Gmail FAIL - What it says about Google and email in general

Gmail FAIL - What it says about Google and email in general

Summary: Following a two-hour outage yesterday, Google has now managed to get its Gmail service up and running. This outage follows two other well-publicized outages in February and May of this year. What do these outages say about Google, and email in general?

SHARE:

Following a two-hour outage yesterday, Google has now managed to get its Gmail service up and running. This outage follows two other well-publicized outages in February and May of this year. What do these outages say about Google, and email in general?

Well, the most apparent conclusion that can be drawn in that Google is not infallible. The company is like every other and is prone to failures. And let's remember that Gmail is primarily a free service, so we as users get what we pay for. This outage has, ironically, been put down to changes to the request routers that direct queries to the service's web servers, changes that were meant to improve service. I doubt that if we all paid Google for the pleasure of using Gmail that you could eliminate downtime totally. Someone made a change, things went wrong, everything came crashing down. It happens.

But the fact that downtime is an inevitable side effect of relying on any technology doesn't mean that every time Gmail suffers downtime, decision makers think twice about turning to Google for web services, especially services that they rely on like email. Email is a critical business service, and downtime not only frustrates, it costs money.

Is it realistic to expect 100% uptime? After all, email is something that's been around for decades. Seriously, no, it isn't, for the reasons I mentioned earlier, but it still doesn't mean that people don't demand it. As Gmail continues to grow, the seriousness of each outage period will grow. That said, even though expecting 100% reliability is unrealistic, it's in Google's interests to improve reliability and add measures to prevent the entire email system from collapsing.

Outages such as this also paint Google's OS aspirations in a different light. Would a Chrome OS put increasing pressure on Google's infrastructure and make outages more commonplace? Will users put up with this?

How did the Google Gmail outage affect you? Do you think that it's time for a company to offer 100% uptime guarantee? Do Gmail outages make you suspicious of relying on Google?

Topics: Cloud, Browser, Collaboration, Google, Outage

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.

Talkback

138 comments
Log in or register to join the discussion
  • A little perspective...

    2 hours of downtime since May. That's at least 3 months of uptime if the May outage happened to fall on the last day of the month. Applying a little math, that tells me that GMail has had an uptime percentage of 99.999%. That's the ever elusive 5 9s many companies can't seem to get on their own internal networks. If users are complaining about that kind of uptime then their expectations are completely unrealistic.
    jasonp9
    • How did you calculate that?

      Here's how I calculate it: three months ago would
      be 92 days, or 2208 hours. 2206 hours of uptime
      would be 2206 / 2208 * 100 = 99.91% uptime, or 3
      9's. I am assuming a 24x7 uptime requirement,
      which seems reasonable for a global mail system
      like gmail.
      JMichaelH
      • Sure...

        Sure...one of those calculations where it depends on perspective. We both got to 2208 in the same manner. I divided 2 by 2208 to get the percentage of downtime (9.057971 x 10^-4). What's left should be logically be the percentage of uptime, so I subtracted that number from 100 and got 99.999094.
        jasonp9
        • bad math

          So why did you subtract from 100? Try subtracting from 1, since perfect up-time would be 2208 divided by 2208. Apples to apples.

          Or, to make it simple, 2206 hours uptime divided by 2008 hours elapsed time in the same period. Three 9's, not five.

          When I worked for Nortel, I ran across the reliability requirements for one of their networking devices for the telephone system. The requirement was no more than 30 minutes of down time over 40 years. Now that is reliability.
          diane wilson
          • they like to dream

            bet they never reached it.
            KrazdKiller
          • Oops...

            you are absolutely right. My brain must have taken the morning off.
            jasonp9
          • We all goof occasionally!

            Here is the algebra behind this:

            (x-2)/x = (x/x) - (2/x) = 1-(2/x) = 0.99909420

            x is the hours of uptime, which in this case is 2208 hours.

            0.99909420 = 99.909420% uptime.

            [b]Oh well, I still think google is doing a great job with gmail. And I am a retired microsoftie. It is much better than hotmail, IMHO.[/b]
            jerryz58
          • bad math

            <When I worked for Nortel, I ran across the reliability requirements for one of their networking devices for the telephone system.>

            Looking back at Nortel, reliability wasn't the answer.

