Does Compliance Drive Patching?

A thought-provoking comment was made in response to Ericka Chickowski's recent article "Best Practices For Oracle And Database Patching."The reader asked for a follow-up discussion on how compliance concerns alter the decision to apply security patches, and went on to advocate why it's much better to scan your databases and plan your patches in advance. Let's look at how database patching is affected by compliance, and why "getting ahead" is simply not possible.

So in response to the comment on the article, how does compliance alter the decision to patch the database? That depends.

Usually it depends on when the next compliance audit cycle begins. The mindset of many Oracle customers, especially within the small and midsize enterprise, is that compliance is the goal. You patch to prepare for the audit, not to be secure. That approach has a tangible reduction in operating costs--and an unknown increase in potential breach costs. Keep in mind many audits don't result in a fine on the first offense; a missing control or simply being one patch revision behind will result in an action item, not a penalty. Since compliance audits are a yearly occurrence, regulatory pressures do not speed up the adoption in a quarterly patch cycle. Fear of hassles from auditors, both internal and external, can alter the decision process in advance of an upcoming audit, but has little to no impact on the other three patch releases.

It's simply not possible to runs scans in advance to stay ahead of the threats. The threats are zero-day issues that are not known in advance. By definition, these problems are uncovered within the past couple of days, weeks, or months. Assessment scans cannot discover a vulnerability they don't have an existing policy or signature for that describes the threat. Granted, there has been tension between the security community and Oracle for being slow to response to known issues: Oracle does not reveal a threat until a patch can be provided. And when it does announce a vulnerability has been patched, it is a little skimpy with threat details and possible workarounds. Hopefully this posture will change when virtual database patching is available to Oracle customers at large, and the signatures needed to block the threat will be made available before the patch is ready. Regardless, it's simply not possible to get ahead of patches you don't know exist.

It is understood that Oracle offers a quarterly security patch, and many organizations plan their patching efforts around that schedule. But as Michelle Malcher pointed out, you need to test the patches before you install them. This causes a delay that is proportional to the number of database features that are patched. The more security fixes bundled in each patch, the more testing is required, resulting in more delays. Consider that Oracle customers don't always want to install the latest CPU because they want to forgo the cost of testing and patching, and would rather wait. When they see some 80-plus fixes bundled into a patch, they hope no severe issue pertains to them so they can wait one more quarter.

One final comment regarding the complaint that it's really difficult to install individual fixes, and it would be easier on customers if they could pick and choose which bug fixes they wanted to apply. This approach has historically been fraught with problems because each customer is potentially on different database revisions leading to database reliability and compatibility issues. Upgrade scripts are tough enough to produce when upgrading from a known starting point. Upgrades to piecemeal revisions are much harder to get right. From a stability and customer support perspective, in the big picture it's easier on everyone to install the entire database patch.

Adrian Lane is an analyst/CTO with Securosis LLC, an independent security consulting practice. Special to Dark Reading.