Follow Slashdot blog updates by subscribing to our blog RSS feed

 



Forgot your password?
typodupeerror
×
Businesses The Internet IT

Amazon.com Suffers Outage: Nearly $5M Down the Drain? 173

First time accepted submitter Brandon Butler writes "Amazon.com, the multi-billion online retail website, experienced an outage of unknown proportions on Thursday afternoon. Rumblings of an Amazon.com outage began popping up on Twitter at about 2:40 PM ET. Multiple attempts to access the site around 3:15 PM ET on Thursday were met with the message: 'Http/1.1 Service Unavailable.' By 3:30 PM ET the site appeared to be back online for at least some users. How big of a deal is an hour-long Amazon outage? Amazon.com's latest earnings report showed that the company makes about $10.8 billion per quarter, or about $118 million per day and $4.9 million per hour." Update: 01/31 22:25 GMT by T : "Hackers claim credit."
This discussion has been archived. No new comments can be posted.

Amazon.com Suffers Outage: Nearly $5M Down the Drain?

Comments Filter:
  • Hmm... (Score:5, Insightful)

    by Anonymous Coward on Thursday January 31, 2013 @05:14PM (#42755911)

    So, if a website is down, and someone goes to buy something, that means they are unable to purchase it later when the site is back up?

    The logic behind how they arrived at that number is slightly flawed.

    • Re: (Score:3, Insightful)

      Well.. Timothy wrote the headline... what makes you think he has ever used logic? But yes, the amount "lost" is probably a small fraction.

      • Re:Hmm... (Score:5, Insightful)

        by icebike ( 68054 ) on Thursday January 31, 2013 @05:27PM (#42756095)

        But Brandon Butler wrote the summary, and lifted the calculations directly from Network World, so a lot of people just assume that people try exactly once, then give up and never return.

        Its totally silly of course, because Amazon often has the best prices, and people will simply wait. Nothing purchased on line constitutes an emergency to most people.

        The cost to Amazon is probably not really that great. There might be some hourly people sitting around doing nothing. Some network staff might have been called in to work overtime.

        But averaged over a month, I doubt there will be an actual drop in sales.

        • Re:Hmm... (Score:5, Insightful)

          by cayenne8 ( 626475 ) on Thursday January 31, 2013 @05:39PM (#42756267) Homepage Journal
          Yeah, but don't discount the number of people who "impulse buy"...and if given time to think about it...they may decide they don't really need it.

          With a company doing the volume of business that Amazon is doing, I have to think that would be significant with that segment of the market alone?

          • by sjames ( 1099 )

            On the other hand, some percentage of those impulse buys would have had enough buyer's remorse to cancel or return their order. That would have cost money to process.

          • by Genda ( 560240 )

            Or the folks who couldn't wait and went with eBay instead, or NewEgg or whatever. There are a dozen reasons why a one hour delay might have impacted whether a particular purchase happened or not. Still you wouldn't expect that to impact more than several percent, which if you think about it would still be a few hundred thousand dollars.

            One thing that might have had an even bigger impact, might be the fact that it would have impacted the ACTUAL cashflow today, even if the total sales is impacted only slightl

            • Or the folks who couldn't wait and went with eBay instead

              Couldn't wait and went with eBay? Haha...

              So you have to participate in an auction, wait until it is over (several days), and see your deal sniped away from you at the last minute while you're away. Ok, rinse, lathe, repeat.

              After 3 or more attempts (taking each a couple of days, due to the auction's durations), just pick a "buy now" offering that looks juicy.

              Unfortunately, the seller is an amateur who simply forgets to send the goods for two weeks. After reminder, he overnights it to you.

              But then, surpri

              • Ebay has several "Buy Now" products sold by vendors. Ebay is specifically trying to compete with Amazon. Several niche products are only available via Ebay "Buy Now" merchants.

                This isn't news to anyone but you, by the way.
                • And that is going to help me how, exactly?
                  • To understand your comment was wrong? Ebay is not simply about auctions anymore. They are trying to compete with Amazon. How could I be clearer?
                • by Skater ( 41976 )

                  Are you referring to the seller stores that they've had for years? I don't get how they're really any different than the rest of ebay, aside from the buy-it-now versus auction format. If that is what you're referring to, they have the same problem - we don't trust ebay for the issues ArsenneLupin mentioned.

