Feed on

Why do I care about defining Tier 0 and why is it a problem to have a large Tier 0? It is all part of minimizing the attack surface. You want to minimize the places where it is possible to find a domain administrator account and exploit that. It is far easier to secure 20 computers rather than 200 or 2000. But with GPOs you can manage 10 as easy as 10 000 computers so what’s the big deal with Tier 0?

It comes with the inherit problems of authentication and authorisation that we need to make sure that it is possible to work without having to type your password for every transaction. This is done by placing a hash that is calculated using your username/password/token etc. This is reused for some time to validate your credentials. This makes the system easier to use but also creates a possible vulnerability. Sadly, the hashes are possible to hijack and reuse therefor the problem with credential theft exist.

The hashes are stored in memory and is possible to steal with the right tools meaning that we have to focus on removing the places where domain administrators have logged in to as few as possible. This means that implementing privileged access workstations is imperative to minimize credential theft. Such workstations mean in short that you have one computer for mail and one for administrative tasks. This of course create a cost if you have 100´s of administrators that needs to have two computers.

Defining Tier 0

Credential Theft is a bid problem today. Many of the attacks we see are targeting accounts rather than the individual computers. This is due to the cost of exploiting. As soon as you have a valid account it is much easier to travel around and try to find a domain admin account. As soon as you have domain admin you have it all (this goes for root etc. as well).

One of the problems I´m challenged with is the definition of Tier 0. What is this then? How do you define Tier 0. The simple definition is: Every computer that either define or manages domain administrator accounts or can managed those computers in such a way that they either have physical access or administrator access on the computers. Compare with the PCI DSS definition of system components.

This for example means that any computer where a domain admin has logged in to recently or where a service account is run with domain admin privileges is also part of Tier 0. At a sales call recently after a brief chat we identified that the client had 2/3 of their computers part of Tier 0.

WPA2 breached

May you live in interesting times! Using WPA2 apparently is not a good idea anymore. This caught my interest as it is a breach on a protocol level rather than just a function and there are many companies that have moved to WIFI and rely of WPA2-Enterprise to secure the communication. So many WIFI units that are out there that needs to be upgraded to keep the company secure. Not to mention all the POS-systems that rely on WIFI for managing the payment using credit cards.

We will most probably see a surge of attacks against payment systems to harvest credit card information and a new tool to crack the corporate networks and look for vulnerabilities to expose.

We should, however, have in mind that no network is secure so if you asume breach in the first place and have implemented protection in layers this will not be one of your bigger problems.

Here in Sweden GDPR is one of the hottest topics within security. There is a lot of confusion regarding what is needed to be done and what different parties need to do.

First of all, GDPR is a law. Any lawyers out there would probably want to correct me as it´s an EU thing, but in essence it needs to be abided by. Hence, the company lawyer has the last say about what is right or wrong. For me, as an architect, my goal is to deliver possible solutions to the problems that arise when you try to abide by the law.

That said, GDPR, is not the only law to follow. On the contrary, for a public entity there are a number of laws that normally are a lot stricter. GDPR more or less, just enforces those laws more, meaning that the security needs to be ramped up quite a bit to ensure that the risks are managed.

GDPR, from a public entity point of view, should be regarded for all information that is not managed by other laws, and as security and operational requirements for the more sensitive information residing in processes governed by the more stricter laws of secrecy or health data.

As many public entities, especially municipalities, have a rather complex setup of services, including partly outsourcing systems to other municipalities, the amount of information that potentially could fall under GDPR is large meaning that, from a threat perspective, there is a large need of security services.

From my point of view, as a security architect, the first action to take is to conduct a risk analysis, either on a larger scale or more focused on the technology. In the end GDPR demands that a lot of legal issues are managed as well, but as stated before, I only try to sort the problems that have a technical or procedural solution.

Second step is to create a governance structure for managing changes that make sure that GDPR in general and security in specific is adhered to and that mapping to the risk analysis is done. GDPR mandates security by design and privacy by design. This is translated in operational terms to map the risk analysis to the solution so that you could show that the risks are managed in a correct way.

