Bruno Dillenseger  -Research and Development engineer for France Telecom R&D and very active member of CLIF project's Community- gave us an exclusive interview to let us know about CLIF Project. Let's discover the project, the story behind and which features have made an open-source cloud project of it.

What is the story behind Clif ?

CLIF comes from research work on component-based distributed software architectures. Actually it has been launched as a use case of the Fractal component model from ObjectWeb.org open source consortium (OW2.org today) - see fractal.ow2.org. This research on component models was jointly carried out by France Telecom (aka Orange today) and INRIA (French public research institute).

The idea of addressing the specific field of load testing comes from a number of specific performance test cases encountered in the context of R&D projects. No existing open source platform in the field of load testing was sufficiently flexible and powerful for our needs: efficient support of distributed load testing, millions of virtual users...

The key idea behind using a component-based architecture for CLIF is to maximize adaptation capabilities by minimizing the adaptation effort and always avoiding modifying its code, just by adding, removing or replacing a component implementation. This is always done at the smallest possible scale thanks to a recursive component design (components have sub-components, and so on). This is described in details in this paper: "CLIF, a framework based on Fractal for flexible, distributed load testing", Annals of Telecommunications, Vol. 64, No 1-2, January-February 2009 (available from SpringerLink).

The goal is to be able to quickly get an operational load testing tool for any new test case, whatever the technical context.

How many companies use Clif ?

This is very hard to figure out for a freely downloadable software. Orange, of course, as a project leader and main contributor, is using CLIF a lot. We have partners for commercial support (see http://clif.ow2.org/partners.html), but it is not easy to know the amount of business they make based on CLIF. 

Does any large corporation use Clif ? How many users?

Orange makes a wide use of CLIF: more than 40 R&D projects have been using CLIF for performance testing through a great variety of protocols. Sometimes, CLIF is also used for load testing Orange's on-line end-user or infrastructure services.

If you look at CLIF's mailing list archive, you will see e-mails from Bull and RedHat, which are big companies, but I can't tell you more about what they have been doing or what they are still doing with CLIF.

From your point of view, what brings the open-source distribution to Clif ? How do you contribute to open-source?

As a corporate R&D product, going open source is a practical and efficient way of providing the product with durability and confidence, which is key to conquer internal, corporate users and contributors. Since the source code is open and freely available on a public forge, users drastically limit the risk of losing their investment on learning, developing test scenarios and possibly extending the platform (custom load injectors, probes, report generators...).

Moreover, we could gather contributions from the community (load injectors for support of extra protocols, porting to other operating systems or system architectures), and we also greatly improved the platform and the confidence we have in it, thanks to the feedback from the CLIF's users community. This last point is particularly essential to testing tools: the more user feedback you get, the more confidence you have in your test results.

Last, open source is also an efficient way to cope with Intellectual Property constraints when we collaborate with other companies or academic labs.

What do you think about TIO Libre definition here: http://www.tiolibre.com/guideline/tiolibre-Libre.Definition?

Personally, I find it interesting, like a natural prolongation of the free Software Principles, in the sense of "libre", extended to operating this software. In any case, users must always be owners of their data, for free. I think that will might be a requirement for the vision of cloud computing, since you might dynamically change of provider or mix providers.

Interview by Elodie Pot - writer

Clif - http://clif.ow2.org

OW2 - http://forge.ow2.org

(c) 2010 TioNews

Follow us on Twitter: @tio_news