Beer 30 had quickly come and gone on a sunny Austin afternoon. After sitting in the uncomfortable seats, I was ready to get my drink on at the Rackspace party. If only I could get through the last question of the last panel of the SXSW day.
My mind wandered to wine bottles, thankfully allowed in the slow carb diet, when David Michaels snapped my attention back to the Scalability: Covering your Rear with a Good Backend Panel. “My favorite tool would be a software engineer with a strong grasp of computer science fundamentals and a deep empathy for user problems.”
That one statement catapulted the excellent panel into the “this is why I LOVE spending $1200 to attend SXSW” stratosphere. Moderated by the infamous Scobleizer, the panel profiled three distinct, successful consumer facing startups, Mint.com, Shopkick, and ImageShack.
David Michaels, director at Mint.com, Jack Levin, CEO of ImageShack (and Google’s 20th employee!), and Aaron Emigh, CTO of shopkick, were discussing the design requirements and tools for highly scalable architectures. My mind noticed several parallels to Tim Ferriss’s keynote promoting his latest book, The Four Hour Body: An Uncommon Guide to Rapid Fat Loss, Incredible Sex, and Becoming Superhuman.
A mouthful but what a title. As the panel started discussing infrastructure buying strategies, I realized they were prescribing a Minimum Effective Dose (MDE). Tim Ferriss references Dr. Doug McGuff who speaks extensively on MED in his book, Body by Science. Minimum Effective Dose is quite simply the smallest amount that will produce a desired outcome.
Anything more is wasted. In IT infrastructure acquisition, the panel recommended buying only the best hardware that meets your platform needs but to strategically defer purchasing excess capacity as the MDE. But, how do you know what those requirements are for your business? The panel agreed. You hire the right systems people.
The right people for the job will fundamentally understand the business and manage the metrics that matter. And that’s where the art in finding and evaluating talent will make or break the business.
At Mint.com, David stated a secure co-located data center to run a highly asynchronous set of transactions for scalability and performance was of utmost importance to ensure users financial data was not only secure but accessible. Whereas ImageShack hosts their own data center, purely an economical decision, to run their data center like its own ISP.
ImageShack transfers 20GBps of data at peak supporting 55 million unique visitors a month. The profiles of these two products necessitated entirely different but highly scalable architectures. These architecture decisions were made by talented systems people focused on optimizing the environment.
Whether you’re trying to become the next extra human through the 4 Hour Body or scale a large startup to 100,000 users overnight, measurement matters. As Tim mentions, there’s an expression in statistics, “All models are wrong but some are useful”.
Thank you George E. Box. In IT infrastructure management you can not get away from the 4 fundamental metrics, CPU, Memory, Disk I/O, and Network utilization. The team even scoffed at the idea that by deploying in a content delivery network (CDN), entrepreneurs could get by without proper systems administration teams.
Perhaps your business model is such that you could get away without a network engineer, sorry guys, but the role of the systems administration teams changes but does not go away as more applications are deployed in the cloud. If anything, they said, it gets more challenging due to abstracting the systems by cloud providers.
Choosing to deliver application services through a cloud platform abstracts the systems management. Instead of directly managing the servers and infrastructure, the systems administrators have to work very closely with their cloud partner to facilitate management tasks.
They lose some of the control and losing that control leads to a whole separate post about loss of visibility in the cloud and how to work well with cloud vendors to ensure you know what hardware you’re really getting. Ahem, Reddit. Levin used the Teoma story to illustrate how hiring the right systems people helped Google dominate the search engine market.
If you haven’t heard of Teoma, they were originally a startup out of Rutgers University founded by an eccentric, funny professor Apostolos Gerasoulis and colleagues. They had the right product teams. Their algorithms were better than Google at producing accurate search results. But Google excelled.
Sadly, Teoma’s search results, although more accurate, were seconds to minutes slower. Teoma is now the search engine behind Ask.com after being purchased for $4.4 million in 2001. And Google dominated not for their superior technology but for their people dedicated to monitoring and tuning the environment to meet the demands of the business.
In a world where I’m inundated with employment numbers, outsourcing, and cutbacks, it was refreshing to see that SXSW people understand the right people enabled with the right metrics are the cornerstone for success.
Share This Post
- Houston, We Have a…Connection! They needed more than a typical monitoring tool. ScienceLogic was able to establish an effective IT monitoring solution between Earth and a space station. Keep reading…
- VMware’s Future: Cloud-native Applications and Containers What does the future of VMware look like? We sifted through all the marketing hype presented at VMworld to get to the core of VMware's vision for road ahead. Keep reading…
- MSPs: What Will Your Next Revenue Stream Be? Don’t look back on the revenue stream that could have been. Partner with MSP JumpStart to define, deploy, market, and sell new managed services your competitors cannot offer. Keep reading…