Virtualization in the CIP Environment Drives Discussion of Applicable Systems Classifications

Here's a draft of a Whitepaper I intend to submit to NERC.

Alt Title:

More Problems With EACMS

Summary:

One of the topics entwined in the NEC CIP virtualization discussion is the risk posed by the virtual management consoles. This is related to consolidated interfaces and automation where management console access can change or delete entire infrastructures including virtual servers, networks, and storage. A short-hand phrase has been coined calling this the “Fewer, Bigger Buttons” problem.

Because this is a valid concern that is not really addressed at all in NERC CIP, the scope of discussion quickly grew to address similar Centralized Management Systems (CMS) in the physical arena as well, and from there to all kinds of systems which pose, or appear to pose similar risks.

Statement of the Issues:

The 2016-02 SDT has been discussing several options publicly. One option is to classify CMS (a heretofore undefined term in NERC CIP, and not an industry standard definition in Cyber Security either) as applicable systems and apply specific security requirements to them.

Another option the SDT has discussed is including CMS into the existing EACMS definition. This is more or less by default the approach taken to date with all of those legacy enterprise automation tools. Obviously, this capability and risk has existed, largely unacknowledged in CIP standards, for a long time without being confined to virtualized environments. HP Openview, IBM Tivoli, Solarwinds Orion, and CiscoWorks (to name just a few enterprise automation tools) have had the ability to affect the entire enterprise all at once for literally decades now.

Anti-malware and software patching systems likewise. If a system in my network operates via a system account with administrative privileges that allow it to modify the configuration of a BCS, isn't that tool both a target and a potential attack vector? And if such a system inside my network that performs this function is an EACMS, then isn’t Microsoft’s Software Update Services site, and Ubuntu’s Linux Repository, or Symantec or McAfee’s antivirus signature update sites on the Internet as much an EACMS as any SCCM or Anti-virus server inside my network?

However, this perpetuates and exacerbates an issue where EACMS has become a "catch-all" category of CIP-related Cyber Assets with one-size-fits-all requirements regardless of the degree of risk or technical constraints posed by the particular system.

An example is the Intermediate System. In order to make the IS subject to CIP requirements it has to be categorized somehow as a type of applicable system. To function as an intermediate in practical terms (and by definition per the NERC Glossary) it has to be outside the ESP. Apparently the IS has therefore been categorized as an EACMS simply because that is the only category currently available that allows for applicable systems to be subject to CIP requirements outside the ESP.

Thus, the otherwise-unconnected phrase “This includes Intermediate Systems” was tacked onto the end of the EACMS definition. It is notable that no other examples had previously been given.

The Definition of Intermediate Systems:

“A Cyber Asset or collection of Cyber Assets performing access control to restrict Interactive Remote Access to only authorized users. The Intermediate System must not be located inside the Electronic Security Perimeter”

Here we see a presumably audit-able CIP requirement set in the definition of Intermediate System rather than in a table of requirements or in a security objective. We see a cyber security function (Authentication and Authorization) defined as a specific type of applicable device, and we see the security benefit of such a function truncated to apply only to users who interactively access Cyber Assets inside the ESP from outside the ESP rather than applying the benefit of robust authentication, authorization and accounting to all remote access.

Another example of the problem with the one-size-fits-all approach to compliance requirements for EACMS is what is known as the “hall of mirrors” effect. Specifically, there may be some types of Ecyber Security Systems that should be required to be protected behind a firewall. However, that requirement can’t exist for all EACMS without defining a new category because a firewall is itself an EACMS. Without defining a new category, the result would be every EACMS needing to be inside an ESP and protected by another EACMS which creates a recursive "hall of mirrors" effect without end.

In addition to the catch-all and recursive problems I've just noted, there is also a missing component: Risk-based assessment and mitigation. For example, a system that only monitors and logs access (such as monitoring systems like Splunk, Tripwire, etc) does not pose the same level of risk as a management console for a large virtualized Control Center infrastructure. In addition, the technical controls to mitigate the risk may differ. A SIEM system presents a risk of leaking BES Cyber System Information; an electronic access control (AAA) system presents a risk of unauthorized access to or modification of a BES Cyber System’s operational parameters; and a Centralized Management Console (physical or virtual) presents an infrastructure reliability risk.

