Spock's public beta debut goes splat
People search engine Spock opened its doors today in a public beta. I posted about it this morning, calling the service a "bit slow." It has slowed to a crawl (15 to 20 seconds for the results) for most searches I have tried today. It's a worst nightmare scenario for the startup, which has garnered a lot of positive press.
It would be good to get some statement from the company about what is going on. I pinged Spock's PR maven, Renee Blodgett. She said the Spock team is working on the problem and will post something on the Spock blog at some point. It would have been better post something on the Spock front door or blog sooner than later to let the flood of users know that the site is getting hammered and the company is working on resolving the issues. It's obvious, but communicating with the audience is better than silence in times of stress. Spock uses Amazon's Elastic Computer Cloud for some of its service, which may have something, or nothing, to do with the performance problems.
Spock won't be irreparably harmed by the bad performance today. It's a good lesson that won't be forgotten, and we are in the age of the perpetual beta. In the long run, Spock will succeed if it can deliver great results, but as Google figured out early, speed is a key component of the user experience. Slow and steady doesn't win the race.
Update 3:15 pm PST: Spock co-founder and CEO Jaideep Singh posted this note on the Spock blog::
Dear Readers,
After a very successful private beta, we launched the Spock public beta today with great fanfare. During the private beta period we created a strong community of users, bloggers, and reviewers. For this we are very grateful.
In anticipation of our public beta launch, we had catered for peak capacity of 100 page views a second, which translates to 300 million page views a month. However, since this morning we have been getting a consistent request rate of 300 to 400 page views a second, which translates to nearly 1 billion page views a month.
Although this level of demand is gratifying, we sincerely apologize for not being able to serve it all. The entire Spock team is working hard to add more capacity today and tomorrow. Please bear with us while we add more bandwidth to meet the needs of our global user base.
Sincerely,
Jaideep Singh
Co-founder & CEO
In an email Singh said that Spock employs about 200 Amazon Elastic Computer Cloud server instances, mostly for crawing and indexing. The Amazon server are not used for serving up the site he said because they are not very "performant." The company is reallocating some of its non-production servers to bring up more capacity to meet demand and get the site performance on track.
Update 2 6:15 PM PST: Just checked and the Spock site is now zippy.