Federated Vulnerability Management

Posted July 14th, 2009 by

Why hello there private sector folks.  It’s no big surprise, I work in the US Federal Government Space and we have some unique challenges of scale.  Glad to meet you, I hear you’ve got the same problems only not in the same kind of scale as the US Federal Government.  Sit back, read, and learn.

You see, I work in places where everything is siloed into different environments.  We have crazy zones for databases, client-facing DMZs, managment segments, and then the federal IT architecture itself: a loose federation of semi-independent enterprises that are rapidly coming together in strange ways under the wonderful initiative known as “The TIC”.  We’re also one of the most heavily audited sectors in IT.

And yet, the way we manage patch and vulnerability information is something out of the mid-80’s.

Current State of Confusion

Our current patch management information flow goes something like this:

  • Department SOC/US-CERT/CISOs Office releases a vulnerability alert (IAVA, ISVM, something along those lines)
  • Somebody makes a spreadsheet with the following on it:
    • Number of places with this vulnerability.
    • How many have been fixed.
    • When you’re going to have it fixed.
    • A percentage of completion
  • We then manage by spreadsheets until the spreadsheets say “100%”.
  • The spreadsheets are aggregated somewhere.  If we’re lucky, we have some kind of management tool that we dump our info into like eMASS.
  • We wonder why we get pwned (by either haxxorz or the IG).

Now for how we manage vulnerability scan information:

  • We run a tool.
  • The tool spits out a .csv or worse, a .html.
  • We pull up the .csv in Excel and add some columns.
  • We assign dates and responsibilities to people.
  • We have a weekly meeting togo over what’s been completed.
  • When we finish something, we provide evidence of what we did.
  • We still really don’t know how effective we were.

Problems with this approach:

  • It’s too easy to game.  If I’m doing reporting, the only thing really keeping me reporting the truth is my sense of ethics.
  • It’s slow as hell.  If somebody updates a spreadsheet, how does the change get echoed into the upstream spreadsheets?
  • It isn’t accurate at any given moment in time, mostly because changes quicker than the process can keep up.  What this means is that we always look like liars who are trying to hide something because our spreadsheet doesn’t match up with the “facts on ground”.
  • It doesn’t compare with our other management tools like Plans of Action and Milestone (POA&M).  They usually are managed in a different application than the technical parts, and this means that we need a human with a spreadsheet to act as the intermediary.

So this is my proposal to “fix” government patch and vulnerability management: Federated Patch and Vulnerability Management through SCAP.

Trade Federation Battle Droid photo by Stéfan.  Roger, Roger, SCAP means business.

Whatchu Talkin’ Bout With This “Federated” Stuff, Willis?

This is what I mean, my “Plan for BSOFH Happiness”:

Really what I want is every agency to have an “orchestrator” ala Ed Bellis’s little SCAP tool of horrors. =)  Then we federate them so that information can roll up to a top-level dashboard for the entire executive branch.

In my beautiful world, every IT asset reports into a patch management system of some sort.  Servers, workstations, laptops, all of it.  Yes, databases too.  Printers–yep.  If we can get network devices to get reported on config info via SCAP-enabled NMS, let’s get that pushing content into our orchestrator. We don’t even really  have to push patches using these tools–what I’m primarily concerned with at this point is to have the ability to pull reports.

I group all of my IT assets in my system into a bucket of some sort in the orchestrator.  That way, we know who’s responsible when something has a problem.  It also fits into our “system” concept from FISMA/C&A/Project Management/etc.

We do periodic network scanning to identify everything on our network and feed them into the orchestrator.  We do regular vulnerability scans and any findings feed into the orchestrator.  The more data, the better aggregate information we can get.

Our orchestrator correlates network scans with patch management status and gives us a ticket/alert/whatever where we have unmanaged devices.  Yes, most enterprise management tools do this today, but the more scan results I have feeding them, the better chance I have at finding all my assets.  Thanks to our crazy segmented architecture models, we have all these independent zones that break patch, vulnerability, and configuration management as the rest of the IT world performs it.  Flat is better for management, but failing that, I’ll take SCAP hierarchies of reporting.