Third step is to manage the standard risks that exist due to the fact that you have an infrastructure to manage, no matter the type of technology used, those are the risks that are classified as You Have if you follow my You3 model. One of the most important steps to do here is to secure your environment against credential theft and lateral movement. You could find an official Microsoft description on Securing Privileged Access here but I strongly advice you to take external help here as it is rather complex to do it yourself.

Fourth step is to secure your databases with encryption. There are many ways of encrypting your databases and if you are running Microsoft SQL Server you´ll find the process rather straight forward with only a small amount of time needed to implement the technology. The processes for key management is a rather different thing however.

Fifth step is to ensure fully working backups of all important data and that all of those are secured as well. Remember to backup your keys as well, otherwise you have built a Do-It-Yourself-wiperware.

Sixth step is to ensure that you monitor your environment and make sure that you catch any attackers sooner rather than later. There are many tools out in the market but make sure that you go for those with built in machine learning and those that utilise information of attacks globally to protect your organisation. Add a good antimalware with ransomware blocking capabilities as well.

There are many more steps to take into account to build a secure environment but the steps taken above will be more than enough to keep you busy for months to come.

Yet a former client of mine has been hit by a ransomware. They used an online backup system that used mapped drive so they was partly encrypted as well. Still they were immensely lucky to having tried Azure Recovery Vault. Before joining Microsoft I had very little knowledge of the inner workings of Azure. Currently I´m supposed to attend a mandatory training in Azure and has to sit through 60 h of training in Azure at http://openedx.microsoft.com (yes, free for all). I came in contact with Azure Recovery Vault and tried it out with a free account. I called a friend of mine and asked him about it and if they tried it. He started laughing a told me that he had tried it for backup purposes about three weeks ago and that they where hit by a ransomware the week after. But thanks to this they could recreate their data. I think I´ll start looking at using Azure as backup solution as a way to protect against ransomware.


And ransomware has been weaponised.

If you remember my post a few months back regarding the future of ransomware we now see the emerging Wiperware, malware that´s sole target is to create mass destruction rather than take files for ransom. The article mentions Maersk loss of more than $200 million to NotPetya and that ransomware will land somewhere around $5 billion this year in costs. That’s a lot of money that could be better invested in simple security functions instead. We are still looking at way to many companies having a very unsecure setup.

One of the best things with working at Microsoft is all the things you get to know and one of the worst things is all the things you are not allowed to tell (yet).

Still I took a look at Credential Guard today to understand how it works and I found this document that describes it on a bit more technical level.

Looking at it from a more architectural point of view it enables us to put a bit more trust into the clients. The viewpoint until now is that the client devices are unsecure by default and sensitive information should not be allowed to be stored on the clients. Not even with encryption software installed. Now there is a possibility that we could revisit that as Credential Guard protects the user credentials from being access by malware hence blocking possible lateral movement.

Today I start my first day at Microsoft! Wish me luck!

An architect asked me yesterday how you use a reference architecture when it comes to security architecture. ‘How can I be sure that it is applicable for me?’ was the simple and yet interesting question.

The answer is not that obvious. First of all, we need to position your own security to the reference architecture. We do this with the risk analysis. Without the risk analysis you cannot know what risks you actually have. A risk analysis needs to have both an administrative part and a technical part to be valid. When you know your first batch of risks you´ll start to map them towards the risks that the reference architecture mitigates. A skilled security architects does that in a few minutes.

When you done that you know if the reference architecture is applicable for you. In some cases (like Microsoft’s SPA concept) the risks are valid for everyone having an Active Directory making it very easy to tell if it fits you.

This is sadly not the end of your peril. When you know the reference architecture to use you now have to map the cost of implementing it with the values you are trying to protect. First when that is done you know if the reference architecture is the right thing for you.

I don´t know how many of you that spend your time reading about security issues during your vacation but you have probably heard about the struggles at the Swedish Transport Agency.

There are quite a few things not right in the current rounds of the news but that will be sorted eventually. All in all, the problem is that the one responsible have decided that a few laws governing security of information was possible to opt out from (they aren´t) and allowed a number of systems to be outsourced to IBM giving a few foreign administrators that was not approved by SÄPO (Swedish Security Police) full access to all information.

The news currently say that information has leaked but that is to mistrust IBM quite a bit but the real problem is that information security is not regarded as an important topic when outsourcing. Using a cloud service is not the problem, not managing your security is a problem.

Older Posts »