Lazybit

Here is a set of points that emphasize the benefits of a smart-cart or token-based authentication solution, coupled with biometric authentication; the example is focused on Dekart Logon for Citrix, but it also applies to other user authentication software by Dekart.

Q: what are the benefits of using your product? Am I simply substituting a PIN for a user/password combination? And can an external user without a flash drive or smart card still access the server?
A: Dekart Logon for Citrix is not a server-side application, it should be used on the clients.

The benefits can be summarized as:

• users don't have to enter passwords manually
• therefore you can use extremely complex passwords (they won't have to memorize them anyway)
• since they don't have to memorize them, you can be sure your passwords won't show up in clear-text on sticky notes, or written somewhere under the keyboard, etc
• the fact that the credentials are stored on a smart card means that brute-force attacks are out of the question
• optional biometric authentication takes that one step further

• the software can also be used with flash disks, being entirely self-contained:

  • the Citrix client itself can be migrated to the USB disk
  • and the same applies to our program
  • as a result, an end-user can plug the USB drive into any computer (even where the Citrix ICA Client is absent), and log on to the Citrix server. (Of course, this is also possible if you use a web-based client, but in that case you have to beware of keyloggers [note: we have a solution for that too])

In this case the user is immune to keyloggers. Even if the keylogger manages to capture the PIN:

• they won't obtain the connection credentials themselves, therefore they won't be able to connect to the server without actually having the smart card (token, or USB drive)
• and even if they do - you can use biometry as an added layer of security
• further, the program can be installed in 'simple' and 'advanced' modes. In the first case, end-users can only connect to predefined servers, they cannot change the credentials, nor see the connection details - this makes the system fool-proof.

And as a side effect, this also means that unloyal end-users won't be able to disclose confidential data even if they want to. In other words, you can implement the "need to know" approach, by not giving users more information than they actually need to get their work done.

The data stored on USB drives are encrypted with AES-256 bit, our implementation of the algorithm is certified by NIST. This is much stronger encryption than the one used by the Citrix client itself.

Q: And can an external user without a flash drive or smart card still access the server?
A: Technically, this is possible, but you can counter that by:

• using extremely complex passwords, or randomly generated ones
• not disclosing them (just write the credentials to a key and issue it to an end-user)

You will probably want to take a look at Key Manager, this is the tool that allows you to write credentials to keys, make copies, edit contents of a key, etc.

Note - you can do these with Dekart Logon for Citrix itself, but if you're planning to operate with many keys (in a corporate environment), you'll find Key Manager very useful. A license for the tool is given for free if a certain number of licenses for Logon for Citrix is purchased.

Q: Couldn't someone with a citrix client installed on their machine get to my server logon screen on the remote machine and execute a brute force attack there?
A: Although that is technically possible, it is not an optimal scenario for the attacker to use:

• most network admins configure the servers in a way that prevents any host from connecting again if they've connected N times during the last M minutes (to prevent brute force attacks, conserve CPU cycles and network bandwidth, thus avoid denial-of-service attacks)
• from the attacker's point of view, the bottleneck of the procedure is the network bandwidth. When brute forcing a local resource, it goes much quicker because reading/writing RAM is much faster than crafting a packet, sending it over the network, then waiting for a response from the server, etc.

In other words, a local brute-force attack can take thousands or millions of years, while doing it over the network is totally insane. It may only work for trivial passwords such as '11111' or ones that can be found in any dictionay. But even in that case, a dictionary attack won't be feasible if the network admin took the right measures and prevents one from physically connecting to the server if they've had too many unsuccessful attempts.

Finally, the last detail is that you can use randomly generated passwords, which are extremely long - brute forcing THAT is impractical.

If I were an attacker, I'd try to find alternative ways, such as social engineering applied against a naive employee.