Hyperic Gives hi5 What it Needs to Succeed

In February of 2006, after reading about Hyperic on a simple Google
search, the hi5 team signed up for an enterprise trial. They downloaded
Hyperic HQ directly from Hyperic’s site, and within minutes were using
the software to inventory hardware, software and services in the hi5
infrastructure.
Immediately the team recognized that Hyperic provided far more
coverage then its predecessors. The HQ product monitored both Resin
and PostgreSQL natively – two key technologies that Big Brother did
not recognize. Additionally, the team could now see the performance
heuristics of their infrastructure over time, which was critical to being
able to pinpoint the genesis of a problem. Since HQ was designed to be
extensible, the team was able to write additional metric collectors that
layered feature usage and business metrics on top of the performance
data. The hi5 website serves up a variety of media fi les including photos,
music and video – and the addition of these metrics enable the company
to correlate performance with a spike in not just user traffi c, but fi le
downloads of the hottest new tunes or raciest new video.

hi5 Networks Grows to Support World’s

hi5 is a global social networking site where more than 60 million members
come to stay in touch with friends, meet new people, create and explore
content, and express themselves. Since launching in 2004, the company
quickly became one of the fastest growing social networks, and the 10th
most highly-traffi cked website in the world according to Alexa.
With its 60 million members and 4.5 million unique visitors generating
over 200 million page views per day, constant uptime and high
performance are absolute musts for hi5.
“Our business, like many other online businesses, is based on our user’s
experiences. We make money off of paid advertising, and that advertising
is based on how many users it reaches. Uptime is critical,” said Akash Garg,
CTO of hi5. “If our service goes down, we can’t serve ads, and we risk
adversely aff ect our users’ opinion of our service, causing them to stop
coming back to the site. Losing advertising momentarily isn’t good, losing
users is far worse.”
In early 2006, hi5’s popularity started to really take off . The IT infrastructure
grew increasingly complicated as it scaled to support thousands of new
users each day. The potential for problems—downtime, and performance
issues—increased alongside the complexity. hi5 had been using a
combination of Big Brother and Nagios as their systems monitoring
software, which helped the company know when servers were down,
but lacked the historical charting perspective to tell them when server
performance was starting to degrade. Without the ability to trend
performance and use it as a baseline for an early warning system, hi5
couldn’t keep up with their rapid growth rate and risked costly outages.
They needed a new approach.