                  Why would someone buy something off ebay that isn't an item that's no longer made or some cheap import (I bought some LEDs once that would fall into that category)? I'd never consider buying, say, a

                  • I wouldn't consider buying many items from Ebay either. However, his main point (topic sentence) was that having to wait several days for the auction to end negated the chance that someone would shop on Ebay during Amazon's outage. That is not a valid point, since (as you point out) Ebay has had seller stores for years.
        • Its totally silly of course, because Amazon often has the best prices, and people will simply wait. Nothing purchased on line constitutes an emergency to most people.

          Your definition of "emergency" and mine radically differ. I call it 'Tuesday' when there's five feet of fresh snow on the ground, over my house, my car, and I need to be to work in an hour. I'm more concerned about a broken coffee machine at work than some website going tits up for a few hours. Even if all the websites went tits up for a few hours, or days, it's not an emergency in my book. Emergency for me qualifies as "significant and immediate risk to life and safety," not "I can't order a copy of Call o

          • by icebike ( 68054 )

            Which is exactly why I said:

            Nothing purchased on line constitutes an emergency to most people.

        • Re:Hmm... (Score:5, Interesting)

          by eWarz ( 610883 ) on Thursday January 31, 2013 @07:18PM (#42757089)
          I happen to be IT for a large (nowhere near as large as amazon, but big enough...) eCommerce site and I can tell you, people don't buy later. A lot of those sales disappear. Users try to buy it on amazon, if that doesn't work they hit up buy.com or newegg or some other site.
          • by icebike ( 68054 )

            How the fuck could you POSSIBLY know this?

            Your site is down, you have no idea how many tried and went elsewhere.

            • by eWarz ( 610883 )
              Because we have tools like Google Analytics as well as hourly/daily/monthly revenue reports on a graph format. We know exactly WHO hits our site, when, where. Which days are the busiest, which are the slowest, etc. We even know your browser, screen resolution, OS, the color underwear you wear, etc.
              • by eWarz ( 610883 )
                To build on that, lets say you get 250 orders ON AVERAGE per hour during the waking 14 or so hours of the day. some hours you might get 240, others you might get 260, but your minimum never falls below 200. Each order has a dollar average of $100. That's $20,000 minimum per hour. If you never drop below 200 orders for a given hour, it's safe to say you've lost at least $20,000/hour for each hour you are down. Furthermore, lets say your busy time is 5pm-7pm. Lets say during that time you average 500 or
              • Re: (Score:2, Insightful)

                by rubycodez ( 864176 )

                you mean you are cluesless as to what those tools are really telling you. Amazon lost essentially zero, and yes most of the buyers came back later at their convenience.

                • Wouldn't it be a matter of looking at the averages for slightly longer periods? I.e. if the downtime was 1 day, just look at the volume of the week's starting with the down day for example. If this week had noticably less sales then any other week, he probably did lose sales (if people waited, they would probably have waited less than a week,). If the average was roughly the same, people would have delayed.

                  Of course, all depends on how steady the volumes usually are. If there are usually large swings from

              • by Rich0 ( 548339 )

                You don't have those figures for the times that your site was down. And I doubt your noise floor is so low that you could pick up the customers who couldn't buy something during that one hour stopping back over the next few days.

                Suppose that they'd normally sell $5M/hr. They're down for 15 minutes - that is about $1.25M in lost sales. People come back over the next 48 hours to make their purchases - that is an increase in net sales over that period of $26k/hr - or 0.5%. Are you really going to notice wh

              • Yes, we know what GA does. Which of those metrics gives you insight into how many customers go elsewhere during downtime?

              • We even know your browser, screen resolution, OS, the color underwear you wear,

                ... and then you wonder why people are so mistrusting of javascript...?

                (Technically, the browser and OS can be read from the user-agent header, and needs no javascript. But the screen resolution and color of underwear sure does need javascript!)

        • This. For me and mine, purchasing through a leviathan retailer like Solimoes gives a little bit of security in an insecure Networld. I am under the impression Amazon would make things right were I to be dissatisfied with a purchase, which has never happened when I order through them. And they have everything....it's like Willey's Store in Greenboro.
      • Comment removed based on user account deletion
    • No, people can purchase items once the site is back up, and I would agree that the number provided isn't accurate. But it may have still cost a lot of sales; some people will not go back to the site once it is up. They may buy from another site that is up (I've done this before when a retail site was down), go to a brick and mortar store, or just forget to go back later because what they were looking for wasn't terribly important.
      • Re:Hmm... (Score:5, Interesting)

        by xevioso ( 598654 ) on Thursday January 31, 2013 @05:29PM (#42756143)

        Some people will go back, some people will not. The point is there does not seem to be any indication that this number was taken into account when the 4.9$ million number was estimated. Perhaps the first hour it's offline they "lose" 4.9$million, but the next hour it comes back on they make 8$million when they normally would have only made 4.9$million. There's no way to know.

