Instagram and Tumblr grapple with growing pains

@CNNMoneyTech May 25, 2011: 4:07 PM ET

NEW YORK (CNNMoney) -- Remember a year ago when Twitter's Fail Whale kept making a splash?

Or back in December when cranky users took to Twitter to complain about Tumblr outages?

For popular startups, growing pains are a rite of passage. Scaling often means fail whales and tumbeasts. At TechCrunch Disrupt 2011, Tumblr founder David Karp and Instagram's Kevin Systrom took the stage to discuss the pains and gains of hockey stick growth.

"Exponential growth is unfathomable until you cross over to the next side," Karp said.

The blogging site has faced a host if problems that come with the scaling territory: downed servers, database bottlenecks and other glitches.

But with 250 million page views a day, the pressure is on Tumblr to keep the site up and running smoothly. Karp admitted that when the servers fail, nobody's happy.

"It's absolutely heartbreaking," he said candidly. "It's the most gut-wrenching feeling in the world."

Since Tumblr crashed for 24 hours in December, Karp said the company has "quadrupled" its engineers.

Say cheese! Instagram's Kevin Systrom says his idea for a photo-sharing app was a risk. "It wasn't clear that it was going to take off," he said.

But less than a year later, Instagram has experienced instant growth. With only 4 employees, the company has over 4 million users and an average of ten photos are posted per second.

Still, both Systrom and Karp conceded that it is one thing to grow rapidly right out of the gate. Sustaining that growth is a different matter.

"You can manufacture short-term growth, but I don't think that long-term growth you can manufacture." Systrom said.

Karp added that it's difficult to even know what lies ahead past a certain point. "Looking forward [over the] next 6 months, planning is an interesting exercise but often feels futile," he said.

But while long-term growth may be difficult to manage for a startup, Karp said he wishes he'd anticipated some of the challenges earlier.

"It took me too long to realize we needed more resources, data centers, more [of an] engineering team," he said. To top of page

  • -->