The Department takes a National Vulnerability Database feed and pushes down to the Agencies what they used to send in an IAVA, only they also send down the check to see if your system is vulnerable.  My orchestrator automagically tests and reports back on status before I’m even awake in the morning.

I get hardening guides pushed from the Department or Agency in SCAP form, then pull an audit on my IT assets and have the differences automagically entered into my workflow and reporting.

I become a ticket monkey.  Everything is in workflow.  I can be replaced with somebody less expensive and can now focus on finding the answer to infosec nirvana.

We provide a feed upstream to our Department, the Department provides a feed to somebody (NCSD/US-CERT/OMB/Cybersecurity Coordinator) who now has the view across the entire Government.  Want to be bold, let Vivek K and the Sunlight Foundation at the data feeds and have truly open and transparent, “Unbreakable Government 2.1”.  Who needs FISMA report cards when our vulnerability data is on display?

Keys to Making Federated Patch and Vulnerability Management Work

Security policy that requires SCAP-compatible vulnerability and patch management products.  Instead of parroting back 800-53, please give me a requirement in your security policy that every patch and vulnerability management tool that we buy MUST BE SCAP-CERTIFIED.  Yes, I know we won’t get it done right now, but if we get it in policy, then it will trickle down into product choices eventually.  This is compliance I can live with, boo-yeah!

Security architecture models (FEA anyone?) that show federated patch and vulnerability management deployments as part of their standard configuration.  OK with the firewall pictures and zones of trust, I understand what you’re saying, now give me patch and vulnerability management flows across all the zones so I can do the other 85% of my job.

Network traffic from the edges of the hierarchy to…somewhere.  OK, you just need network connectivity throughout the hierarchy to aggregate and update patch and vulnerability information, this is basic data flow stuff.  US-CERT in a future incarnation could be the top-level aggregator, maybe.  Right now I would be happy building aggregation up to the Department level because that’s the level at which we’re graded.

Understanding.  Hey, I can’t fix everything all the time–what I’m doing is using automation to make the job of fixing things easier by aggregation, correlation, status reporting, and dashboarding.  These are all concepts behind good IT management, why shouldn’t we apply them to security managment also?  Yes, I’ll have times when I’m behind on something or another, but guess what, I’m behind today and you just don’t know it.  However, with near-real-time reporting, we need a culture shift away from trying to police each other up all the time to understanding that sometimes nothing is really perfect.

Patch and vulnerability information is all-in.  It has to be reporting in 100% across the board, or you don’t have anything–back to spreadsheets hell for you.  And simply put, why don’t you have everything in the patch management system already?  Come on, that’s not a good enough reason.

POA&Ms need to be more fluid.  Face it, with automated patch and vulnerability management, POA&Ms become more like trouble tickets.  But yes, that’s much awesome, smaller, easily-satisfied POA&Ms are much easier to manage provided that the administrative overhead for each of these is reduced to practically nothing… just like IT trouble tickets.

Regression testing and providing proof becomes easier because it’s all automated.  Once you fix something and it’s marked in the aggregator as completed, it gets slid into the queue for retesting, and the results become the evidence.

Interfaces with existing FISMA management tools.  This one is tough.  But we have a very well-entrenched software base geared around artifact management, POA&M management, and Security Test and Evaluation results.  This class of software exists because none of the tools vendors really understand how the Government does security management, and I mean NONE of them.  There has to be some weird unnatural data import/export acts going on here to make the orchestrator of technical data match up with the orchestrator of managment data, and this is the part that scares me in a federated world.

SCAP spreads to IT management suites.  They already have a footprint out there on everything, and odds are we’re using them for patch and configuration management anyway.  If they don’t talk SCAP, push the vendor to get it working.