Risk-based assessments and mitigation would help with this. It would allow for acknowledging that there are a number of systems that both monitor and provide part of the solution of controlling access, but which do not actually control traffic at the point of entry. These devices or systems may or may not benefit from being inside a protected boundary, or they may form part of the strategy that protects BES Cyber Assets. The technical means of implementing some multi-part systems may require that components be outside or that they span the ESP.

All of this appears to be an unfortunate artifact of the single-level security mindset inherent to the ESP approach (hardened perimeter defense) and the “All-In” nature of CIP Applicable Systems. Part of this problem (creeping scope of applicability to increasingly peripheral systems) stems from using the ESP to define scope of applicability, rather than using risk to BES Cyber Systems and security objectives to define the scope of necessary controls. FERC does not allow a Registered Entity to assess and accept their own risks due to the interconnected nature of risk to the BES. At the same time, NERC and the Regions have zero incentive to accept risks on behalf of Entities. NERC and the Regions bear none of the cost of mitigation, and would receive the lion's share of criticism in the case of a failure of reliability or security breach. It is very difficult to create standards that are effective, comprehensible, and inclusive of different technical capabilities, based upon the existing definition of EACMS or even new definitions structured with the same ESP-as-hardened-perimeter mindset. ESP is easy to visualize. Drawing a “red dotted line” around assets needing protection is convenient. It’s simply not sufficient. In contrast, a modern defense-in-depth, systems- rather than device-based approach doesn’t require torturing definitions.

With these examples in mind, the issues caused by having EACMS as a “catch all” category are obvious. Some applicable systems were defined into existence for compliance purposes and these standards are incompatible with broader standardized cyber security best practices, and with one-size-fits-all requirements applied to the whole artificial group.

Recommendations:

So what do we do about it?

1. The audit process needs to envision an approach more concerned with meeting an objective than a performance requirement. Then it doesn’t matter where something resides, as long as the applied combination of protections achieves the objective. It doesn’t matter much what is providing the control, as long as the assets needing protection receive the control. For example, the implied security objective of CIP 5 is not to “have inbound and outbound rules”. That is a limited method of achieving the real objective, which is to protect the BCS from unauthorized and potentially malicious traffic. If you have a better method, you should be allowed to use it.

2. It might be necessary to break up the EACMS category of applicable systems into discrete functions (bullet points below) so that the appropriate security objective and requirements for each can be derived and applied whether the systems in question are physical or virtual.

• SIEM - Security Incident & Event Monitoring systems.

This would subtract the “M” for “Monitoring” from EACMS. These are systems that strictly monitor and collect information about the ESP and BES Cyber System electronic communications or status but do not control access. A great deal of literature, discussion, guidance, and best practices are published across a broad range of industries as to how to securely implement SIEM. The risk presented by compromise of these systems revolves around the information (such as configurations and event logs) they contain. The crux of the concept here is that the protections already defined for BES Cyber Systems Information (BCSI) are adequate and effective at providing protection for this information, and it is the information that needs protecting, not necessarily the SIEM system.

A rather large issue with EACMS is CIP-004 and its applicability of most personnel-oriented controls (training, background checks, etc.) to anyone with potential access to an EACMS. This kills sharing any service whatsoever (AAA, SIEM, etc.) because anyone with any form of “access” to an EACMS gets sucked into CIP-004. Have an account on a EACMS AAA server but NO access to any BCS? Too bad, you still must have a CIP background check and be trained on the CIP program (beyond your ‘need to know’). It’s a symptom of the “all-in” issue.

Splitting these monitoring systems out and adjusting the requirements would may allow entities to more easily use outsourced managed security service providers or global/enterprise-wide SIEM systems and correlate event information in their CIP operational environments with those in their non-CIP environments to provide increased security and reliability benefits. The concern is that under current standards the CIP program and device-level CIP audits might be deemed to encompass Cyber Assets which do not actually affect the reliable operation of the BES in the wider enterprise network or at the service provider and could therefore dilute attention from BES Cyber Security functions in favor of paperwork exercises.

