The engineering blog of Shutterstock, covering Perl, templating, traffic control, color search, and engineering culture.
HN links it for technical posts such as TCP slow start tuning and a charting tool; a commenter from Shutterstock also recommends RJMetrics for analytics.
If you need to analyze things that live in your database (users, transactions, actions on your site), I recommend RJMetrics.
I work for Shutterstock, a company that is now large enough to have gone through several phases of data maturity (we recently became a publicly traded company, which requires a high standard of data integrity and auditability).
We use a variety of tools to look at data: Google Analytics for traffic, a home grown BI and A/B testing framework, Great Plains for financials, our own open source toolkits for client side event tracking and data visualization (http://bits.shutterstock.com/?p=277 | http://code.shutterstock.com/), and RJMetrics to track and explore KPIs for our smaller business units that don't have dedicated analyst teams.
RJMetrics makes it incredibly easy to explore and visualize data without having to spend dev time pulling different views. They also have native support for things like cohort analysis and customer lifetime value (which saves a TON of time in Excel). In addition to the dashboards they provide (that a non-technical person can change and improve) they also have regular automated emails that help you and the team feel the rhythm of your business.
Basically: unless you are ready to devote a team of BI analysts to your product, RJMetrics gets you 95% of what you need with 5% of the effort of rolling something on your own. You'd be crazy to try and replicate this kind of functionality at an early stage of your business.
cached copy of the post article (seems down now)
http://google.com/search?q=cache:http://blog.habets.pp.se/2011/10/Optimizing-TCP-slow-start
A little history for those wondering what this is about
http://blog.benstrong.com/2010/11/slow-start-follow-up.html
http://blog.benstrong.com/2010/11/google-and-microsoft-cheat-on-slow.html
http://lwn.net/Articles/427104/
http://bits.shutterstock.com/?p=74
http://research.google.com/pubs/pub36640.html
https://docs.google.com/gview?url=http://research.google.com/pubs/archive/36640.pdf
https://docs.google.com/gview?url=http://www.ietf.org/proceedings/78/slides/iccrg-7.pdf
Thanks for taking a look!
There's a little more info in our recent blog post about this too: http://bits.shutterstock.com/?p=103
In particular, here's an example where you can switch between stacked, percentage, and line views; toggle on and off each data series; re-order the stack with drag-and-drop; zoom in to a time slice; add smoothing; add annotations, etc:
http://shutterstock.github.com/rickshaw/examples/extensions.html
It can make a difference, having this functionality around when you're trying to dig in and read your data.
Yes, tons of credit due there. In my post (http://bits.shutterstock.com/?p=74) I linked to Jonathan Corbet's coverage (http://lwn.net/Articles/427104/) of the topic on LWN, which is also a good jumping-off point for reading on the topic.