Where Life Gets Surreal

Then I woke up and realized that if I provide my Department CISO with near-real-time patch and vulnerability mangement information, I suddenly have become responsible for patch and management instead of playing “kick it to the contractors” and hiding behind working groups.  It could be that if I get Federated Patch and Vulnerabilty Management off the ground, I’ve given my Department CISO the rope to hang me with.  =)

Somehow, somewhere, I’ve done most of what CAG was talking about and automated it.  I feel so… um… dirty.  Really, folks, I’m not a shill for anybody.

Similar Posts:

Posted in DISA, NIST, Rants, Technical | 12 Comments »

Security Automation Developers Conference Slides

Posted July 2nd, 2009 by

Eh? What’s that mean?  Developer Days is a weeklong conference where they get down into the weeds about the various SCAP schemas and how they fit into the overall program of security automation. 

Highlights and new ideas:

Remedial Markup Language: Fledgeling schema to describe how to remediate a vulnerability.  A fully automated security system would scan and then use the RML content to automagically fix the finding… say, changing a configuration setting or installing a patch.  this would be much awesome if combined with the CVE/CWE so you have a vulnerability scanner that scans and fixes the problem.  Also needs to be kept in a bottle because the operations guys will have a heartattack if we are doing this without any human intervention.

Computer Network Defense: There is a pretty good scenario slide deck on using SCAP to automate hardening, auditing, monitoring, and defense.  The key from this deck is how the information flows using automation.

Common Control Identifier:  This schema is basically a catalog of controls (800-53, 8500.2, PCI, SoX, etc) in XML.  The awesomeness with this is that one control can contain a reference implementation for each technology and the checklist to validate it in XCCDF.  At this point, I get all misty…

Open Checklist Interactive Language: This schema is to capture questionaires.  Think managerial controls, operational controls, policy, and procedure captured in electronic format and fed into the regular mitigation and workflow tools that you use so that you can view “security of the enterprise at a glance” across technical and non-technical security.

Network Event Content Automation Protocol:  This is just a concept floating around right now on using XML to describe and automate responses to attacks.  If you’re familiar with ArcSight’s Common Event Format, this would be something similar but on steroids with workflow and a pony!

Attendance at developer days is limited, but thanks to all the “Powar of teh Intarwebs, you can go here and read the slides!

Similar Posts:

Posted in NIST, Technical | 3 Comments »

Sir Bruce Mentions FDCC, World Goes Nuts

Posted May 7th, 2009 by

Check out this blog post.  Wow, all sorts of crazies decend out of the woodwork when Bruce talks about something that’s been around for years and suddenly everyone’s redesigning the desktop from the ground up.

Quick recap on comments:

  • 60-day password changes suck
  • You can do this at home, the GPOs are available from NIST
  • My blue-haired sheepdog can’t use the FDCC image, it’s broken for commercial use!
  • You wouldn’t have to do this in Linux
  • Linux is teh suxx0rz
  • My computer started beeping and smoke came out of it, is this FDCC?

Proving once again that you can’t talk about Windows desktop security without it evolving into a flamewar.  Might as well pull out “vi v/s emacs” while you’re at it, Bruce.  =)

Computer Setup photo by karindalziel.  Yes, one of them is a linux box, I used this picture for that very same reason.  =)

But there is one point that people need to understand.  The magic of FDCC is not in the fact that the Government used its IT-buying muscle to get Microsoft to cooperate.  Oh no, that’s to be expected–the guys at MS are used to working with a lot of people now on requests.

The true magic of FDCC is getting the application vendors to play along.  To wit:

  • The FDCC GPOs are freely available from NIST
  • You can download images from NIST with a preconfigured FDCC setup
  • Application vendors can test their product against FDCC in their own lab
  • There is no external audit burden (yet, it might be coming) for software vendors because it’s a self-certification
  • FDCC-compatible software doesn’t require administrative privileges

In other words, if your software works with FDCC, it’s probably built to run on a security-correct operating system in the first place.  This is a good thing, and in this case the Government is using its IT budget to bring the application vendors into some sort of minimal security to the rest of the world.

