Direct Access at NEBytes
Tonight was the second NEBytes event, and after the launch event I was looking forward to it. Unfortunately the turn out wasn’t as good as the first event, 56 were registered but I only counted approximately 22 in the audience. The topic I was most interested in was a discussion of Microsoft’s Direct Access (DA), this was billed as an ‘evolution in remote access capabilities’. Being a security guy, obviously this piqued my interest.
Tonight’s speaker covering DA was Dr Dan Oliver, managing director at Sa-V. Before I start I want to state that I have/had no prior knowledge of DA, and my entire understanding comes from the presentation/sales-pitch by Dan tonight, if anyone with more knowledge once to point out any inaccuracies in my understanding or thoughts I’d more than welcome getting a better understanding of the technology.
DA is an ‘alternative’ to VPNs (discussed more later) for a Microsoft environment. The premise is that it provides seamless access to core resources whether a user is in the office or mobile. The requirements are fairly steep, and as Dan discussed on several occasions may be a stumbling block for an organisation to implement DA immediately. These are (some of) the requirements:
- At least one Windows 2008 R2 server for AD and DNS services
- A Certificate Authority
- Recent, high-end client OS: Windows 7, Ultimate or Enterprise SKU only.
- IPv6 capable clients (DA will work with IPv6 to IPv4 technologies)
As few organisations have a complete Win7 roll-out, and even less have the resources available to roll-out the higher end versions Dan was asked why the requirement. Answer: ‘Microsoft want to sell new versions, sorry’.
With DA pitched as an alternative to VPN at numerous points in the presentation the was a comparison between the two solutions, and to me the sales pitch for DA seemed schizophrenic. Dan kept switching between DA being an improvement to the current VPN solutions completely, and DA being suitable for access to lower priority services and data but organisation may prefer to remain with VPNs for more sensitive data. At this point I couldn’t help thinking ‘why add DA to the environment if you’re still going to have VPN technologies as well’. This was especially the case as Dan stated (and I can’t verify) that Microsoft do not intend to stop providing VPN functionality in their technologies.
From a usability and support perspective DA is recommended as it does not require additional authentication to create a secure connection to ‘internal’ services. Apparently having to provide an additional username/password (with RSA token/smartcard/etc.) needed to establish a VPN connection is beyond the capabilities of the average user.
One aspect that I did agree with (and if you listen to Exotic Liability you will be familiar with) is the concept of re-perimeterisation. The concept that the traditional perimeter of assets internal to a firewall is no longer relevent to protect resources in the modern environment, and that the modern perimeter is where data and users are, not tied to a particular geographical location or network segment. However, rather than the perimeter expending to encorporate any end user device that may access or store sensitive data, Dan claimed that DA would shrink the perimeter to only include the data centre, effectively no longer being concerned with the securityof the client system (be it desktop, laptop, etc.).
This point made me very concerned for the model of DA, if the client machine has seamless, always on access to ‘internal’ corporate services and systems I would be even more concerned for the security of the end user machine. If a virus/trojan/worm infects the system with the same access as the user account, then it too has seamless, always on access to the same internal services. I’m hoping this weakness is only my understanding of the technology, seems like a gaping whole in technology. If anyone can shed any light on this aspect of DA I’d appreciate some additional pointers to help clear up my understanding.
At this point I still can’t see an advantage to implementing DA over more established alternatives, my gut feeling is that DA will either become ubiquitous over the coming years or disappear without making an impact. Due to the fact it doesn’t play nice with the most widely implemented MS technologies, let alone ‘nix or OSX clients and the strict requiremented making a roll-out expensive I expect it to be the latter, but I’ve been wrong before.
At this point I decided to make a speedy exit from the event (after enjoying some rather good pizza) as the second event was dev based (Dynamic consumption in C# 4.0, Oliver Sturm) and I definitely fit in the ‘IT Pro’ camp of NEBytes audience.
Dispite my misgivings from the DA presentation I still enjoyed the event and look forward to the next. If you were at either of the events please let the organisers know your thoughts and ideas for future events by completing this (very) short survey. Thanks Guys.
— Andrew Waite