'needs to analyse end user, not just data centre, performance'

The performance of has shown no signs of improvement despite the US government publishing a report saying that the website now had uptime of more than 90 percent, an analysis has found.

Experts at application performance company Compuware previously found that standard website design optimisation practices had not been applied to, and the company has been monitoring the site’s performance from an end-user perspective via a real-time app using its tools.

“In the last three days, the site has been down for two days,” Andreas Grabner, technology strategist for Compuware told ComputerworldUK.

“We did an analysis as an end user and we had three failed attempts [to get onto the website]. Whatever they claim is just not the truth.”

Grabner added that when he was able to finally access the site, performance was slow.

“In the latest measurement we took, only three out of 50 states have ‘acceptable’ performance [website response times] of four seconds or less,” he said. “The majority take longer than eight seconds, and even worse, most Americans can’t get on the site.”

The US Department of Health and Human Services (HHS) website, which launched on 1 October 2013, allows uninsured US citizens to shop for new health insurance plans. However, nine weeks on the website is still seeing outages and slow page loads, affecting users’ ability to complete applications.

This is despite the US government releasing an assessment report at the weekend, where it said it had installed new “technical monitoring instruments” that allows it to constantly analyse site performance in real time.

It said that over the last five weeks, more than 400 bug fixes and software improvements had been applied, hardware had been upgraded and 12 large dedicated servers had been deployed. The system response time had consequently improved from eight seconds in late October to under one second, and the system’s uptime was “consistently surpassing 90 percent”.

“Ninety-plus percent uptime is a strange figure to be quoting," said Grabner. "If you look at most commercial websites, they operate at 99 to 100 percent uptime. In fact, we benchmark uptime, or, as we call it, availability each month for sites in the retail sector and most operate at 99.9 or 100 percent availability. We also benchmark healthcare providers in the US and the top 21 sites have an uptime of over 99 percent.

"This is just part of the story though. Availability is a fairly basic measure. A site can be available, but if it takes 10 seconds to get to it it's fairly pointless. That’s why we also look at measures such as end user response time, which measures how long it take for a website to load for a user."

However, despite the report saying that the site was able to cope with 50,000 concurrent users, on 2 December the tech team behind the site decided to launch a queuing feature with just 35,000 concurrent users.

“Looking at the report, it seems that they’re just looking at the wrong end - they are just looking at data they capture in their data centre,” said Grabner.

“They look at things coming into the servers, not at the end user. If you look at website performance, you need to start at the browser, where the end user is.”

He suggested that the government’s assessment was limited in its scope.

“There’s a lot of stuff between the data centre and the browser. There’s network connectivity, CDN (Content Delivery Network), third party components. If they are not looking at those, they’re not looking at the full picture,” Grabner said.

Copyright © 2013 IDG Communications, Inc.

8 highly useful Slack bots for teams
Shop Tech Products at Amazon