        • Indeed, Amazon should be able to use statistical methods to work out, within some reasonable margin of error, just what percentage of people did come back and which went to buy elsewhere. It would be interesting to see that figure, though I don't suppose they'll release it.

    • Re:Hmm... (Score:5, Insightful)

      by NatasRevol ( 731260 ) on Thursday January 31, 2013 @05:34PM (#42756211) Journal

      Not to mention that the $10.8B number is sales revenue, not earnings. So it's NOT how much they make in a quarter.

      They actually only earned $97M last quarter. Or $7.5M per week or $1M per day. Or $41,000 per hour.

    • by cod3r_ ( 2031620 )
      Exactly what I just thought lol.. They didn't make 4.9 that hour they made 9.8million the hour later lol.. All about averages.
    • The logic behind how they arrived at that number is slightly flawed.

      I'd say they used RIAA & MPAA math.

    • by MAXOMENOS ( 9802 )

      I'm in perfect agreement with this comment. Some lost sales are inevitable from this downtime, but the $4.9 million number is based on unwarranted assumptions.

      We'll know better what the impact is at their next quarterly statement (if they're honest)

    • Also, were all of Amazon's services down?

      The country-specific sites (e.g. Amazon.co.de)?
      Zappos.com?
      Audible.com?
      How about the cloud services (Amazon makes money on those, also)?

      Also, why should we believe Thursday afternoon at the end of January represents an average shopping hour for Amazon?

    • I was going to say the same, people merely placed there orders a couple hours later, not dropped the order altogether or decided to drop amazon altogether. Just a small glitch that won't even be noticed when their earnings come out. As pitiful as those earnings are, given the company's market cap

    • The actual cost is hard to calculate :
      Most people don't 'stumble' upon Amazon : they go there for a reason. So they will wait until the site is back up.
      People who did accidentally stumble upon Amazon ( just googling for a product ) , will see a site that doesn't work, and ignore it, so that might be a lost purchase.

      However, I think the biggest cost might come from the knowledge that Amazon got hacked : even if hackers didn't get anywhere near the users' credit card information, most people won't understand

    • So, if a website is down, and someone goes to buy something, that means they are unable to purchase it later when the site is back up?

      The logic behind how they arrived at that number is slightly flawed.

      The logic initially applied to airline reservation systems (Sabre, etc.). Apparently, in that industry, if a customer is unable to book a flight on one system (with one alliance), many will then just book with a competitor, rather than waiting and trying again at a later time... Especially if they aren't doing this from the comfort of their home, but from a travel agent's office, and might have to come back there later to try again.

      With amazon, this seems rather unlikely. People like to comparison shop, an

    • by bitt3n ( 941736 )

      So, if a website is down, and someone goes to buy something, that means they are unable to purchase it later when the site is back up?

      after I got the error message I just assumed Amazon had gone out of business. it's lucky I saw this article!

  • I'd hate to be the one developer responsible for that bug...

    • Re: (Score:2, Insightful)

      by Anonymous Coward

      The $5M figure assumes that 100% of the people that would have shopped at Amazon that hour purchased what they wanted elsewhere, rather than just trying back later, which I would bet a vast majority did.

    • Re:I'd hate to be... (Score:4, Informative)

      by Enderandrew ( 866215 ) <enderandrewNO@SPAMgmail.com> on Thursday January 31, 2013 @05:26PM (#42756089) Homepage Journal

      Don't work. Jeff Bezos is known as a very understanding, forgiving and easy to work for boss.

    • Actually, it seems it was a DDoS, as admitted by the people claiming responsibility:

      we used a 7kbotnet running hoic 100 threads each. 80servers in botnet and a 16gbps booter

      (From the update link in the summary: http://www.foxnews.com/tech/2013/01/31/amazoncom-website-offline/?test=latestnews [foxnews.com])

      • by penix1 ( 722987 )

        And you want to tell me why again ISPs can't cut off botnet infected machines and warn the customers to clean their crap before allowing them back on the net? If ISPs really cared so much about their networks as they claim they do (the reason they give for usage caps and throttling) then you would think they would want to rid themselves of useless and destructive network chatter such as infected machines. You can't tell me they can't detect a machine sending thousands of page requests a second. You can't te

        • Because that 7k spread across all the different ISPs and providers represents about .0001% of each ones bandwidth. It would require far more effort and resources then you can possibly believe. Now some of the ISPs have DPI and could do it, but I'd rather they drop the DPI and just give internet access.

