Posts tagged ‘Symantec’

Full Disk Encryption versus Sleep

As part of my Symantec Endpoint Encryption (SEE) upgrade, I verified that the new version worked with our main computer models.   During that testing, I looked at how boot/shutdown times changed, and verified that the system could still reboot and enter/exist sleep and hibernate correctly.  The only problem that came out of that testing was when standby was used, the user no longer had to do a preboot authentication before logging into Windows.  Previously GuardianEdge Hard Disk  (GEHD) Encryption provided full disk encryption on a cold boot and on returning from standby or hibernate.    Further testing found GEHD 9.5.3 also had this problem, I just didn’t know it.  

Since the cold boot was announced, it has been best practice to not allow computers using Full Disk Encryption (FDE) to enter sleep.  The encryption key typically remains in memory while the computer is in sleep mode and is thus susceptible to the attack.  

As a side note, hibernate is also dangerous if your FDE product doesn’t encrypt the hibernate file.   SEE should not have this issue.

Personally I refrain from using hibernate or sleep.   In my experience, it is unreliable.  So I’m not the most sympathetic person here.    There is always a tradeoff between security and usability.   If sleep is insecure and you fail to disable sleep then you only have the appearance of security.   You’ve checked the encryption box without actually protecting the data.

Management’s first response to this documented need to disable sleep is to ask for a report comparing sleep versus quarantine shutdown times.    In my experience, getting off XP is the first step toward having decent shutdown times.   The second step is reloading the system once it has the crud.    The effect of creeping crud will be more noticeable now that we are on three-year leases. 

For more information on protection against cold boot attacks with Symantec Endpoint Protection see their knowledgebase.

Symantec EndPoint Encryption Installer

Symantec Endpoint Encryption (SEE) 8.0.1 is my first upgrade since Symantec purchased GuardianEdge.  It is a newer version inspite of being a lower number than GuardianEdge 9.5.x.    I guess it is really too soon to expect big changes.   I was hoping they would address some of the installer annoyances.

With SEE, you install a management server, then create the client install packages.   These are MSI packages.

Separate 32 bit and 64 bit Installer Files
I don’t do MSI package generation myself, but the software I’ve seen allows you to put both 32 bit and 64 into one installation file.   I would think this would make things a lot easier.   The main drawback would be the size of the install file.    I end up putting both 32 and 64 bit files into one installation package and call the appropriate one based on the CPU architecture.   So it doesn’t save me space and instead requires extra scripting work.   Is there some technical limitation I’m not aware of?

Upgrades
Upgrading versions of Symantec Endpoint Encryption (or GuardianEdge Hard Disk Encryption) requires using special switches.   When performing an upgrade, I need to use the command line MSIEXEC /i “\.msi REINSTALL=”ALL” REINSTALLMODE=”vomus”.   This upgrades the client by reinstalling all features of the product.

On the other hand, a regular install of Symantec Endpoint Encrytption, where it was not installed previously, uses more familiar switches (/qb).   In previous upgrades I’ve found that if I try to use the reinstall/reinstallmode switches on a fresh install it will not work.   I then have to script the install to use different command line options based on installation status in addition to 32 bit versions 64 bit.

To make matters worse,  some computers in my environment have Removable Storage Encryption and others don’t.   My install script is getting too complicated.

Full Control
When creating the installation package, you must save the client installation package to a local or network volume with Full control permissions set.  The SEE instructions say “This ensures the success of the upgrade package, as it will retain the Windows permissions of the location to which it is saved.”    Again, I don’t create MSI packages often.   Adobe Reader and Symantec Endpoint Encryption both create packages where a setup.exe calls the MSI.   However in neither case am I advised to change permissions on a folder.

I think I actually have seen issues that support has blamed on not creating the package in a directory with Full Control.    But I’m not sure what the actual problem is.

The old Files
In the past, when I’ve upgraded using the switches provided, I found I needed to have all the old install file available.   While people are generally on one version, I found it better to leave every version I’ve ever used in the install directory.   That can take up a lot of room.

I’m really lost as to the cause of this issue.   Shouldn’t MSI files be cached locally so even if it did need the original installation files it should have them, not require me to have them.   I am going to try one more time and removing the old install files from my SEE8.0.1 package and see if I still have issues.   Perhaps that was a problem only with older versions of the product.

Doesn’t Symantec own a MSI packaging company?   Hopefully some in-house expertise can cross divisions of the company to create a better product.

SEP 11 RU6 MP3 Released

Symantec released Maintenance patch 3 for SEP 11.0.6. this week.

Changes and fixes are listed in the Symantec knowledge base.

Release notes are here.

Win7 SP1 SEP Support

Ouch!

Symantec has posted a knowledge base article.   Symantec Endpoint Protection will not support Service Pack 1 for Windows 7 or Windows 2008 R2 until SEP 11.0.7 (11.0 Release update 7).

There are no known issues.   They just aren’t going to certify it until 11.0.7.

