This is page 2 of:
Windows XP End-of-Life Could Cripple PCI Compliance
Scenario 3: “We don’t know the impact yet, because we have a complicated POS environment.” This group includes any retailer with self-serve kiosks, gasoline pumps or similar environments that run on Windows XP. These kiosks can involve multiple interfaces that connect payment acceptance, receipt printing, product delivery robots (e.g., in a vending machine), display screens and Internet access. That means changing the operating system may be a complicated project. This group of retailers also includes franchisors with a large number of different franchisee environments, where changing an operating system could be far from trivial.
The questions for these retailers include all of those listed in scenarios 1 and 2.
Could retailers just keep their end-of-life Windows XP operating system and develop a compensating control for Requirement 6.1? That approach could be difficult. Retailers are not likely to have either the resources or access to the Windows XP source code to identify new vulnerabilities or to develop their own security patches and then push those patches to the affected Windows XP systems with the same rigor and robustness as the vendor that developed the product originally.
Developing a compensating control would be difficult, because every compensating control requires a risk analysis. In this case, the control would need to reflect a high likelihood of a maximally damaging system compromise. Conceivably, a retailer could combine a host-based intrusion protection system (IPS) with enhanced monitoring and, say, frequent penetration testing. But what is the rationale for needing the compensating control in the first place?
A compensating control is designed for situations where the merchant “cannot meet a requirement explicitly as stated, due to legitimate technical or documented business constraints.” Although merchants have a fair amount of leeway in defining these “constraints,” both the advance notice of Windows XP’s end-of-life status and the number of operating system (and payment application) options argue against this being an easy case to make. Another reality is that any compensating control is only good until the next assessment, which usually is one year. Retailers contemplating this path need to work closely with their QSA and their acquirer before pursuing any type of compensating control strategy.
Windows XP’s upcoming end-of-life may have a subtle impact on the PCI SSC’s list of validated applications. Normally, a PA-DSS validation on a given application version and platform is good for three years, after which a new report on validation (ROV) must be prepared and submitted to the PCI SSC for review and acceptance. Will the PCI SSC accept ROVs for Windows XP payment applications this fall (some of these were originally validated under the expiring PA-DSS version 1.2) for the full three years? A 2016 renewal date would be almost two years past the operating system’s end-of-life date. Similarly, what about currently listed Windows XP-based applications with 2016 expiry dates? I noticed several of these in my pass through the list. Would—or should—the PCI SSC revoke its validation on April 9, 2014, even though there would be two years left on the original validation?
Having raised this question, let me now knock it down. Regardless of an application’s PA-DSS status, if the application is validated and running on an end-of-life operating system, it will not be compliant with PCI DSS Requirement 6.1 and the retailer is not PCI compliant. The fact that the Windows XP version is PA-DSS validated is a technical detail that does not change the facts on the ground.
Do you have Windows XP POS or payment applications? What are your plans for migrating away from Windows XP in the coming year? Or, do you just plan to head to Tatooine and watch both sunsets? I’d like to hear your thoughts. Either leave a comment or E-mail me.
-Marc
February 6th, 2013 at 12:26 pm
Another possible solution – POSReady 2009, which we are currently investigating. Mainstream support will end April 2014, but extended support will continue to April 2019. Nothing like putting off the inevitable, but a few more years will certainly help.
February 6th, 2013 at 4:33 pm
This is an interesting issue, pivoting largely on the interpretation of PCI 6.1. One could argue – no new vendor patches means no missing patches therefore compliant. The truth is probably in the middle – vulnerability management, mitigating controls, and possibly the messy compensating control path.
February 6th, 2013 at 8:56 pm
Thanks for the comments, Seth and Lyal.
@Seth, I’m not too familiar with POSReady 2009, but a possible short-term stay of execution might come from XP Embedded, which is not end-of-life until 2016. That option may appear to buy you two years, but check that the application is validated for that specific operating system — either POSReady 2009 or XP Embedded — first.
@Lyal, You raise an interesting point, but I still don’t think your solution will fly. In the case of XP, saying no new patches are available means that nobody is looking for new vulnerabilities and nobody is fixing the system. It seems pretty hard to twist that to imply no patches are available so no patching is needed. Windows 2000, anybody?
While you mention a compensating control, and I tried to address that path in the column because it may technically be possible, actually I was doing my best to dissuade anybody from going there. I cannot see any comp control being effective.
Beyond PCI DSS Requirement 6.1, another factor condemning Windows XP after April 2014 is contained in the ASV Program Guide. One of my colleagues pointed out to me that when you look on page 15, you’ll see it says:
“The ASV scanning solution must also be able to determine the version of the operating system and whether it is an older version no longer supported by the vendor, in which case it must be marked as an automatic failure by the ASV.”
That means your Windows XP earns you an automatic FAIL on the next external scan, so once again, you are not PCI compliant. I can’t imagine an ASV treating an end-of-life operating system as a false positive.
I keep going back to Microsoft’s advice which I quoted: “If your organization has not started the migration to a modern desktop, you are late.”
Walt
February 7th, 2013 at 12:37 pm
POSready 2009 is based of the Windows XP SP3 codebase. It’s the successor of Windows Embedded POS with was initially launched based on XP Embedded.
Windows Embedded Standard 2009 (“standard” is the new name for the toolkit version of embedded, in this case based of XP Embedded SP3).
Both solutions will add many years to your devices without any changes on you side.
February 9th, 2013 at 8:02 am
We have no plans to move off XP after April 2014. Later Windows products do not meet our requirements.
February 13th, 2013 at 6:54 pm
@Walt – Windows XP (i.e. workstations) scannable by a ASV?
I think that has bigger issues that just being EOL – logging and firewall/Modem NAT issues to start with.
On a related note, I have a client where the ASV scan regularly detects several servers as the incorrect version. The “record as a false positive” option for this ASV service is too hard to do every 3 months, so the finding gets ignored, once QSA validation of the actual environment takes place.
March 30th, 2013 at 12:51 am
First, Windows XP is still around because people like it! I wonder if anyone has considered Ubuntu Linx? The OS is straight forward and works extremely well and it’s FREE! Well, except for the profesional online support but $250 per year no bad.
May 19th, 2013 at 1:23 pm
What about placing the XP POS terminals on an intranet network without any comunication or connection to internet by IPS, NAT, Firewall, etc.? They would only communicate with the main server on the intranet (running Windows 7 or 8) and only allow the main server communication to internet for needed functions like Credit Card transactions or like EDI to other main servers.
June 6th, 2013 at 12:17 pm
You may want to look at this article:
http://storefrontbacktalk.com/securityfraud/out-of-date-os-causes-pci-violation-no-but-why-let-facts-trip-up-a-marketing-letter/
It is older but completely contradictory of this article. Change of opinion or interpretation? Either way PCI compliance doesn’t clearly state one way or another.