This statement is from the FDCC FAQ, comments in parenthesis are mine:

“How are vendors required to prove FDCC compliance?
There is no formal compliance process; vendors of information technology products must self-assert FDCC compliance. They are expected to ensure that their products function correctly with computers configured with the FDCC settings. The product installation process must make no changes to the FDCC settings. Applications must work with users who do not have administrative privileges, the only acceptable exception being information technology management tools. Vendors must test their products on systems configured with the FDCC settings, they must use SCAP validated tools with FDCC Scanner capability to certify their products operate correctly with FDCC configurations and do not alter FDCC settings. The OMB provided suggested language in this memo: http://www.whitehouse.gov/omb/memoranda/fy2007/m07-18.pdf, vendors are likely to encounter similar language when negotiating with agencies.”

So really what you get out of self-certification is something like this:

Similar Posts:

Posted in Technical | 4 Comments »

NIST Framework for FISMA Dates Announced

Posted April 10th, 2009 by

Some of my friends (and maybe myself) will be teaching the NIST Framework for FISMA in May and June with Potomac Forum.   This really is an awesome program.  Some highlights:

  • Attendance is limited to Government employees only so that you can talk openly with your peers.
  • Be part of a cohort that trains together over the course of a month.
  • The course is 5 Fridays so that you can learn something then take it back to work the next week.
  • We have a Government speaker ever week, from the NIST FISMA guys to agency CISOs and CIOs.
  • No pitching, no marketing, no product placement (OK, maybe we’ll go through DoJ’s CSAM but only as an example of what kinds of tools are out there) , no BS.

See you all there!

Similar Posts:

Posted in NIST, Speaking | 1 Comment »

Et Tu, TIC?

Posted October 7th, 2008 by

Let’s talk about TIC today, dear readers, for I smell a conspiracy theory brewing.

For those of you who missed the quick brief, TIC is short for “Trusted Internet Connections” and is an architecture model/mandate/$foo to take all of the Internet connections in the Government (srsly, nobody knows how many of them really exist, but it’s somewhere in the 2,000-10,000 range) and consolidate them into 50.  These connections will then be monitored by DHS’s Einstein program.

No, Not That Kind of TIC photo by m.prinke.

Bringing you all up to date, you’ll need to do some homework:

Now having read all of this, some things become fairly obvious:

  • If you have the following people needing connections:
    • 24 agencies, plus
    • DoD with 2 points of presence, plus
    • Intelligence agencies with a handful of Internet connections, means that:
  • That basically, everybody gets one Internet connection.  This is not good, it’s all single point-of-DOS.
  • Agencies have been designated as Internet providers for other agencies.  Sounds like LoB in action.
  • Given the amount of traffic going through the TIC access points, it most likely is going to take a significant amount of hardware to monitor all these connections–maybe you saved 50% of the monitoring hardware by reducing the footprint, but it’s still hardware-intensive.
  • TIC is closely tied with the Networx contract.
  • In order to share Internet connections, there needs to be a network core between all of the agencies so that an agency without a TIC access point can route through multiple TIC service provider agencies.

And this is where my conspiracy theory comes in:  TIC is more about making a grand unified Government network than it is monitoring events–Einstein is just an intermediate goal.   If you think about it, this is where the Government is headed.

We were headed this way back in ought-two with a wonderful name: GovNet.  To be honest, the groundwork wasn’t there and the idea was way ahead of its time and died a horrible death, but it’s gradually starting to happen, thanks to TIC, FDCC, and Einstein. 

More fun links:

If you want to get a reaction out of the OMB folks, mention GovNet and watch them backpedal and cringe,–I think the pain factor was very high for them on GovNet. So I think that we should, as a cadre of information security folks, start calling TIC what it really is:  Govnet 2.0!  =)

Similar Posts:

Posted in Technical | 2 Comments »

Comments on SCAP 2008

Posted September 24th, 2008 by

I just got back from the SCAP 2008 conference at NIST HQ, and this is a collection of my thoughts in a somewhat random order:

Presention slides are available at the NVD website

I blogged about SCAP a year ago, and started pushing it in conversations with security managers that I came across.  Really, if you’re managing security of anything and you don’t know what SCAP is, you need to get smart on it really fast, if for no other reason than that you will be pitched it by vendors sporting new certifications.

