Uh oh.  Is the site is down?

Yahoo! site inaccessible

Site downtime is rare these days, but it still happens, and when it does, thousands of people can be affected.  But how do you know that an entire web property is down, and that it’s not just down for you?  How can you figure out who’se affected by the outage?

I tried to load Yahoo! a few moments ago, but I was unable to do so.  To figure out if it was just my machine that had an issue, or if others were affected as well, I did the following:

First, I launched tests from webpagetest.org and Gomez’s free synthetic monitoring service called Gomez Info-Center (since they’re free tools, I used both to make sure that I was getting a second opinion).  I use these services to understand if I’m the only one experiencing problems with websites.  If a site is slow, or if I’m getting “Address Not Found”, it’s the first place that I’ll go if there’s nothing obviously wrong with my own Internet connection.

Here’s what webpagetest.org told me:

webpagetests.org results for Yahoo.com

Weird stuff!  At least I know I’m not the only one.  This was further confirmed by a few twits that I received confirming that people were experiencing mixed results when accessing Yahoo!.

Twitter response

So is it *just* Yahoo!, or might there be other sites affected as well?  I went over to Keynote’s Internet Health Report to have a look.

Keynote's Internet Health Report

I didn’t see anything yellow or red, indicating that all backbone providers seem fine.  So chances are that the problem is local to Yahoo!, or Yahoo!’s data center.  If I had to guess, I suspect that they had some momentary DNS issues, because sure enough .. after a few minutes, Yahoo! was back up and kickin’:

Yahoo! is alive!

Sure enough, as I was writing this blog post, an alert from the folks at AlertSite kindly informed me that something happened in Washington DC.

AlertSite notification - way to go, ServerBeach!

This isn’t really a post about Yahoo! being down (chances are that the Yahoo! data center closest to me was in Washington, the same data center that AlertSite was announcing availability issues with).

Incidently, I actually use Yahoo!’s main landing pages as a benchmark for success.  Their pages load quickly, and their availability (uptime) is generally excellent.

This post is about something much more important.  How can you know if your blog/site is slow, or even worse, simply not loading for others all around the world.

We’re currently in the process of getting the synthetic monitoring and the real-user monitoring chapters reviewed; they’re almost done!  They talk about two things:

1. Making sure your site is accessible from around the world
2. Making sure that your site is accessible for all of your users

Synthetic Monitoring

This technology checks your site for you at intervals (every minute, five minutes, fifteen minutes, hour, whatever), and gather all kinds accessibility (is the site up?) and performance (how fast did it load?) stats, from many different types of machines located all around the world.  You can tell synthetic monitoring technologies to alert you as soon as something happens that you don’t want (like a page not found or taking more than X seconds to load).

Thanks to the distributed nature of many synthetic monitoring platforms, you can also figure out if certain parts of the world are having more issues accessing your site than others (which is what I suspect happened in Yahoo!’s case – the west coast seemed fine, but those of us in the east seemed affected).  In fact, if Yahoo!’s downtime wasn’t expected, I’m positive that some form of automated alert informed them that they were experiencing issues.

Real User Monitoring

But sometimes, it’s not enough to test from a whole bunch of desktops around the world, because they don’t necessarily represent real traffic – the actual traffic that you have on your site.  That’s where RUM comes in.  RUM tools monitor every single transaction on your site, and tell you how fast (or slow) transactions were for your users.  RUM data is extremely insightful.  For example, you can figure out exactly how many users in NYC have page load times over 9 seconds.  Maybe you have a widget on it being served from Australia that is dramatically reducing performance for users in North America.

We show you what you need to know to implement these kinds of solution, and go into many of the pros and cons of these technologies; whether you’re a blogger or an SVP of Operations for a Fortune 500 company.

In the meantime, if you haven’t seen this video, load it next time your favorite website is down. http://www.thewebsiteisdown.com/

No related posts.