From spring 2007 to 2008, Twitter handled its messages through Starling, a Ruby persistent queue server.
Starling was part of Twitter’s early infrastructure, when the service was growing rapidly and its original architecture was under pressure. A persistent queue temporarily stored and passed messages between components, helping Twitter absorb bursts of activity instead of requiring every operation to complete immediately. The system was written in Ruby and became closely associated with Twitter’s early scaling story.
Starling was not the same as RabbitMQ, Sidekiq, or Resque. Those are separate queueing or background-job technologies, many of which became prominent later or were used in different application contexts. Twitter gradually moved important infrastructure away from Ruby and toward Scala and the Java Virtual Machine from 2009 onward. That transition was intended to improve throughput and support a more distributed service architecture.