• EACS - Electronic Access Control System. The fundamental defining characteristic of an Electronic Access Control system is that it performs authorization of traffic or users. This is the gatekeeper function- the classic Authentication and Authorization functions of standard AAA.

In many cases these systems do not perform any active filtering of the traffic passing through any particular interface. The primary duty of EACS is to authenticate and authorize. Additional components of electronic access security strategies are accounting (logging) systems and gateways which actually pass or drop traffic. In comparison to the SIEM discussion above, EACS & EAG move beyond the risk of unauthorized access to meta-information about an environment to unauthorized access to and modification of operational parameters of the actual BES Cyber Systems.

In contrast, theoretically an application level IPS could send a control signal from anywhere in the enterprise to anywhere in the enterprise telling a specific host not to respond to a given type of packet, with a certain payload, from a particular address- and it could do this dynamically based upon heuristics rather than signatures, but this outstanding security solution would not be a compliant solution under our current regime.

Or a network level intrusion protection system combined with dynamic firewall rules may send a control message to a firewall instructing it to dynamically change an Access Control List (ACL) in response to traffic patterns indicating a threat. The IPS does not itself filter traffic in this scenario, but it is involved in controlling access. An Active Directory server may enforce a lock-out on a user account after hours or subsequent to a number of incorrect password attempts. It is involved in controlling access (authentication and authorization), but it is not blocking traffic at layer 3 and from the current NERC CIP ESP perspective is therefore irrelevant in boundary protection. All boundary protection requires access controls, but not all access control is boundary protection.

A metaphor for Access Control Systems that do not reside in or on an ESP/ESZ is: a pair of military units with interlocking fields of fire supporting each other against frontal assaults and flanking movements. Defense relies upon being positioned to assist one another, not on “being inside a fence”. Electronic Access Control Systems (AAA) can work to protect Cyber Assets inside the ESZ from anywhere.

• EAG - Electronic Access Gateway. The fundamental defining characteristics of an EAG are that it hosts the EAP and performs the active function of filtering or forwarding traffic at the demarcation point (boundary protection). Primarily it is firewalls and routers that perform gateway functions at the layer 3 ESP boundary demarcation point. Virtual firewalls and virtual routers inside a hypervisor perform the same function in the same manner. However, hypervisors themselves may not be EAGs if they are not configured with a virtual firewall or virtual router function to provide a gateway function.

Modern security methods typically employ a defense in depth strategy using distributed AAA (Authentication, Authorization, and Accounting) systems to authenticate and authorize access to the Electronic Security Zone based upon characteristics of the user and traffic, while the EAG subscribes to the AAA service for user permissions and filters (permits or denies) traffic based on the source, destination, and port or protocol. Further, Electronic Access Control strategies often employ multiple devices, each containing a part of the AAA solution, such that compromise of one element of AAA does not result in the entire system failing. Vendor and platform diversity within a defense-in-depth systems-based approach to Access Control are generally an element of securing the entire system from vulnerabilities common to specific classes of devices (e.g. all-Windows or all-Linux environments may have common configuration or malware vulnerabilities). Often the Accounting (logging) function is used to determine (and formulate a strategy to correct) any failures in Authentication and Authorization. 

Although some EAG devices are also capable of performing various levels of functions to authenticate and authorize traffic, many are not capable of complete AAA solutions in themselves and therefore differ enough from EACS to warrant different technical control measures. Requirements that acknowledge the difference and allow for handling them differently will prevent any "hall of mirrors" effects as described above only. 

For conceptual discussion purposes EAG acts somewhat like the legacy ESP as a logical demarcation point for conceptual discussions to delineate PCA and BCS from non-CIP-Applicable Cyber Assets. It may even be useful to replace ESP completely using ESZ with EAG for demarcation points.

• CMS - Centralized Management System. System using an elevated privilege account either on behalf of an interactive user or in an automated fashion, allowing mass modification of BES Cyber Systems.

As discussed, these systems are the ones driving the SDT's apparent thought process. The risk posed by these systems is not just unauthorized access to information or BES Cyber Systems, but the ability to modify or destroy the infrastructure the BCS rely upon, or the BCS themselves. These will have unique requirements over and above the others. It would obviously not be beneficial to simply create a reclassification and documentation exercise for entities who would not see sufficient benefit.

Proposed definitions:

AAA: Authentication, Authorization and Accounting systems are Cyber Systems that control ‘Gatekeeper’ functions (electronic access methods and permissions, e.g. authentication of users or control messages that alter dynamic ACLs) for Cyber Systems.

CMS: Centralized Management Systems are Cyber Systems that perform automated management tasks and mass configuration of BCS whether scheduled or on demand using a dedicated service account credential.

EACMS: Deprecated.

EAG: An Electronic Access Gateway is a Cyber Asset that performs active electronic traffic control (filtering and/or forwarding) for ESZ boundary protection based upon the criteria given to it.

EAP: An Electronic Access Point is the logical interface on an EAG where traffic filtering operations take place. 

ESZ: An Electronic Security Zone is the logical container providing separation or isolation from threats or attack vectors to grouped Cyber Assets, said Cyber Assets being characterized by similar operational criticality, or sensitivity to compromised data confidentiality and integrity, as well as needing similar access controls, audit logging and/or monitoring requirements.

SIEM: Security Incident & Event Monitoring is a Cyber System that performs electronic monitoring of Electronic Security Zone(s) or Cyber Systems. 

Rapidly evolving threats & slow-moving regulatory standards

Over at the Anfield Group Blog, Chris Humphreys posts: 

The DOE’ s Quadrennial Energy Review Report states that:

“The current cybersecurity landscape is characterized by rapidly evolving threats and vulnerabilities juxtaposed against the slower-moving prioritization and deployment of defense measures.” I lump regulatory standards and requirements into the “slower-moving prioritization and deployment of defense measures” as one of the key components to preventing a truly proactive stance on cybersecurity. Additional focus on recovery and resiliency needs to be a foundational element of any cybersecurity program because the idea that an organization can combat against 100% of cyber intrusions is false. What becomes critical is the recovery of the system if/when a successful cyberattack occurs."

I couldn't agree more. We will never eliminate all risk.So it behooves us to have a backup plan- resilient recovery strategies. NERC CIP's specific language around redundancy doesn't dismiss the importance of redundancy, but a lot of NERC CIP compliance folks do. The language says one cannot exclude a Cyber Asset from scope of CIP simply because the system is redundant. Fair enough. Redundancy doesn't protect from software vulnerabilities, malware, or mis-configuration. But too many people seem to think that this means redundancy doesn't matter, and in fact, there doesn't appear to be any requirement to have redundancy for Cyber Assets. 

Something that NERC CIP doesn't do well: make clear that assessing technical controls for high availability at the systems level rather than at the device level can provide a more accurate perspective on real cyber security, and this high availability is achieved through redundancy of underlying infrastructure (perhaps switching and virtual network systems, or hypervisor infrastructure) that has little or nothing to do with BES functions, BES Information etc. Building resiliency in and eliminating reliance upon single devices (or as I like to call them, "single points of failure") is a key part of virtualization's benefit.

The entire mindset behind and promoted by the NERC Glossary and the definition of BES Cyber Asset is to blame for this lack. Add that to the prescriptive requirements, device-centric example measures, and the device-oriented Severity Level tables, and you get a self-reinforcing  echo chamber about how to achieve reliability that makes it difficult to look outside the way it has always been done.

A Tale of Two Viewpoints: Mixed Trust vs Shared Infrastructure

In NERC CIP Standards Drafting efforts, industry chatter, and auditing, there has been quite a bit of talk about “mixed trust”; meaning an environment that has both BES Cyber Systems and Cyber Assets not subject to CIP standards. Non-CIP Cyber Assets may be systems that are under the Responsible Entity’s control, and may even be providing functions related to Grid Operation, but not functions which “if rendered unavailable, degraded, or misused” for 15 minutes will affect the reliability of the Bulk Electric System. For example, corporate business systems are not BES Cyber Assets, even though they are under the control of the same Responsible Entity as the BES Cyber Systems.

