After More Downtime, We Ask: Can Twitter Truly Scale?

On Monday, a database hang-up on a long-running query was causing problems for both Twitter.com and the service�s API, which in turn affected Twitter clients outside of the official website.
The site has grown at an incredible rate since its inception, and it�s always struggled to keep up (technically speaking) with user adoption. Ever since its userbase spread beyond the tech elite to more mainstream social media users, the site has been subject to a seemingly unending string of growing pains.
In response to a wave of failures last month during the World CupWorld Cup, TwitterTwitter engineer Jean-Paul Cozzatti posted that the company�s network was improperly configured, leading to downtime and trouble with features.
Cozzatti claimed that the company had doubled the capacity of its internal network, improved its monitoring and rebalanced its traffic to prevent future downtime.
�For much of 2009, Twitter�s biggest challenge was coping with our unprecedented growth (a challenge we happily still face),� he wrote. �But as this week�s issues show, there is always room for improvement.�
Apparently, the improvements made last month were not enough to keep up with user growth and their demands on the service.
Cozzatti posted again today to address Twitter�s issues Monday, comparing the engineering team�s work on scaling the app to �riding a rocket.� To give you an idea of how fast that rocket is soaring through the social media universe, we reported in April that the service had 105 million users. In his post today, Cozzatti said Twitter now has 124 million registered users.
Scalability and uptime are the team�s top priority; in fact, Cozzatti noted, other projects are being put to the side until these problems can be thoroughly solved. He stated that Twitter�s engineers have made more than 50 performance and optimization tweaks since the site�s World Cup woes, including doubling throughput to the database that stores tweets, improving how the app uses memcache and improving page caching of the front and profile pages, which helps to reduce page load time.
Nevertheless, �There are still times when we run into problems unrelated to Twitter�s capacity,� Cozzatti admitted.
Monday�s issues serve as a perfect example. During the database malfunction and restart � which covered a 12-hour period � users were unable to login, sign upor update their profile information and design. �In the end,� wrote Cozzatti, �this affected most of the Twitter ecosystem: our mobile, desktop, and web-based clients, the Twitter support and help system, and Twitter.com.�
While Twitter is getting its own data center this fall and is actively recruiting more engineering talent, the company clearly needs to implement the long-term solutions we�re reading about. To have the cultural cachet of a web service such as GoogleGoogle search, GmailGmail, FacebookFacebook or any of the apps we rely on for day-to-day work and life � and we have the distinct impression that Twitter does, indeed, hope to be part of that cadre � it must first and most importantly achieve an acceptable uptime ratio. All the partnerships, revenue and media buzz in the world can fall a bit flat when the app itself doesn�t work.
That being said, we�re certain Twitter can and will make the needed improvements for scalability, and growing pains are good pains to have.
Are you confident that Twitter can become reliable and stable in the near future? Or do you foresee more significant downtime from this service? What will it take for Twitter to grow as quickly as its users require?
Read More here