You are here

Server Architecture and Dashboard Improvements - Summer 2014

Over the past few months, a few improvements have been made to Server pages, as well as some major underlying infrastructure changes which have helped to dramatically reduce false positives.

More information about server outages

For every server in your account, data about which server reported an outage, and the reason for the outage (a 'status code'), are shown on your Server's main dashboard page, in the 'Outages' section. This should help you determine if your site may be having trouble only in a particular geographical region, or if your site is having a particular issue (like DNS trouble, or the page is returning a specific non-200 HTTP code).

Hopefully this extra data will be of use! We're working on integrating the data into email communications, and possibly SMS messages as well, so be on the lookout for that—or not, if you have 100.0% uptime!

General architecture improvements

Some users were reporting periods of frequent false-positives, where a server would reportedly go down-and-up many times in a short period of time. This was understandably very annoying, and was also extremely hard to reproduce, since the problem had to do with network connection issues between check servers (which are spread around the world, and among different hosting providers and networks!).

We have modified Server's check architecture so it will now perform redundant confirmation checks on servers that are in different areas than the original server that reported the outage, instead of confirming the outages from only one or two servers (potentially in the same region).

So far, it's been over two weeks since we deployed this improvement, and reports of false positives have decreased to zero!

As part of this work, we also did extra load testing to ensure that Server will be able to continue to quickly scale as our user accounts grow—the original version of Server could've only supported a few hundred customers at 10 minute checks, but the current architecture can support at least a thousand customers with 1 minute checks!

Thanks for using Server, and please contact support if you have any issues or questions!

This is a blog post from Server, a website and server uptime and performance monitoring service. Sign up for Server to know when your server is down.

Subscribe to RSS - Server Blog