  • by Scarletdown ( 886459 ) on Thursday January 31, 2013 @05:15PM (#42755935) Journal

    Sounds more like that was 5 million in potential dollars not earned, not 5 million lost. You can't lose what you do not yet have.

    • Re: (Score:2, Insightful)

      by Anonymous Coward

      Tell that to the RIAA and MPAA

    • by Anonymous Psychopath ( 18031 ) on Thursday January 31, 2013 @05:18PM (#42755967) Homepage

      Sounds more like that was 5 million in potential dollars not earned, not 5 million lost. You can't lose what you do not yet have.

      I'm glad you aren't in charge of payroll.

      • by s.petry ( 762400 )

        Because a company pays people with speculative or borrowed money does not mean that they should pay them on speculative borrowed money. Many of the financial practices we see being used today were illegal in the past. Of those that were not, taking loans to pay people was not a practice by healthy companies. It was done in desperate times sure, but today people think your crazy if you don't borrow money to pay your employees. That is beyond baffling. Not only do you pay your employee wages and taxes, b

        • To the person's point, did you read Amazon.com's latest earnings report showed that the company makes about $10.8 billion per quarter, or about $118 million per day and $4.9 million per hour."

          Another little point that may be of note here...

          If, when those quarterly reports come out it shows that Amazon had less than $10.8 billion in costs, then they made a profit and not a loss. Sure, it may have been 4.9 million less than expected, but is was still a profit and not a loss despite what the methods of Hollywood Accounting or Bistromathematics or whatever ethically challenged system they may use shows.

          • by s.petry ( 762400 )

            yeah, I actually missed that until I saw it pointed out. Makes the "issue" even more of a non-issue as you look at the claim in detail.

          • How does that logic hold?

            I lose my wallet in the street but because my income for the year is still net positive I didn't actually lose anything after all?

            Don't think so - a loss is a loss even is the long term impact is just less profit.

            Now - whether there was a $5M loss - that is a whole different question ;-)

      • Ah, but your paycheck is money owed to you. That does not work the same way.

    • by 54mc ( 897170 )

      You can't lose what you do not yet have.

      Unless you're the RIAA/MPAA

      • You can't lose what you do not yet have.

        Unless you're the RIAA/MPAA

        Yeah, I thought that when posting originally, but figured that goes without saying. And when something goes without saying, I don't say it (unless I am dealing with the Knights Who Until Recently Say Nee, then I will say it.)

  • Assumptions... (Score:5, Insightful)

    by Anubis350 ( 772791 ) on Thursday January 31, 2013 @05:20PM (#42756015)
    This assumes that amazon makes a constant amount every hour, as opposed to peak vs. off-peak business hours. This also assumes that the bulk of their business took their purchases elsewhere while Amazon was done, which I'm not inclined to believe is necessarily true.
    Amazon probably lost money, I'm in doubt that it's anywhere close to 5M
    • This also assumes that a person wanting to make a purchase on Amazon will not just wait an hour for Amazon to come back up, and will instead make the purchase elsewhere. In some cases that's probably true, but if it were me I'd probably just try again when Amazon came back up. I shop at Amazon out of laziness as much as anything else.
  • Chicken feed for Amazon.

  • Interesting statistic, though that assumes that a one hour wait is sufficient to make $5 million worth of sales redirect to a competitor. Amazon has some level of brand loyalty and reputation compared to others, and I'd bet that Amazon sales are not equally distributed throughout the day.

    I guess "Amazon might have suffered $5 million in losses" doesn't sound as interesting as claiming they actually did.

    • Probably not $5M, but for example I bought a SSD the other day, I shopped a few sites, Amazon and Newegg were two with the best prices in particular. Newegg also had a slightly better deal on another product I wanted too, I bought the SSD and other product from them. If Amazon wasn't up I wouldn't even have had the chance to comparison shop, Online competition is fierce between the large retailers.

  • to have a Sears catolog again.I could call and just place an order to a live operator..sigh.
    • to have a Sears catolog again.I could call and just place an order to a live operator..sigh.

