Following the migration of a number of my websites from physical servers to virtualised hosting, many of the sites experienced worrying outages.  With no sense of the irregularity or duration of these outages (only the emotion-heavy complaints from clients), I went in search of a decent web server monitoring service that could shed some light on the actual downtime.

About six online services were extensively trialled, from which I subscribed to three premium/paid packages.  The following is an overview of my impressions of these services, the top 3 of which delivered reliable monitoring:

Overview of Top 3

The table embedded to the right (or PDF version) provides an overview of features and prices based on monitoring of two sites at 5 minute intervals and one at a 1 minute interval.  Wormly stands out in terms of value by the inclusion of free 1 minute sensors, however, generally, the user-pay model seems to work well in all cases.

All services offer varying email alternate details so that, for example, tech level contacts receive a higher granularity of report and frequency, than say business managers.  Each of the services offers a ‘expected string’ and response code option, however, none of these seems to work as expected - for this evaluation and simple ‘check for response code 200′ (found) was employed.

Whilst the focus of my trial was uptime, these services also provide useful additional features such as performance monitoring, email server availability and pingability (ahem).

Each of the above is explained in further detail below:

Wormly

This is a very good service with intelligent interfaces in both the web control panel and simple email reports.  Surprisingly, subject lines and sender email addresses are an important point of differentiation amongst the top 3 products - a simple sender ID and summary subject line is of high value in terms of a quick understanding of the problem server, the nature of the issue, the status and action that needs to be taken.

It’s likely that Wormly will be our ‘Editors Choice’.

In short: simple, yet cable

Site24×7

This service seems to be the most comprehensive, however, when stacked up against Wormly, it loses out slightly due to cost-blowouts arising from high frequency checks. The email reports are superb. The interface allows nomination of specific and additional monitoring locations (say, Australia and London - as shown below) whilst the “Status View” provides an informative snapshot of performance over a given period.

Notably, Site24×7 was the only company to respond when a preview of this comparison was emailed to all companies.  This is not to say that the others were not interested in feedback, however, the support experience with Site24×7 was positive.

Pros: Good service; excellent email reports; good support

Pingability

A good service, let down by a below average interface, fairly unrefined emails reports and confusing server check terminology options. If you chose to go with Pingability, you won’t go far wrong, however, you may find the service has a somewhat steep learning curve.  With a consolidated glossary, simplified interface and a review of email reporting, this service could easily rank with Wormly and Site24×7.

Cons: no raw data export; an ‘unrefined’ experience

Other services

Alteryx is a free service (with some premium options, such as Twitter alerts) which was brought to my attention by Gigaom.  Whilst it had some nice options (like emailing your webservice provider automatically) and easy to setup, it’s accuracy was woeful:- about 90 of the first 95 checks were false positives, all of which were on the second and third monitored sites suggesting to me that, along with some cues from the layout and interface, the service is more of an alpha than a “beta”.

Siteuptime might be a worthwhile service, however access to a decent set of features for evaluation let it down.  What probably wont change is the poor subject line information such as the over-simplistic “DOWN” and “OK” (without any indication of the website in question). The interface too was tired and tricky to navigate.

XPL’s service didn’t work at all and appears only to be a hook for their other services. Don’t waste your time.

Update 10th July: Pingdom wasn’t included in the round-up but gets some positive vibes from Read Write Web

Update 14th July: After trawling our server logs, it was discovered that Wormly, unlike Site24×7 and Pingabilty, does not identify their crawler/robot with a user-agent string (noted in the CS-User-Agent field in IIS logs).  This makes the exclusion of legitimate traffic from robot traffic difficult and has been raised with Wormly Support.

Update 15th July: Wormly Support replies: “You can define the user agent string to be whatever you’d like in the HTTP sensor setting” - [if that's what I think it is] this will prove a very nice feature - because it allows you to customise (or whitelabel) the service name. As the reply suggests “User Agent” strings are set on a per sensor basis, so if you’re account is monitoring say ten clients’ servers, your User Agent could be modified to read “Acme Websites Monitoring Robot”.