Structure 09: How Some of the Busiest Sites Manage Their Web Ops

10 Comments

structure giants Web operation strategists at high-traffic sites such as MySpace (s nws), Yahoo (s yhoo) and Google (s goog) have to make hard-and-fast decisions on a daily basis to ensure their sites run at high speed and user data is stored securely. As the representatives of those companies, along with executives from Microsoft (s msft) and LinkedIn, can attest, no single solution exists to ensure some of the web’s largest and most visited sites stay up and running.

Losing user data is the ultimate nightmare for anyone who manages a site’s web operations and architecture, agreed the members of the panel devoted to web ops management, which was moderated by Facebook VP of Technical Operations Jonathan Heiliger. MySpace’s Richard Buckingham, who’s VP of Tech Ops for that social networking site, has experienced this first-hand. Though Buckingham said MySpace restored the data from its backup and the site was only affected for a few hours, it took extra precautions to make sure it was better prepared in case another error happened. “Since that incident, we’ve built out stronger snap restore and geographical replication of data,” he said.

To protect against using loser losing user data, Microsoft classified it into three buckets — high, medium and low business impact — and built three different data strategies around them, said the company’s GM of Global Networking Services, Najam Ahmad.

On the innovation side of web infrastructure, MySpace is particularly interested in the development of flash technology. “Fusion-io is the product we’ve been looking at and it’s pretty groundbreaking and revolutionary,” said Buckingham. “It’ll be exciting to see how that changes.” However, unlike Yahoo, Google and Microsoft, who all said IPv6 technology will be adopted within the next 2-3 years, MySpace isn’t interested in it for now.

“We don’t care; it’s not a business priority right now,” said Buckingham. “When it is, we will deal with it.”

Finally there are the challenges that result from mergers and acquisitions, which largely involve figuring out the best way to integrate two companies’ infrastructure. Yahoo VP of Infrastructure and Operations Strategy Raj Patel said migrating the agility of a startup into an established business’ massive architecture is best done methodically so that the innovation of the site isn’t affected. Microsoft’s Ahmad said the company decides whether to merge various types of infrastructure on a case-by-case basis. “Sometimes we integrate (acquired companies’ infrastructure) and sometimes we don’t,” he said, but noted the majority of the technology Microsoft builds is, not surprisingly, Windows-based.

Video of the panel is here:

Photo by James Duncan Davidson.

10 Comments

Carolyn Pritchard

Hi Wolfang —

We’ve just posted it — enjoy!

best, Carolyn

Wolfgang Kandek

This was an excellent panel at Structure 09, good information on how different these companies go about scaling – any chance that you will put the video online ?

gp

previous year it was social networking ……….this year flavor is Cloud computing

Om Malik

gp

not sure what is your point? what are you trying to say. please elaborate and expand on why you are making a statement. i would love to know what is your reasoning.

a

This is a useless piece of article with no real insights. The title says…. “How Some of the Busiest Sites Manage Their Web Ops”.. can you sincerely answer the question after reading the article??

Carolyn Pritchard

Chris — Indeed, and that is fixed. Was a long day :) best, Carolyn

Comments are closed.