There’s a magic line of comedian Eddie Izzard’s in his Glorious tour about plane safety and the top-up valve on life jackets:
“… and they have these life jackets and … they’ve got a little pipe here for top-up. And I don’t know about you, but I don’t want top-up. I want stays-up!”
Here’s he is in a clip from YouTube. It’s about 1:07 minutes in:
As it is with planes, so it is with websites. Or so it should be. But sometimes, for all sorts of reasons, websites don’t always stay up. In my experience this is usually because the hosting company is doing server rebuilds or, less favourably, the quality of their service is taking a dip. No matter how carefully one selects a hosting service, things can and do go wrong. When that happens, I need to know.
Pingdom to the rescue
Pingdom is a service that provides website up-time and performance monitoring and I’ve had a subscription with them for some while now.
They allow me to attach a trace to any website and specify a range of notification options. Typically, I set them to test a website every five minutes. In rare cases where we’ve had particular issues, I increase this to every minute. I can then have them email me when a site drops and repeat that email for every 15 minutes of continuously dropped time. I then have them email me when the site comes back up.
At the end of each month, Pingdom send me a report summarising all the sites I’ve attached traces to, giving me stats on the total amount of downtime for that month, together with a percentage of up-time and the average speed with which each site responds.
Not only do I get a clear picture of which hosting companies I recommend for my clients are doing well and which aren’t, I can quote authoritative figures for a site’s performance. Here is an example report from December 2014:
|Pingdom monthly report for 2013-12-01 to 2-13-12-31|
|Site A||99.88%||0h 54m 58s||8||361 ms|
|Site B||99.88%||0h 53m 27s||31||325 ms|
|Site C||99.92%||0h 37m 00s||4||494 ms|
|Site D||99.93%||0h 32m 08s||15||675 ms|
|Site E||99.99%||0h 05m 48s||3||500 ms|
|Site F||100.00%||0h 02m 00s||1||403 ms|
|Totals||99.93%||3h 0m 05s||62||460 ms|
This isn’t necessarily a typical month. Where sites from a particular hosting company show up with relatively poor performance (as can occasionally be the case) then my recommendation for companies that host my client sites will change. With client businesses being reliant on their websites, Pingdom helps give me what I need to make informed decisions.
November 2014 update
As an indication of the value of these reports, during the course of 2014 there’s been some change of hosting companies for some of my clients based upon this sort of Pingdom feedback. Since these moves - and with new sites going to a different hosting company - here’s the new performance data. See the difference!
|Pingdom monthly report for 2014-11-01 to 2014-11-30|
|Site A||99.94%||0h 25m 00s||3||399 ms|
|Site B||99.97%||0h 14m 00s||2||175 ms|
|Site C||99.97%||0h 13m 00s||3||531 ms|
|Site D||99.98%||0h 09m 00s||2||176 ms|
|Site E||99.98%||0h 07m 00s||3||203 ms|
|Site F||100.00%||0h 00m 00s||0||218 ms|
|Site G||100.00%||0h 00m 00s||0||351 ms|