgame compared to designing data center hardware.
Page 1 / 2   >   >>
Multicloud Infrastructure & Application Management
Multicloud Infrastructure & Application Management
Enterprise cloud adoption has evolved to the point where hybrid public/private cloud designs and use of multiple providers is common. Who among us has mastered provisioning resources in different clouds; allocating the right resources to each application; assigning applications to the "best" cloud provider based on performance or reliability requirements.
Register for InformationWeek Newsletters
White Papers
Current Issue
InformationWeek Tech Digest - August 20, 2014
CIOs need people who know the ins and outs of cloud software stacks and security, and, most of all, can break through cultural resistance.
Flash Poll
Video
Slideshows
Twitter Feed
InformationWeek Radio
Sponsored Live Streaming Video
Everything You've Been Told About Mobility Is Wrong
Attend this video symposium with Sean Wisdom, Global Director of Mobility Solutions, and learn about how you can harness powerful new products to mobilize your business potential.