Slashdot Mirror


How Facebook Is Eating the $140 Billion Hardware Market

mattydread23 writes: It started out as a controversial idea inside Facebook. In four short years, the Open Compute Project has turned the $141 billion data-center computer-hardware industry on its head. This is the comprehensive history of the project, including interviews with founder Jonathan Heiliger and members of the financial services industry who are already on board, plus a dismissal from Google's own data center guru Urs Holzle.

22 of 89 comments (clear)

  1. Congrats Mark Zuckerberg by binarylarry · · Score: 2

    I hear OCP is currently planning a move to Detroit to cut down on Silicon Valley overhead costs.

    --
    Mod me down, my New Earth Global Warmingist friends!
    1. Re:Congrats Mark Zuckerberg by phantomfive · · Score: 2

      I hear OCP is currently planning a move to Detroit

      I know you are joking, but I know of at least one SV company that is planning on doing exactly that (I don't expect it to turn out well, but I'm watching).

      --
      "First they came for the slanderers and i said nothing."
    2. Re:Congrats Mark Zuckerberg by Anonymous Coward · · Score: 5, Interesting

      I work in Detroit as an embedded software engineer (although, I've picked up some solid hardware skills in the last few years). I have just under 10 years experience and I make a CH under $100k. Auto industry pays me well. I have a job offer in SoCal which I am considering. I have a detailed spreadsheet, and after rent for a smaller place (I'd be selling a home to rent in CA) and taxes, I will have about the same amount of money. Given the cost of living out there, I will have less money after bills out there. It's a pretty hard sell financially. But, the beach is free! Ditto with hiking.

      However, the problem with this job market, lucrative though it may be at the moment, is 2 fold: (1) A lot of the work is very boring. My skills are not growing at the pace they would in a faster paced business. And, boring is boring! (2) I have no idea what is going to happen to the American automakers. The industry is going to have one hell of a decade sooner rather than later. We have a business culture that's about 20 years out of date at least. My guess is we'll see even less vertical integration than we have now, and the traditional automakers will focus on what they do best: engines, materials, manufacturing, assembly, systems integration. I do not see a world where the automakers are going to catch up to the big boys with regards to electronics and software. They are just too far behind. We're already seeing Apple and Google invade the automotive electronics, which, IMO, is more of a vote of no confidence in the traditional auto industry's ability to make decent, useable consumer facing electronics. Engine controllers. We got that. Infotainment systems that don't make you want to drive off a bridge? Not so much. Sensors and automation? Maybe, but I don't see too much impressive at the deep technical level coming out of Detroit.

      It gets murkier with algorithmic driving. There are some big boys that will fare just fine (like Bosch), but there are just too many small players moving too fast. Detroit has the skill set to integrate all that high technology, but we're losing a war of time with regards to creating cutting edge tech. People might argue that we're in a tech bubble, but the internet of things is not purely hot wind. Embedded is growing REALLY fast, and it's going to continue at a very nice clip. I think we'll see more market fragmentation for a little while before a reversal with cross market, embedded technology consolidation on a totally new scale. For consumers, that might be a good thing (probably a mixed blessing). For my career, it might hurt to stick with the losing side. There's a reason embedded companies aren't building tech centers here in Detroit where its cheaper and we have a massive technological workforce: our industry (and therefore our labor force) has a complacent culture compared to many other places in the country. If I were going to invest, I'd be much happier to train enthusiastic self starters than hire from an existing talent pool with a mediocre culture.

  2. Re:Summary plz by Smallpond · · Score: 4, Interesting

    Open source design of data center hardware - power, compute nodes, storage, rack and cable layouts. The problem is that everyone's needs are slightly different.

  3. Re:Not relevant? by Cytotoxic · · Score: 4, Insightful

    The guy in charge of selling data center computing as a service thinks that most companies should buy their data center computing from a company like his instead of rolling their own.

    And this is surprising or controversial why?

    In other news, the guy from Cisco thinks that companies will be looking to Cisco for fast, stable networking. And the guy from Intel thinks that companies will be looking to Intel for power efficient data center solutions.

    This doesn't make them luddites.... it makes them salesmen.

  4. Re:1. FB wins. 2. Customers win by Anne+Thwacks · · Score: 2
    Open Hardware is not exactly a new concept. And the fact that it can have a huge influence with big players is not new.

    LTO - Linear Tape Open - has been a mega success in driving up cross manufacturer compatibility, and driving down the cost of tape backup. As a consequence, tape use has gone up. (Possibly assisted by increased amounts of data, and the fact that it is now obvious than not even NSA and GCHQ can keep their data secure "in the cloud").

    LTO is made by players like HP, IBM, Sony.

    --
    Sent from my ASR33 using ASCII
  5. FB & GOOG - different tiers by QuietLagoon · · Score: 4, Interesting
    From the looks of TFA, it seems that Facebook's direction is about commoditizing the hardware, and google's direction is about commoditizing the services.

    .
    That makes sense because Facebook's service requirements are not transportable to other industries, but Facebook's hardware needs may be.