Introduction to SCAP:  SCAP is a collection of XML schemas/standards that allow technical security information to be exchanged between tools.  It consists of the following standards:

  • Common Platform Enumeration (CPE): A standard to describe a specific hardware, OS, and software configuration.  Asset information, it’s fairly humdrum, but it makes the rest of SCAP possible–think target enumeration and you’re pretty close.
  • Common Vulnerabilities and Exposures (CVE): A definition of publicly-known vulnerabilities and weaknesses.  Should be familiar to most security researches and patch monkies.
  • Common Configuration Enumeration (CCE): Basically, like CVE but specific to misconfigurations.
  • Common Vulnerability Scoring System (CVSS): A standard for determining the characteristics and impact of security vulnerabilities.  Hmmm, sounds suspiciously like standardization of what is a high, medium, and low criticality vulnerability.
  • Open Vulnerability and Assessment Language (OVAL):  Actually, 3 schemas to describe the inventory of a computer, the configuration on that computer, and a report of what vulnerabilites were found on that computer.
  • Extensible Configuration Checklist Description Format (XCCDF): A data set that describes checks for vulnerabilities, benchmarks, or misconfigurations.  Sounds like the updates to your favorite vulnerability scanning tool because it is.

Hall of Standards inside NIST HQ photo by ME!!!

What’s the big deal with SCAP: SCAP allows data exchanges between tools.  So, for example, you can take a technical policy compliance tool, load up the official Government hardening policy in XCCDF for, say, Windows 2003, run a compliance scan, export the data in OVAL, and load the results into a final application that can help your CISO keep track of all the vulnerabilities.  Basically, imagine that you’re DoD and have 1.5 million desktops–how do you manage all of the technical information on those without having tools that can import and export from each other?

And then there was the Federal Desktop Core Configuration (FDCC): OMB and Karen Evans handed SCAP its first trial-by-fire.  FDCC is a configuration standard that is to be rolled out to every Government desktop.  According to responses received by OMB from the departments in the executive branch (see, Karen, I WAS paying attention =)   ), there are roughly 3.5 Million desktops inside the Government.  The only way to manage these desktops is through automation, and SCAP is providing that.

He sings, he dances, that Tony Sager is a great guy: So he’s presented at Black Hat, now SCAP 2008 (.pdf caveat).  Basically, while the NSA has a great red-team (think pen-test) capability, they had a major change of heart and realized, like the rest of the security world (*cough*Ranum*cough*), that while attacking is fun, it isn’t very productive at defending your systems–there is much more work to be done for the defenders, and we need more clueful people doing that.

Vendors are jumping on the bandwagon with both feet: The amount of uptake from the vulnerability and policy compliance vendors is amazing.  I would give numbers of how many are certified, but I literally get a new announcement in my news reader ever week or so.  For vendors, being certified means that you can sell your product to the Government, not being certified means that you get to sit on the bench watching everybody else have all the fun.  The GSA SAIR Smart-Buy Blanket Purchase Agreement sweetens the deal immensely by having your product easily purchasable in massive quantities by the Government.

Where are the rest of the standards: Yes, FDCC is great, but where are the rest of the hardening standards in cute importable XML files, ready to be snarfed into my SCAP-compliant tool?  Truth be told, this is one problem with SCAP right now because everybody has been focusing on FDCC and hasn’t had time yet to look at the other platforms.  Key word is “yet” because it’s happening real soon now, and it’s fairly trivial to convert the already-existing DISA STIGs or CIS Benchmarks into XCCDF.  In fact, Sun was blindsided by somebody who had made some SCAP schemas for their products and they had no idea that anybody was working on it–new content gets added practically daily because of the open-source nature of SCAP.

