With the end of the 30 day Cybersecurity Sprint that U.S. CIO Tony Scott announced on June 12th rapidly approaching, I thought it would be a good time to look both the 30 day goals as well as the goals of the Cybersecurity Sprint team that will report out at the end of the sprint.
Four focus areas that must be reported on to the Department of Homeland Security (DHS) and to the Office of Management and Budget (OMB) by the date are:
- Fix all critical vulnerabilities within 30 days (from the issuance of the May 21 directive from the Department of Homeland Security)
- Tighten policies for privileged users
- Accelerate the adoption of multi-factor authentication – especially for privileged users but also for wider employee sets
- Immediately deployment of indicators supplied by DHS to scan systems/logs to detect attack or the possibility of a breach (possibly this is the Einstein 3A software discussed here)
I have to think that most of what we’ll see is “reporting” at the end of the sprint. Given the scope of these changes, we can’t realistically expect that organizations will meet all of these requirements in time for the deadline.
Fix all critical vulnerabilities. This is harder than it looks, unless machines are enrolled in with a configuration management system. OS patches can be easily enough done with standard patch and update software, but application and configuration vulnerabilities depend on application software, and on organizations understanding how to find, and deploy security level patches for the apps. If the definition of “critical vulnerabilities” is narrow enough, the window can be met. But software and processes have to be in place for applications as well. It isn’t enough to just roll out changes as they come in from the application vendor, as there are too many dependencies in production systems. A full change management process has to be followed to reduce the risk that untested patches to application level software will crash a critical application. There is a good chance that internet facing systems (as the most critical ones) can have the majority of critical vulnerabilities found and fixed, but full compliance for all internet facing systems, and then all systems connected behind them on the network will be longer than a 30 day process.
Privileged users - Privileged user “policies” for how they operate (on paper) for instance, can change, but enforcement requires selection, purchase and deployment of solutions to the problem. What’s more, implementing the hardware and software required for multi-factor authentication of privileged user accounts has the dual problem of getting enough hardware and enforcement on the software platforms. While the software side is more easily done for domain level roles, system level roles (such as that of root users on UNIX and Linux systems) will require a different approach. So while a start can be made, fully implementing these policies will be time consuming and expensive.
Multi factory authentication – While it is “somewhat” more feasible to at least get the hardware for privileged users, It also isn’t realistic to expect that the millions of federal employees without smart cards or other multi-factor authentication solutions can have something in place by the deadline. Just the hardware purchase and procurement problems prevent it. Something simple along the lines of what Google does with account access – requiring an authentication code sent by via a text might happen sooner, but definitely is not as secure.
Scan networks for vulnerabilities and apply software that can detect some attacks in progress – This is perhaps the most feasible of the lot to complete in time – Since the software tools are available and appear to require only log level access.
The Cybersecurity Sprint team – With their 30 day review of the Federal Government’s cybersecurity policies, procedures, and practices – Hopefully they will be coming up with a phased approach to changing our IT security stance to be more proactive.
So to sum it up - What’s likely change by the deadline?
- Networks and systems will get scanned by the Einstein 3A tool (or the other tools required under the program) for signs of attacks previously, or of an attack in progress.
- Internet facing systems and applications will be properly patched and updated for current threats, at least at the system level, if not at the application level.
- Procedures for privileged user access will be drawn up and put in operation at the “paper” level, while enforcement software lags – and some privileged user roles will require multi-factor authentication that did not previously.
- The rest will have to be “best efforts”, reporting and planning.
And what about the next step after the deadline? The OMB says that the Federal CIO (Scott) will create and operationalize a set of action plans and strategies to further address critical cybersecurity priorities and recommend a Federal Civilian Cybersecurity Strategy around a set of key principles:
- Protecting data: Better protect data at rest and in transit.
- Improving situational awareness: Improve indication and warning.
- Increasing cybersecurity proficiency: Ensure a robust capacity to recruit and retain cybersecurity personnel.
- Increase awareness: Improve overall risk awareness by all users.
- Standardizing and automating processes: Decrease time needed to manage configurations and patch vulnerabilities.
- Controlling, containing and recovering from incidents: Contain malware proliferation, privilege escalation, and lateral movement. Quickly identify and resolve events and incidents.
- Strengthening systems lifecycle security: Increase inherent security of platforms by buying more secure systems and retiring legacy systems in a timely manner.
- Reducing attack surfaces: Decrease the complexity and amount of things defenders need to protect.
This is a good list, but I have to think that the priorities need to be on the data that is the target of these attacks. A detailed discovery process needs to identify where it is, lock down the access to it at both system levels (OS and file systems) and from within applications, and then accounts with data access need to be watched. This combination, which is best done with encryption, access controls to encrypted data and then monitoring of access patterns for accounts and users that have this access is the best first step to take to limit the damage from penetrations to the network that WILL HAPPEN and then stop the extraction of data before it becomes too critical.