      How long did you have to wait to talk to an operator? What if a bunch of modems automatically dialed their sales number and flooded their systems?

  • by sribe ( 304414 )

    There went 2013's profit!

  • by tscheez ( 71929 ) on Thursday January 31, 2013 @05:27PM (#42756101)

    They'll just stay open an extra hour to make up for it???

  • This assumes that people who couldn't order something from Amazon right at that moment will immediately go order it somewhere else. I'd bet that most people would just try again later and order what they want.
  • 404 Error? (Score:5, Funny)

    by dnahelicase ( 1594971 ) on Thursday January 31, 2013 @05:33PM (#42756179)
    Maybe Amazon should consider moving to the cloud...
  • by Anonymous Coward

    I was going to buy something during that period. I waited until Amazon came back, and bought it later. Problem solved. I'm not going to take my ecommerce business elsewhere because elsewhere doesn't have prime.

  • In what universe?

    Here's Amazon's latest financial release:
    http://phx.corporate-ir.net/phoenix.zhtml?c=176060&p=irol-newsArticle&ID=1779049&highlight= [corporate-ir.net]

    Here's the important bit:

    Net income decreased 45% to $97 million in the fourth quarter

    Amazon is barely breaking even. Whether or not this is an intentional strategy is another discussion, but it sure as hell ain't making 10.8 billion a quarter. It's not even making ONE HUNDREDTH of that.

  • People need to quit looking at whole numbers and think about this in real terms.

    One hour. That's all, they lost business for one hour, they're still up for many thousands (maybe millions) of other hours incurring the revenues consistently.

    As for everyone who says "but it's so much money!" you're missing the point, absolutely no reasonable business anywhere is spending so much on just running their business that one hour of lost revenues is actually going to cause so much as a blip on their books.

    Th
    • by Gertlex ( 722812 )

      People need to quit looking at whole numbers and think about this in real terms.

      Ok.

      One hour. That's all, they lost business for one hour, they're still up for many thousands (maybe millions) of other hours incurring the revenues consistently.

      Wut? Amazon hasn't been around for multiple centuries. If we're going to talk real terms, skip the hyperbole that results from vague statements :)

      Other than that, you're preaching to the choir, I'd say.

      • I said maybe millions because I didn't feel inclined to do the math to figure out what 5-10 years ranges in hours and took a wild ass stab heh, 5-10 years being my other wild-ass stab of guessing how long it's been since their last real outage having not followed. Wasn't trying to be hyperbolic, but since it turned out I was, maybe I can become a pundit or something, sweet!
  • by new death barbie ( 240326 ) on Thursday January 31, 2013 @06:02PM (#42756475)

    Good luck with that. First you have to pack it up in the ORIGINAL packaging, then fill out a Return For Credit form, and then wait at least 10 days for processing...

    THEN maybe you can claim your credit.

    • by pbjones ( 315127 )

      I hope they do better than I did, the cost of return mail was worth more than the book, bah.

    • by Kergan ( 780543 )

      Not to mention, adding to the funny aspect of the claim, the group deems itself as Nazis and got reported by Fox News.

      Conspiracy 101: people who believe in conspiracy theories are usually very, very, very, very, very thick people.

      • Usually, then again after reading about the money market manipulation scam posted on ./ yesterday, keeping an eye out for fraud and conspirators is a good idea.

  • "Amazon.com's latest earnings report showed that the company makes about $10.8 billion per quarter, or about $118 million per day and $4.9 million per hour."

    No they don't, Amazon makes barely any profit at all. They do have high (and growing) revenues though.

    • Net/Gross

      Amazon grosses close to $5M per hour. 'Makes' is a terrible term to use because it doesn't define if your making a profit or loss.

  • something and the web site I usually shop at is temporarily down rather than trying again later I go elsewhere or don't buy it at all.

    Oh shiny...

  • by afgam28 ( 48611 ) on Thursday January 31, 2013 @10:02PM (#42758023)

    calling bs on the hackers who claimed responsibility.

    http://gizmodo.com/5980618/amazon-is-down [gizmodo.com]

  • They must use EC2.

  • First line of TFS: "Amazon.com, the multi-billion online retail website".

    Oh, right, THAT Amazon. Thanks for avoiding any potential confusion with...well, nobody.

I THINK THEY SHOULD CONTINUE the policy of not giving a Nobel Prize for paneling. -- Jack Handley, The New Mexican, 1988.

Working...