Changing Government role: This is going to be controversial.  With NVD/CVE, the government became the authoritative source for vulnerabilities.  So far that’s worked pretty well.  With the rest of SCAP, the Government changes roles to be a provider of content and configurations.  If NIST is smart, they’ll stay out of this because they prefer to be in the R&D business and not the operations side of things.  Look for DHS to pick up the role of being a definitions provider.  Government has to be careful here because they could in some instances be competing with companies that sell SCAP-like feed services.  Not a happy spot for either side of the fence.

More information security trickle-down effect: A repeated theme at SCAP 2008 is that the public sector is interested in what Big SCAP can do for them.  The vendors are using SCAP certification as a differentiator for the time being, but expect to see SCAP for security management standards like PCI-DSS, HIPAA, and SOX–to be honest here, though, most of the vendors in this space cut their teeth on these standards, it’s just a matter of legwork to be able to export in SCAP schemas.  Woot, we all win thanks to the magic that is the Government flexing its IT budget dollars!

OS and Applications vendors: these guys are feeling the squeeze of standardization.  On one hand, the smart vendors (Oracle, Microsoft, Sun, Cisco) have people already working with DISA/NSA to help produce the configuration guides, they just have to sit back and let somebody turn the guides into SCAP content.  Some of the applications vendors still haven’t figured out that their software is about to be made obsolete in the Government market because they don’t have the knowledge base to self-certify with FDCC and later OS standards.  With a 3-year lead time required for some of the desktop applications before a feature request (make my junk work with FDCC) makes it into a product release, there had better be some cluebat work going on in the application vendor community.  Adobe, I’m talking to you and Lifecycle ES–if you need help, just call me.

But how about system integrators: Well, for the time being, system integrators have almost a free ride–they just have to deal with FDCC.  There are some of them that have some cool solutions built on the capabilities of SCAP, but for the most part I haven’t seen much movement except for people who do some R&D.  Unfortunately for system integrators, the Federal Acquisition Regulation now requires that anything you sell to the Government be configured IAW the NIST checklists program.  And just how do you think the NIST checklists program will be implemented?  I’ll take SCAP for $5Bazillion, Alex.  Smart sytem integrators will at least keep an eye on SCAP before it blindsides them 6 months from now.

Technical compliance tools are destined to be a commodity: For the longest time, the vulnerability assessment vendors made their reputation by having the best vulnerability signatures.  In order to get true compatibility across products, standardized SCAP feeds means that the pure-play security tools are going to have less things to differentiate themselves from all the other tools and they fall into a commodity market centered on the accuracy of their checks with reduced false positives and negatives.  While it may seem like a joyride for the time being (hey, we just got our ticket to sell to the Gubmint by being SCAP-certified), that will soon turn into frustration as the business model changes and the margins get smaller.  Smart vendors will figure out ways to differentiate themselves and will survive, the others will not.

Which leads me to this: Why is it that SCAP only applies to security tools?  I mean, seriously, guys like BigFix and NetIQ have crossover from technical policy compliance to network management systems–CPE in particular.  What we need is a similar effort applied to network and data center tools.  And don’t point me at SNMP, I’m talking rich data.  =)  On a positive note, expect some of the security pure-play tools to be bought up and incorporated into enterprise suites if they aren’t already.

Side notes:

I love how the many deer (well over 9000 deer on the NIST campus) all have ear tags.  It brings up all sorts of scientific studies ideas.  But apparently the deer are on birth control shots or something….

Former Potomac Forum students:  Whattayaknow, I met some of our former students who are probably reading this right now because I pimped out my blog probably too aggressively.  =)  Hi Shawn, Marc, and Bob!

Old friends:  Wow, I found some of them, too.  Hi Jess, Walid, Chris, and a cast of thousands.

Deer on NIST Gaithersburg Campus photo by Chucka_NC.

Similar Posts:

Posted in DISA, FISMA, NIST, Technical, What Works | 2 Comments »

« Previous Entries

Visitor Geolocationing Widget: