Colin’s IT, Security and Working Life blog

October 27, 2009

Providing very secure webmail

Filed under: Government IT Security, Uncategorized — Tags: , — chaplic @ 3:46 pm

 

Most office workers are familiar with the concept of “webmail”. It allows the employee to access their email from any web browser, on any internet connected PC. This gives staff flexibility, may remove the need to supply some staff with a laptop and allows access anytime and anywhere – for example, on holiday (if they are keen). Webmail looks similar to email in the office and allows the user access to their inbox, calendar and attachments.

Technical configuration is fairly straightforward – encryption is provided by the same type of system used to secure web-banking, and users get logged in either by using their office username and password, or occasionally a more sophisticated mechanism like SecurID (little fob with changing numbers). All major email packages include a webmail server and it’s straightforward to configure.

It is cheap to provide, easy to use, and popular with staff.

Some clients cannot accept the risks of providing a “vanilla” webmail solution. Why not?

The stereotypical answer of “security” is often used. But to understand why this answer is used, it’s necessary to look at aspects of a webmail system.

Firstly, the encryption. As the data travels across the public internet and untrusted systems, it’s necessary to encrypt it. This encryption is a flavour of “SSL” or Secure Sockets Layer – websites identified by a padlock and starting with https in a web browser. This is the exact same technology used by online buying and banking.

Whilst for most intents and purposes SSL is pretty secure, some organisations do not consider it secure enough, and if you are a man-in-the-middle you can potentially read encrypted data quite easily.

The other challenge is the “endpoint” – otherwise known as the PC or laptop. With a organisations own PC it’s possible to be reasonably confident that software patching is up-to-date, there’s no malware software installed, and anti-virus is up-to-date. This cannot be claimed of computers that are likely to be used for webmail access.

Computers used for webmail are likely to be home PCs (perhaps crawling with nasties) and public web-cafes. Web-cafes in Airports are a well-know target for people installing keylogging software as they are commonly used by businessmen. Such nasties can capture information and send it back to the attacker. It’s unlikely to be a targeted attack – the malware will be on millions of PCs – but it is unknown what the attacker will do with the information.

The attacker is probably seeking ebay login details or credit card numbers. But, potentially, for that session and maybe beyond, they can access what the user can access via webmail.

Finally, there is also the issue of data remnance. When a web page is loaded, all the information is stored locally on the PC to speed up access. This is especially true if the user accesses an attachment. This information is typically not encrypted (and, there is no way of controlling). Thus, the next user of the machine may very easily find information they are not intended to see.

Predictably, the market has developed solutions. Webmail can be accessed via a number of products all of which can check the endpoint to ensure anti-virus is up-to-date, ensure it passes a number of other tests, and wipe attachments when the session ends. It’s also possible to control what operating system and web browser the user is connecting from, though the PC may spoof this.

It’s also possible to setup filtering based systems where the webmail system either filters emails the user can see based on a label (i.e. do not show this email unless it’s labelled as “UNCLASSIFIED”), or the webmail system is a duplicate of the normal environment, but only containing non-sensitive emails.

Ultimately, the decision to implement such a solution lies with the organisations risk owners. Clearly, they need to be in full possession of the facts, risks and countermeasures. They will also need to support the development process because a novel solution like this is likely to attract attention.

The impact of not providing this facility needs assessed. How many staff effectively do this already by emailing documents to their hotmail account so they can work at home? Recently, transport unions have proposed short-notice 5-day strikes. How would the organisation cope if a key transport route was closed? What would be the impact of not providing this facility to carbon neutral and efficiency savings targets (need to buy 1000’s of people a blackberry or laptop?).

A likely technical solution would have the following aspects:

· A “front end” webserver in a secured (DMZ) network

· Use of best-commercial-grade SSL encryption

· Registration with companies on the internet that allow ongoing and continual penetration testing of websites

· Endpoint checking – a small software component would have to be downloaded to the untrusted endpoints. This checks the machine to ensure software patch levels and antivirus is at acceptable levels, and perhaps the operating system is agreeable. If the endpoint check fails, or cannot be loaded, access is denied

· A high degree of protective monitoring – user access would be closely logged and anomalies alerted (perhaps in real-time)

· There is an element of end-user responsibility therefore terms and conditions would have to be maintained and agreed. It may be necessary for the solution to be “opt in”

· Use of a one-time-password (RSA SecurID) to replace or complement a password

· It may be desired to redact some information or remove some webmail functionality – for example the ability to download or upload attachments

· It may be desirable to control what machines can access the webmail by performing an enrolment procedure and using certificates. This removes the ability to access from any PC but allows access from pre-agreed PCs (e.g. users home PC)

I’m a big fan of the Microsoft IAG product. At its most basic level it’s an SSL VPN, and brings with it the endpoint checking functionality – so we can ensure the client PC is at a certain patch level.

It also allows us to dip into the data being accessed – in real time- and perform filtering based on rules we set. Finally, it sits atop ISA Server 2006 which is a firewall that’s Common Criteria EAL4 evaluated – in other words, it’s a robust firewall.

A simplified solution architecture is shown below:

clip_image004

In conclusion, the effort required achieving this and the friction creating a solution that steps outside the normal security paradigm for a high-security organisation should not be underestimated. Technology to create a robust solution exists and is  commercially heavily used.

Technorati Tags: ,,
Advertisements

2 Comments »

  1. Interesting article Colin.
    It made me think about how secure our webmail server is (pretty good I reckon).

    I thought of another webmail security solution and was wondering what you think about it.
    Instead of using a web browser to go straight into a company’s webmail, would it perhaps be more secure to load a virtual desktop through Microsoft’s new Remote Desktop (terminal Services) virtualisation solutions? Or would this be susceptible to the same security risks?

    Comment by Thom — October 28, 2009 @ 9:38 am

    • There’s obviously a functionality hit, as not all organisations will allow RDP out to the internet. But if they do, it obviously reduces problems with installed applications on the client PC.

      From a security PoV, I don’t see major benefits. There’s possibly slightly less data remnance issues because the only thing left lying on the client PC is likely to be RDP cache rather than HTML files and attachments. But that’s tempered by a protocol that’s considerable more functional, unless you whack yout terminal servers in the DMZ too…

      Comment by chaplic — October 28, 2009 @ 10:54 am


RSS feed for comments on this post. TrackBack URI

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Create a free website or blog at WordPress.com.

%d bloggers like this: