Category Archives: Windows XP

Windows XP – Looming End of Life (E0L) – What are the risks ?

As I’m fairly sure that you may have gathered, Windows XP is going to go end of life very, very soon. 8th April this year (2014) in fact. This is proving to be a little bit of an issue for more than one organisation, many people have come to love Windows XP – and the old mantra of “if it ain’t broke, don’t fix it” has, until now – meant that there was little reason to move to Windows 7 or, even less to the poorly received Windows 8(.1). This has meant that, right now, across the world there are IT departments who are having a little bit of an issue – how to upgrade _all_ the machines in the organisation to Windows 7 as soon as is humanly possible. I have to say though that of the multiple organisations that I know of, not a single one is going to have finished their upgrade by the 8th …

So, realistically, where does this leave them in terms of risk ? I was actually asked this by a customer this morning – please quantify our risk. Well, I’m a big fan of statistics, not a great mathematician, but the concept definitely amuses me. So, thought I, what is the probability of there being a certain type of vulnerability this month. A quick Google didn’t throw up many sites with statistical data for XP patches, and I didn’t want to go through all of the Microsoft stuff myself, so I’ve borrowed the data from the excellent guys over at Secunia1. The following graph shows the vulnerability severities ( 356 in total over the last 10 years )2.


This isn’t entirely helpful as these don’t map directly to the Microsoft Classifications ( Critical, Important, Moderate, Low ). Let’s make one or two assumptions here then – we’ll assign the top four Secunia categories to the equivalent Microsoft ones, and we are going to assume that the vulnerabilities were evenly distributed over the 10 year period ( 120 months ). So, 1% of the vulnerabilities is equivalent to 3.5 vulnerabilities ( roughly – it’s 3.56, and I know that I should round up, but this is all assumption anyhoo ! ) So, each of those segments above equates as follows:

  • Critical ( Extremely) – 4% or 14 vulnerabilities
  • Important ( Highly) – 38% or 133 vulnerabilities
  • Moderate (Moderately) – 24% or 84 vulnerabilities
  • Low (Less) – 28% or 98 vulnerabilities

If we continue with our assumption that these have been evenly distributed over the lifetime of XP ( 10 years/120 months) we can see that the percentage probability of a given criticality occurring in a given month is equivalent to the total number of (vulnerabilities / 120) * 100 which give us the following:

  • Critical – Approx 10%
  • Important – Approx 110% ( More than certain !)
  • Moderate – Approx 70%
  • Low – Approx 80%

Well, that’s not very good news – it would suggest that each month moving forward would increase the number of vulnerabilities by these amounts, so at the end of 1 year you’d expect to see, on average:

  • Critical – 1.4 vulnerabilities
  • Important – 13.3 vulnerabilities
  • Moderate – 8.4 vulnerabilities
  • Low – 9.8 vulnerabilities

Ok, that’s fine – but what we are actually interested in is the residual risk isn’t it – what are we left with after we have considered our controls & countermeasures – what mitigation is in place. Well, the following information gives us a bit of a better idea:


Using the base assumption of even distribution we get the following probabilities of it occurring in a single month:

  • Remotely exploitable – 180% ( Nearly two guaranteed to be remote exploits !)
  • Local Network – 50%
  • Local System – 60%

Or as above, that’s:

  • Remotely exploitable – 1.8 a month, or 21.7 a year
  • Local Network – 0.5 a month, or 6 a year
  • Local System – 0.6 a month, or 7.2 a year

You can keep going if you want to, the following graph shows the actual impacts:


I’m not going to do that here, I don’t really think that it has much value. The point is that there is a distinct bias towards the first vulnerability being announced being an “Important, Remotely Exploitable” one.

So what ? Well, that’s interesting actually. Microsoft has given up patching XP, but that’s not the same thing as being left defenceless. Both Anti-Virus and Firewall technology for XP is going to continue being supported for some time – and if these countermeasures have been implemented, then there is a good chance that any given vulnerability will be completely mitigated by them – the trouble is, until the vulnerabilities are actually announced – you aren’t going to be able to tell how effective your controls will actually be – and you may need to do some fairly rapid reconfiguration of your firewalls &/or AV signatures to ensure that you are detecting and preventing those attacks.

Please don’t take that as permission to slack off in your upgrades, or even worse, decide that you can accept that risk – the best course of action is to upgrade to a patched and supported OS, however, the above at least has a stab3 at quantifying the level of the problems !

Just for the record by the way, I have confirmed with Microsoft that there will be patches for XP released on Tuesday the 8th April – these will be the last ever XP patches, but for those of you who have a monthly patching policy, you won’t actually breach your policy until the following month …

1. Guys, if you read this link, you should definitely bring back the free Vulnerability Alert mailings – but if you don’t a free subscription for plugging you would be welcome 😉

2. Ok, I realise that this is not 100% legitimate, there wouldn’t be an even distribution over the years, so this really is a generalisation. The distribution over the years actually looks like this …


If I had paid attention in University Mathematics lectures, I would remember how to do this more accurately, but I didn’t and I don’t…

3. You should look carefully at what your organisational risk appetite is, and also the full business impact of a vulnerability being exploited. Also, please remember that you may have obligations under other things (PCI/DSS for example) that you need to meet…