            If all this is being done on company time it is no wonder systems have problems.
            781lc
      • how many days?

        and how many days prior to that did gmail have down time?
        KrazdKiller
      • Reliability is generally calculated yearly...

        ... and not monthly or quarterly....

        So far Gmail IS NOT having 9,999 uptime if we take into account the entire year (so far).

        Be fair and calculate uptime the way is calculated in the industry.

        Regards,
        andresvettori
    • 99.9999%

      most companies would be happy with 99.5% and usually end up with around 96.5%. the only reason to make a big deal out of nothing is to either try to make a point for a paid service, or just to generate news where there is no news.
      KrazdKiller
      • 99.9% is not good enough

        99.9% is not good enough.

        If it was good enough two million documents would be lost by the IRS this year. Also, the IRS will incorrectly process 103,260 returns this year.

        What is the actual percentage?:

        For you photo buffs, 811,000 faulty rolls of 35-mm film would be loaded this year.

        In the next 60 minutes, 22,000 checks will be deducted from the wrong bank account.

        Every minute 1,314 telephone calls would be misdirected by telecommunication services.

        Every day 12 babies would be given to the wrong parents.

        As many as 268,500 defective tires would be shipped in the next year.

        Two airplane landings daily at O'Hare International Airport would be unsafe.

        In the next hour, 18,322 pieces of mail would be mishandled.

        In the next year, 55 malfunctioning automatic teller machines would be installed.

        This year, 291 pacemaker operations would be incorrectly performed.

        This is why my company has a 100% SLA.
        bb_apptix
        • Those numbers seem about right...

          Having bought a few bad rolls of 35-mm film, seen bank statements that read like a work of fiction and known several people who work for the the Bells, none of these would really surprise me. I'd venture to say the mail number is actually much higher. 55 malfunctioning ATM machines seems to be way low as well...there's somebody working on the one at my bank every couple of months. Having a 100% SLA doesn't mean you don't have problems...just that you'll fix them when they happen. Nobody, and I do mean nobody, delivers 100% good product. I also work for a company that has several 100% SLA contracts. Product gets returned. We fix it or replace it. 100% SLA is completely unrelated to actual performance of the product.
          jasonp9
      • Five 9's

        This speaks to the need of real supported paid service in mission critical environments where 5 9's are required. I can see Google failing big time there, especially that they do not provide any SLA's and do not even have a support team in place.

        If I am a school system and my email depends on Gmail only to find it is out when I need it most (such as needing to communicate to parents a snow day!), then this system has certainly failed me.

        Afterall, you get what you paid for.. You want free service with no support, you have to live with issues of downtime without being able to do anything about it other than [i]pray[/i] that it won't last long. Expect such episodes to increase as the use of Gmail and Google Apps increases going forward.
        samofdetroit
    • A LITTLE PERSPECTIVE

      I think you have summed it up perfectly and any further discussion is
      irrelevant.
      ucdailoi
  • Geographically isolated redundant system anyone?

    Putting all your eggs in one basket may eliminate the cost of extra baskets, but ...
    kd5auq
    • This is exactly what we should expect ...

      ... from Google! Two periods downtime involving hours in less than six months is pretty dreadful.
      M Wagner
    • Wouldnt help

      1) A Maintenance error would have resulted in both services having an error.

      2) Its very difficult to handle seperate Geographic points without delays. Eg if you create gmail account or message it may take an hour before it works on the other side of the world.
      bklooste
  • Lets see........

    In the 12+ years that I have had my personal AOL email account I have not suffered any problems with getting my email. They may have had downtime, but I have never experienced it. GMail and Blackberry always seem to have downtime every few months. I have a gmail, but only for google voice and really wouldn't rely on it for my business email for sure.
    OhTheHumanity
    • The only BlackBerry downtime I have seen ...

      ... involved the carrier (often AT&T) or the BES service with my employer. I have NEVER experienced BlackBerry BIS downtime on VZW. Good point about AOL.
      M Wagner