Blog Post

Yahoo Doesn't Get It

Yahoo, you have a $40.56 billion market cap and have been serving up web pages, ads and services for over a decade. I assume you get it when it comes to infrastructure. So why is it that you launch new services without proper capacity? Rafe Needleman says that Yahoo Live, your new streaming video service, kept crashing when the users numbered more than 800. This same thing happened when you launched Yahoo Pipes back in February. Maybe you guys should be buying backup capacity from Amazon?

15 Responses to “Yahoo Doesn't Get It”

  1. I agree with the actual post besides the last sentence: “Maybe you guys should be buying backup capacity from Amazon?”

    Just an FYI to Stacey, Amazon’s S3 service could not have helped Yahoo in this case. When it comes to live content you need a service that provides a ton of bandwidth for live content. This means you need a server that can process a server side language, which S3 does not provide.

    I don’t care what Yahoo’s reason is for allowing this service to fail. They have millions of users and they should properly test scalability prior to launching any service. There is a reason for beta testing a product on an invite only status. You test to see where the problems are.

    Someone at Yahoo really screwed up and launched this service prematurely. The service was not ready for prime time and that means someone screwed up.

  2. @ Ericson – your comparison of Adsense to Yahoo’s service is right on. I would add marketing to their problem – you know Yahoo has ads, but you don’t know what they’re called or how to find them. You don’t even know where to find them.

    Google makes it all easy.

  3. Why are they even calling it an experiment? Live streaming works and will likely be very popular. If anyone needs to get in early in a new technology it’s Yahoo. The guys working on the application have done a good job and gone some way to improving the current interface model for live video but someone higher up dropped the ball when they allocated the resources.

  4. smirnoff

    If this breaks at 800 users someone did not do their job. How can Yahoo! managers not properly estimate the number of users and plan to support that number at a minimum. If someone told me “It’s going to be a great product but we don’t expect more than 800 simultaneous users”, I would have kicked them out of my office. Get real.

  5. It does seem to me that Yahoo has not been able to scale as fundamentally as Google has.

    Sure, you can just throw a ton of boxes behind a load balancer, but then that will never scale to the gargantuan proportions in the way how Google does it.

    There are fundamental engineering differences. When you do a search at google, it goes through several steps in several large silos yet come out pretty quickly. They way how they’ve engineered each of those processing steps, I doubt Yahoo has anything like that as yet.

    It doesn’t show in their URL structures.

    I go to gmail, and I get something like this — consistently:

    I go to yahoo mail, and I get this. The first two segments will change.

    That URL tells me i’m on a single box (which I will remain on for the duration of my session). It could be a single small cluster as well. See my point?

    Some friends and I were talking about this sometime last year. Yahoo is a famously open company, and if they had something like this already existing, they would have talked about it. We would have read about it and have the papers to prove it.

    That’s why Google is so far ahead of all these guys.

    You don’t run AdSense (mind you, not Adwords) on small clusters of servers. AdSense is on millions of sites. You don’t run something like Google Analytics and its attendant nightly processing without pretty decent planning and engineering. That’s where Google engineering pays off with MapReduce, Distributed Storage, heck, even considering and planning for failure rates of hard drives.

    See some of the distributed computing papers here:

    And what did they do with advertising? Yahoo screwed up big-time. As an advertiser, I can go to google, signup, login, place my ads, and have them showing up on millions of AdSense sites, or google search in less than half an hour.

    That fact alone is why Google won the search advertising war. Not because the technology was any different. But because of the instant gratification and simplification factors.

    In response Yahoo spent years building the monstrosity that was Panama, and Microsoft built up their beast as well, that throws up HUGE blocks in the face of standard small business owners with 1-20 employees. Their systems are so complex that it takes a full time person or search management firm just to help you get started to place an ad and put up some keywords. They are fully oriented towards medium to large sized businesses — not smaller businesses that make up the large bulk of transactions on the web.

    Every way that they could bring in the long tail of businesses, they’ve opted to focus on just the top tier of medium to large sized business. With ridiculously large minimum spendings and so on.

    As a small business owner, there is NO WAY IN HELL, I could ever run a banner ad on Yahoo. I can go over the AdWords and get one running worldwide in 15 minutes.

    That’s the difference between the two companies. That’s the difference between success and failure.

  6. Agree w/ the comments. If you insist on being able to scale and have all the machines ready by first day, you’d spend a ton of time for an experiment. Getting the concept right requires being agile and iterating fast. Only after that you worry about scaling, which is a solvable problem.

  7. want_OM_back

    I understand where you are coming from, but as a developer I disagree with you. Yahoo may have infrastructure advantages from all the platforms they have build over the years but this project most certainly is using 3rd party servers like the flash media server (unless they build one on their own) and getting things right the 1st time doesn’t always happen in these cases. You can never predict where the bottleneck will occur, could they have done a better job….. yes, but in this world of hard deadlines, sh!t happens.

  8. Hey guys,

    thanks for your comments.

    The folks who i talk to at Yahoo Brickhouse and early stage projects always talk about their infrastructure advantage. Well, not so much. It is a different matter if a tiny start-up with little or no resources buckles under pressure, but this is Yahoo.

    Here is something to ponder about: Yahoo as a company has been building a big grid computing infrastructure and are basically one of the big guns of hosting and infrastructure. They talked about it in their recent quarterly conference call.

    I think it is safe to say that they don’t plan and launch properly services, specially more bandwidth intensive services. I think there is a serious disconnect somewhere.

  9. want_OM_back

    Stacey Doesn’t get it
    Do your research before you start smearing sh!t all over someone. Their blog clearly states that they are a small team almost like a startup and the point is to try out new ideas. Who cares if it scales the first day, look at twitter, gmail they still have issues yet they are tech darlings. Everyone just stop giving yahoo a hard time, every company goes through hard times. Yahoo! finally comes out with something decent and all I hear is it doesn’t scale blah blah, just STFU if you don’t have anything to write. OM we need to you back!

  10. It’s not the size, it’s how you use it. In other words, Yahoo has plenty of capacity and bandwidth, they just don’t seem to allocate enough to new projects and always come up with this problem with scale at time of launch.