The Apache Software Foundation runs its open source projects on a hierarchy of principally three levels, top-level projects (TLPs), sub-projects and incubated projects. Achieving the TLP status is a major milestone for an open source effort and this week Apache announced that six projects were being graduated to TLP status.
This Yahoo-led effort powers more than 400 TB of traffic per day, joining six other projects that have now been elevated to the top level of the Apache open source software project hierarchy.
Among the six new TLPs, is the Apache Traffic Server, a project that was originally an incubated effort by Yahoo (NASDAQ: YHOO) in 2009. The Traffic Server is also being updated to version 2.0 this week as the technology continues to grow under the direction of the Apache model.
“It’s the fundamental goal of being in incubation that you succeed and graduate,” Leif Hedstrom, Chairperson of the Apache Traffic Server project, told InternetNews.com. “It signifies that our community and software have been well-governed under the ASF’s meritocratic, consensus-driven process and principles.”
In addition to the Traffic Server, five former sub-projects of existing TLPs have now moved up. Three of the new TLPs were formerly sub-projects of Apache Lucene. They include the Apache Mahout machine learning algorithms effort and Apache Tika which is a toolkit for content detection and analysis. The Apache Nutch Web search engine is also moving up to TLP status. Rounding out the list of new TLPs are a pair of Apache Hadoop sub-projects including the Avro data serialization project and the HBase distributed database.
Apache Traffic Server
The Apache Traffic Server is already heavily used by Yahoo to deliver 400 terabytes of traffic per day. The Traffic Server is a caching proxy server that can speed up the delivery of Web server content.
“Traffic Server is built for scalability on SMP hardware,” Hedstrom said. “It can take advantage of multiple CPUs very well.”
Extensibility is also a key attribute of the Traffic Server. Hedstrom said the Traffic Server has a very feature rich plug-in API, similar to how people write mods for the Apache HTTPD Web Server. From a competitive point of view, Hedstrom said the Squid project is among the popular open source caching servers that are alternatives to Traffic Server. Commercial proxy caches also exist and many match Apache’s Traffic Server’s features, though Hedstrom said that Traffic Server offers better price/performance.
Benefitting From Apache
By being part of the Apache Software Foundation, Hedstrom noted that the Traffic Server project has grown beyond what Yahoo on its own was able to provide. New developers have joined the project which has helped the technology to improve.
“We’ve directly seen a large number of changes that has improved the features, quality and performance of the software,” Hedstrom said. “A few examples include 64-bit support, IPv6 support, and performance in some benchmarks have more than doubled than what we had starting incubation.”
The Apache Traffic Server is currently at its 2.0.0 release which came out this week. Hedstrom noted that release’s main new feature is that it’s 64-bit compatible, but only supported on Linux.
Moving forward, Hedstrom said that there is a developer version 2.1.0 release which should be out in the next few weeks and a version 2.2.0 public release which has loosely been scheduled for the summer.
“The v2.1.0 Traffic Server has a number of features, including significant performance improvements, much better RAM cache efficiency, support for most common Unix platforms, including numerous Linux distros, Mac OSX, Solaris and FreeBSD,” Hedstrom said. “All these ports to non-Linux platforms is a direct result of us being part of the Open Source Apache community.”
Sean Michael Kerner is a senior editor at InternetNews.com, the news service of Internet.com, the network for technology professionals.