Symantec Endpoint Protection 12 Announced

Today Symantec pre-announced Symantec Endpoint Protection 12.  You can sign up for the public beta now, although the beta bits are not immediately available.   It wasn’t stated whether this beta includes the server install or if it is client only. (update  - Good news! Symantec commenter reports beta will be the full install and not client only).   The full release is “later this year.”  

Why are we excited about this?   SEP11 has grown a bit long in the tooth.   While it gave vast performance improvements over Symantec Antivirus 10, the natives are growing restless.    SEP12 offers performance improvement, improved protection and is better designed for the virtualized environments found in many data centers.

The list of what’s new is at the link above, and then click on the what’s new tab.

Migrating FDE Vendors

I was asked recently via email how to pragmatically uninstall GuardianEdge.   I’d been thinking about something similar, that is how do you migrate endpoint security vendors including Full Disk Encryption.

To a certain extent this problem doesn’t affect very many people.   Is Full Disk Encryption installed at many companies outside the Federal Government and Government Contractors?  I imagine its starting to make more inroads via the encryption safe harbor and regulatory requirements.  

I’ve had Full Disk Encryption deployed for over 3 years.   With many security products that is an eternity.   Features change.  Companies get bought and sold.   What if I decide to switch from yellow (Symantec) to red (McAfee).   Does Sophos have a color?  

As far as uninstalling GuardianEdge specifically, I’m pretty sure the manual says you need to decrypt before you uninstall.   Therefore, I would need to deploy a decrypt policy via Group Policy, then after sufficient time has occurred for decryption, uninstall GuardianEdge and replace it with my new favorite Full Disk Encryption.   The problem with this scenario is 1) The computer is left unencrypted for a period of time 2) this period of time is unspecified 3) The end-user will experience the joyful performance hit of decrypting and encrypting the hard drive.    Not Good!

Another possibility is to introduce the new encryption products as computers are replaced.   This has the benefit of not interrupting the user.   The downside is the helpdesk would have to keep track of two different one-time password programs to allow users to access computers with a forgotten password.   Management is twice as hard.   I’d have to maintain two different systems.   With a three-year lease cycle on computers it would be quite a while before all computers are on the new system.

We’re about to do a rip and replace migration to Windows 7.   This would be an ideal time when you’re already doing a system refresh.   You don’t have to worry about the decrypt/uninstall.   You just back up data, drop the Win7 Ghost load, restore data, encrypt.   It is a rare opportunity.

I don’t like these options.   Readers, have any of you migrated Full Disk Encryption products?   Do you see any alternatives I”m missing?   Comments welcomed below.   First time commenters will be held in the moderation queue.   All comments must clear the spam filter.

Wishlist for SEP 12.1

Symantec Endpoint Protection (SEP) 11 is getting long in the tooth.   It was a huge step forward.   But I’m starting to look forward to the next release.   Symantec released a small business edition with version 12.   So I’m calling the next version of SEP, SEP12.1.    That isn’t official.   Here’s a list of what I’d like to see in SEP11

Full 64 Bit Feature Parity
Enough is enough.   With the release of Windows 7, 64 bit is starting to be adopted by regular users.   Some companies have made 64 bit the standard for their Windows 7 corporate rollout.

Symantec does not currently support application and device control on 64 bit.  Companies don’t want to have different levels of security for 32 versus 64 bit computers.    We use the Device control part of Symantec to disable the wireless card when a wired connection is present.   I see that as critical functionality.   This is causing us to be unable to use 64 bit laptops.   Further the helpdesk wanting to hold down complexity seems to be against 32 bit laptops and 64 bit desktops.   To avoid twice the testing they want all 32 or 64 bit computers.

I can no longer find the knowledge base article, but I recall there being less keylogger protection in 64 bit SEP11 due to kernel protections by Microsoft.   Not sure that one could be fixed without hooking the kernel outside of approved APIs.    (not a good idea).

Wireless Management
As I mentioned, I use Application and Device Control to disable wireless cards when wired connected.   This is an important security consideration to prevent the client from being attacked by someone in the parking lot while they are on our network.  

The problem with the current method (besides the 64 bit issue I covered in the last section), is Symantec leaves it up to the SEPM administrator to manually add the device ID for each device they wish to block.   This is decidedly not cool.   Each time we start bringing in a new laptop model I need to update the block rule with the new device ID.   It’s not just wireless cards.   I’d like EVDO/3G wireless modems disabled as well.   Symantec should be doing this in a more automatic way.  

IPv6
Symantec Endpoint Protection 11 does not understand IPv6.   With the built-in firewall you can only allow it or block it at the protocol level.   You can not have rules based on source/destination addresses/ports.   I don’t think I need to belabor the point.   IPv4 address exhaustion is months away according to some reports.   Some ISPs are already conducting IPv6 tests with end users.  

IPv6 support is listed as in development and to be in the next major release.  

To the Cloud
Symantec did rather well in Gartner’s December 2010 Endpoint Protection Magic Quadrant.   I believe the in the cloud protection was even mentioned.   The problem is in the cloud reputation scoring is currently only available for home users.    I believe all of Symantec’s major competitors already use this sort of community scoring as an extra layer of protection, and have for some time.   

