Joint Statement about GNU/Linux Security

April 4th, 2004

Executive Summary

GNU/Linux vendors Debian, Mandrake, Red Hat, and SUSE have joined together to give a common statement about the Forrester report entitled "Is Linux more Secure than Windows?". Despite the report's claim to incorporate a qualitative assessment of vendor reactions to serious vulnerabilities, it treats all vulnerabilities as equal, regardless of their risk to users. As a result, the conclusions drawn by Forrester have extremely limited real-world value for customers assessing the practical issue of how quickly serious vulnerabilities get fixed.

Full Statement

The security response teams of GNU/Linux distributors Debian, Mandrakesoft, Red Hat and SUSE have assisted Forrester in gathering and correcting data about vulnerabilities in their products. The gathered data was used at Forrester for a report that became titled "Is Linux more secure than Windows?". While the vulnerability data regarding GNU/Linux which is the basis for the report is considered to be sufficiently accurate and useful, Debian, Mandrakesoft, Red Hat and SUSE, from now on referred to as "We", are concerned about the correctness of the conclusions made in the report.

We believe that it is in the interest of our usership and the Free Software community to respond to the Forrester report in the form of a common statement:

We were approached by Forrester in February 2004 to help them refine their raw data. Forrester collected data about the vulnerabilities that affected GNU/Linux during a one year period (June 2002 - May 2003) and looked at how many days it took us to provide corrections to our users. Significant efforts have been put in not only making sure that the underlying dataset for the vulnerabilities was correct, but also to articulate the special technical and organizational care taken in the response processes in the professional Free Software security field. This expertise is greatly appreciated by our usership since it adds a high value to our products, but we see that most of this value has been ignored in the methods used for the analysis of the vulnerability data, leading to erroneous conclusions.

Our Security Response Teams and security specialized organizations of respectable reputation (such as the CERT/DHS, BSI, NIST, NISCC) exchange information about vulnerabilities and cooperate on the measures and procedures to react to them. Each vulnerability gets individually investigated and evaluated; the severity of the vulnerability is then determined by each of the individual teams based on the risk and impact as well as other, mostly technical, properties of the weakness and the software affected. This severity is then used to determine the priority at which a fix for a vulnerability is being worked on weighed against other vulnerabilities in our current queues. Our users will know that for critical flaws we can respond within hours. This prioritization means that lower severity issues will often be delayed to let the more important issues get resolved first.

Even though the Forrester report claims so, it does not make that distinction when it measures the time elapsed between the public knowledge of a security flaw and the availability of a vendor's fix. For each vendor the report gives just a simple average, the "All/Distribution days of risk", which gives an inconclusive picture of the reality that users experience. The average erroneously treats all vulnerabilities as equal, regardless of the risk they pose. Not all vulnerabilities have an equal impact on all users. An attempt has been made to allocate a severity to vulnerabilities using data from a third party, however the classification of "high-severity" vulnerabilities is not sufficient: The mere announcement of a vulnerability by a particular security organization does not necessarily make the vulnerability severe - similarly, the ability to exploit a weakness over the network (remote) is often irrelevant to the vulnerability's severity.

We believe the report does not treat vendors of Free Software and the single closed source vendor in the same way. Free Software is known for its variety and its freedom of choice amongst the standards it defines. Multiple implementations of these standards are typically offered for both desktop and server use, which gives users the freedom to select software based on their own criteria rather than those of the vendor. The openness, transparency and traceability of the source code is added value in addition to the larger variety of software packages available. Finally, the claim that one software vendor had fixed 100% of their flaws during the period of the report should be incentive for a closer investigation of the conclusions the report presents.

signed,
Noah Meyerhans, Debian
Vincent Danen, Mandrakesoft
Mark J Cox, Red Hat
Roman Drahtmüller, SUSE

Additional Information:

Javier Fernández-Sanguino Peña composed a survey in 2001 and discovered that it has taken the Debian security team an average of 35 days to fix vulnerabilities posted to the Bugtraq list. However, over 50% of the vulnerabilities were fixed in a 10-days time frame, and over 15% of them were fixed the same day the advisory was released! For this analysis, all vulnerabilities were treated the same, though.

He has rerun the survey based on vulnerabilities discovered between June 1st 2002 and May 31st 2003 and found out that the median value of delays between the disclosure and releasing an advisory including a correction was 13.5 days (average is 31.10 days). Again, for this analysis advisories were not classified with different priorities.