Electric Cloud Takes Command Of App Production Process
November 1, 2006 —
Automated build tool maker Electric Cloud has announced plans to ship ElectricCommander, a browser-based tool for automating the build, package, test and deploy processes, later this year.
The company positions ElectricCommander as the finishing touch on its suite of so-called production management software, joining parallel-build tool ElectricAccelerator and ElectricInsight, an add-on to ElectricAccelerator that brings detailed analysis and reporting on project builds.
Electric Cloud’s approach distinguishes the design and code phases from later phases. From this perspective, the company claims that 80 percent of the production tasks can and should be automated, to allow people to focus on project-specific tasks such as defining test policies, combing the output, and specifying the build and test commands.
ElectricCommander is meant to replace the brittle, homegrown production processes that many shops employ because they’ve worked so far, according to Electric Cloud CEO Mike Maciag. He argued that agile development methods can’t afford to wait for the old-fashioned makeshifts. Distributed development teams add to the challenge, Maciag observed, adding that the push for more complete governance and corporate compliance demands the replacement of manual processes, as does the increasing use of virtualized systems, whether in a grid environment or not.
At the heart of an ElectricCommander installation is a multithreaded Java-based management server, with attached database for storage of command packages, log files and metadata. The management server runs on Microsoft Windows XP and Windows Server 2003, and Red Hat Enterprise Linux.
The server manages the build and test processes, allows tagged project assets to be shared and reused, and updates the relevant log files at each step in a process for transparent access to how a job runs.
The procedure packages can be set up with time limits on each step, or at selected points. Agents on the build machines execute the job steps, monitor the status of the process and report the results back to the master server. The agent can run in three Unix environments (AIX, HP-UX and Solaris), as well as Red Hat Linux and Windows.