    Meanwhile, google is providing services to companies, and is looking to make those services transportable.

    I wouldn't say that google is "dismissing" Facebook's strategy but instead, google is working a few levels above it.

  6. Re:Summary plz by Hognoxious · · Score: 4, Funny

    what IS "Open Compute Project"

    236 points when playing buzzword scrabble?

    --
    Confucius say, "Find worm in apple - bad. Find half a worm - worse."
  7. Re:explain to me how this threatens cisco? by Sique · · Score: 4, Informative

    Because there are also OCP network equipments, like a switch design from Facebook that lets you do software defined networks easily.

    --
    .sig: Sique *sigh*
  8. Misleading article by Anonymous Coward · · Score: 2, Insightful

    They compare free open source software which is a product itself with open source designs for hardware which are just specifications hardware is still not free

  9. Re:1. FB wins. 2. Customers win by Anonymous Coward · · Score: 2, Interesting

    There have always been industry consortia standardizing this or that, such as VGA, Ethernet and VESA bus back in the PC's pre-internet days.

    But this is different - it's an encompassing standard that basically says that if BrandX can claim that it followed OCP, their design is just as good as Lenovo, HP, or Dell. That is a big difference. Sure, quality control and service could be different, but is that worth paying a huge markup for? Maybe not, for many customers.

  10. Re:Summary plz by ArcadeMan · · Score: 4, Funny

    Given that this is Slashdot, I believe the appropriate reply to your question is "a Raspberry Pi Beowulf cluster, enclosed in a 3D-printed 1U rackmount unit with a cooling system controlled by an Arduino".

  11. Re:Not relevant? by ShanghaiBill · · Score: 4, Insightful

    Do people not see the connection between increases in privacy breeches and the moves to cloud systems?

    The correlation is negative. Cloud companies have better security than a typical small company trying to roll their own solution.

  12. Re:Summary plz by CaptainDork · · Score: 2, Funny

    Monitored by drones.

    --
    It little behooves the best of us to comment on the rest of us.
  13. Re:Not relevant? by CaptainDork · · Score: 2

    I don't disagree entirely.

    Clouds are a huge attack surface populated by some impressive names.

    Hell, the feds can't keep their doors shut.

    I do the best I can with my law firm in-house and I use best of breed off the shelf protection.

    That's my risk assessment.

    I don't sell widgets and it's already in the news that Bubba got in a car wreck and stuff.

    --
    It little behooves the best of us to comment on the rest of us.
  14. Re:Summary plz by KGIII · · Score: 3, Funny

    And not employing enough females and minorities.

    --
    "So long and thanks for all the fish."
  15. Re:Not relevant? by bidule · · Score: 3, Funny

    Do people not see the connection between increases in privacy breeches and the moves to cloud systems?

    Well, I'm not wearing any pants.

    --
    ID: the nose did not occur naturally, how would we wear glasses otherwise? (apologies to Voltaire)
  16. Re:Summary plz by KGIII · · Score: 2

    Do not worry. Someone will be along shortly to tell you that it is attitudes like your's that prevent women from applying to the tech jobs in the first place. Me? I understand the difference between something said in jest here and something said in a work environment. Though I may have now preempted the aggressive comment(s).

    --
    "So long and thanks for all the fish."
  17. Re:Summary plz by Anonymous Coward · · Score: 2, Interesting

    The average person with a brain will realize that they are too stupid to trust with their own data. Only the elite 1% know what they are doing, the 99% just follow what they're told as "best practices" without realizing they're using the wrong best practices for the wrong thing because they don't understand the problem the best practice was meant to solve.

  18. Re:explain to me how this threatens cisco? by silas_moeckel · · Score: 2

    Hate to break it to ya software has always defined networks.

    There is a lot of noise about moving the management side out of the chassis. This does not magically make a L3 switch with a a 32k entry fib work as a core BGP router. In some ways it's nice you can stop paying your vendor for massively overpriced licences to turn on sunk cost features. It gets very scary when they want to throw boxes all over the place but centralize management. It realy seems like an excuse to keep putting in pitifully small CPU on L3 devices where a few hundred bucks for a xeon to run the high level stuff overseeing a mips arm etc that's dealing with low level hardware.

    But I guess it's to be expected these were the same sort of people that though 2 switches in a stack were redundant.

    --
    No sir I dont like it.
  19. Re:Summary plz by CrankyFool · · Score: 2

    Not sure if trolling or just oblivious. This isn't about 'cloud' -- this is about "maybe, if you're going to build a large datacenter, you can not re-invent the wheel and use standardized DC components, courtesy of a ton of research work FB did to optimize the BOM."

  20. Re:Not relevant? by ShanghaiBill · · Score: 2

    Cloud companies will provide the cheapest possible security they can get away with

    Cloud companies have invested billions in data centers. They are not going to skimp on security, risk a big breach, and see all of that investment go down the drain along with their reputation. They have a very strong incentive to provide good security.