25 July 2015

I had a lot of discussions about SIEM rollout in the past weeks. One approach is to watch only Windows server systems that store business critical information or provide critical infrastructure services. Why should we waste time and effort on information not critical for business? That sounds convincing, in particular with a risk based approach in mind.

My approach goes far beyond this. I strongly recommend to watch all windows server system through SIEM.

The reason is quite simple: In a Windows server network lots of user accounts and technical accounts are used for administrative tasks. In general, this accounts are globally defined (in the Windows Active Directory) and member of the individual server’s administrators group. And, in general, this accounts are used for all systems, even for those storing business critical information.

If one assumes the about 10% of a company’s servers manage business critical information, hacking attempts on 90% of the servers will remain undetected. An attacker who hijacks one of the non-critical systems, and starts a DLL injection attack on the Windows Local Security Service lsass.exe to extract plain text passwords from memory, will have access to all of your systems within minutes.

But if you watch all servers through your SIEM system you will get a security incident within seconds after the hacking attempt takes place. With well-defined security incident processes in place you may be able to prevent the worst.

This reminds me of the report ‘Dissecting the Top Five Network Attack Methods: A Thief’s Perspective’ I read this week:

I love breaching a company that spends tons of money on gear but can’t get it working together. I know I leave traces, but by the time the admins connect all the dots, I’m long gone.

In the case above the admins do not even have the chance to connect all the dots because they are almost blind.

Have a good weekend!