With in the cloud protection, there is a community based reputation score assigned to files so they can be treated appropriately.

I understand Symantec is a big company, but it needs to innovate protection, not lag behind while using other parts of the company (consumer) as test beds for new engines and new techniques.

Performance Improvements
I know that Symantec Endpoint Protection was a big step up over Symantec Antivirus 10 in terms of performance.   But that was many years ago.   According to some comparison numbers Endpoint Protection could use some speed improvements.   Not near the top of my list but worth mentioning.

Single Agent/ Single Console
Those of us using GuardianEdge for encryption are hoping to have a unified point of management.   One agent to upgrade.   One less thing to update, one less place to look for reports.

Some of these items are already listed at Symantec Ideas.   Some of them, like IPv6, are already known to be in the next major release.   At Symantec Connect, you can use the Ideas section to suggest a new feature or functionality, and vote or comment on other people’s suggestions.  

I dont have a lot of complaints about SEP.   I do hope that a few of these things get cleared up in the next version.

Adobe Reader X Protected Mode and Antivirus

The sandbox functionality in Adobe Reader X is known to conflict with some antivirus products. 

I’ve installed Reader X at home with no issues.      A post in the Symantec Connect forums indicates Adobe Reader X cannot open on computers that use the Network Threat Protection component of Symantec Endpoint Protection.   The workaround for the moment is to disable Reader’s protected mode.    I don’t use Network Threat Protection at home which is why I didn’t see any issues there.

Opportunistic TLS and MessageLabs

Back in February 2008, I suggested to the Sendmail admins that we look into opportunistic TLS.   Like all encryption there is a performance hit.   Unlike S/MIME or PGP the encryption is only during transit between links.   Additionally there is no guarantee that all links will be encrypted.   Hence the word opportunistic.   While you don’t want to get a false sense of security from it, I don’t see a reason not to implement it on a system that has the performance capacity.  

The Sendmail admins added opportunistic TLS to outbound email pretty quickly.   However they found that to add it for inbound email required recompiling Sendmail.   As a result, this was put on the shelf for a while.   Here we are 2.5 years later, and as part of moving Sendmail to a Solaris blade server, they added opportunistic TLS for inbound email.  

There’s always a but…

We use MessageLabs as our secure email gateway.   I assumed that because I could connect to them on port 25 and initiate the command STARTTLS that meant they supported opportunistic TLS.     The exact phrase I used in Feb 2008 was “I suspect messagelabs would then send our inbound email across a SSL session making our email slightly more secure.”   It turns out my assumptions are incorrect.  

Symantec MessageLabs does not support opportunistic TLS (Solution ID: DA_116296).   Solution ID DA_136900 claims that opportunistic TLS is a security threat rather than a security feature.   Because it only encrypts a connection when it can, unencrypted email can be sent.   This is of course true.   But at a minumum, I would know that the connection from my servers to MessageLabs was encrypted.   Final secure delivery would depend on the configuration of the recipient servers.  

It seems to me that Symantec MessageLabs is trying to force customers to purchase their Boundary Encryption product.

I have been informed via the comments that if MessageLabs receives the message via SMTP/TLS they will attempt to preserve that level of security on delivery to the next hop.   That makes sense.   In most cases adding encryption merely for the last hop is pointless.   Sweetness.  

So the onus is back on other mail providers.   I saw a great rant recently on Gmail not providing opportunistic TLS.   

GuardianEdge Windows 7 Looking Back

Like a lot of companies we are trying to go to Windows 7 sooner rather than later. We skipped Vista and XP is starting to seem a bit old. One of the things holding us back is GuardianEdge’s Full Disk Encryption product. Here’s our timeline.

In October 2009 I asked GuardianEdge about Windows 7 support and Windows 7 64 bit support. They said both would available in version 9.5 due out in December 2009.

When GuardianEdge Hard Disk Encryption 9.5 was released (January or February), I found that there was no support for preboot authentication. Without preboot authentication, I think the encryption is pretty worthless. Support tells me 9.5.1 will include preboot authentication and be available in April 2010.

9.5.1 is released and I find it doesn’t work on my Toshiba Portege with windows 7 32 bit installed. I decide this may be a one-off. I’m the only one using the Toshiba so I try it out on a few Dell E6500 computers with Windows XP and Windows 7. This failed miserably. It turns out this was a known issue with Dell E6500 and GuardianEdge was working on a patch.

GEHD 9.5.1 patch 1 came out. While it fixed the assorted problems with the E6500, I now see in the release notes:

There are known issues with GuardianEdge Hard Disk on various configurations of the following Dell computer models
■ Dell E4310
■ Dell E6410
■ Dell E6510
■ Dell E5410, and
■ Dell E5510

Unfortunately the E6410 and the E6510 are two of the three systems listed on our standard configuration page. The third E4300, I suspect would really be the E4310.

GuardianEdge says this will be fixed in September 2010.

I wouldn’t this be surprised if this led to looking at other solutions and revisiting Bitlocker. I wrote about Bitlocker in March. These pretzels are making me thirsty.