Here’s the thing; “mixed trust” is not the right term. Mixed trust would imply that Cyber Assets of different trust levels can access each other in an uncontrolled fashion. Nobody is proposing a relaxation of security controls between CIP and non-CIP assets. What is being proposed is “shared infrastructure”. At some level we all have shared infrastructure- the same building, the same power, the same Internet connection. Shared Infrastructure doesn’t mean “mixed trust”. Shared Infrastructure can have logical controls and isolation involved.

A few days back, I wrote about Streetlight Effect in relation to Lew Folkerth's "Lighthouse" article in the March-April issue of Reliability First's newsletter. In it he talks about “Zones of Authority” in relation to the audit process for NERC compliance. At the end of the article he makes an assertion about Virtualization being a bad idea not because of actual security concerns, but because of the auditor’s inability to look at things outside of the designated Electronic Security Perimeter (compliance concerns) required for BES Cyber Assets. While I respect Lew's experience and appreciate the viewpoint, I disagreed with that approach pretty strenuously.

Tom Alrich has another perspective on the article. He makes a point about security being enhanced if the RE’s entire network were in scope for NERC CIP Compliance audits. I don’t think I can agree with that either (although he does acknowledge that the average compliance specialist would rather repeatedly hit themselves in the head with a hammer than take this approach because of the burden of paperwork involved, so it doesn’t appear to be a serious suggestion.) Such an approach would only be a net gain if compliance evidence production didn’t overwhelm the efforts to secure things in the first place. And if all of the compliance requirements were strictly security requirements and not just designed to make audits easier.

But realistically, bringing security mechanisms that exist outside the ESP (Electronic Security Perimeter) into scope of auditing doesn’t require making the entire corporate network subject to inspection by the Regions. Here's the problem with CIP-005 and ESP. Most NERC CIP standards require you to have a program or process to accomplish X. CIP-005, on the other hand, requires everything peripherally related to BES Cyber Systems to reside inside the ESP. What we have here is the "hard crunchy shell" concept from 20 years ago. It provides “bright line” criteria for audit and categorizing assets as either in-scope or out, because the auditor can require a diagram with the asset shown inside a neatly drawn “dotted red line” (inside joke for NERC CIP compliance specialists), but it doesn’t provide good security on its own.

Some of this difficulty is definitional. A PCA "protected cyber asset" is any cyber asset "associated" with a BES Cyber System and can exist "in or on" an ESP, but this is only made specific in the NERC Glossary.  A BES Cyber Asset must be contained in an ESP (CIP-005). An Intermediate System must reside outside the ESP. An EACMS can be in or out.

Documenting a mechanism that provides a security function doesn’t require that everything else on that network be in scope for auditing. Authentication, Authorization, and Accounting (AAA) may be partly provided by an RSA Token system. That AAA system could reside inside or outside of the security zone where clients of the system exist. The documented controls for CIP AAA could bring that RSA server into scope without touching other, unrelated non-CIP servers on its network segment.

The key to providing good security for BES Cyber Assets doesn’t rely solely upon whether a device is inside a particular perimeter. It depends upon the controls applied to accessing that device. A layered defense of the device includes boundary identification & control, but it doesn’t stop at the outer boundary.

Let's address a couple specific points:

"Lew says (or implies) that auditors are having differences of opinion with entities on the security of mixed-trust switches. It seems these entities have switches that implement both ESP and non-ESP VLANs. When the auditors tell them this isn’t secure, the entities point out that the non-ESP VLANs have just as good security as the ESPs do. So why aren’t they safe?"

Let's examine a scenario with two VLANs, one named "CIP" and one named "non-CIP". The main point really isn't that the security of "non-CIP" is as good as "CIP". The most important thing is that network isolation is maintained between them. The configuration of devices in one VLAN are irrelevant to devices in the other VLAN. If it's just a Layer 2 VLAN, there is no "mixed trust" because the traffic from "non-CIP" doesn't mix with that of "CIP". If it's a Layer 3 switch, and routing takes place between the 2 VLANs, then an EAP must be identified (probably the VLAN's logical or virtual interface) and inbound/outbound access controls applied at that point. If it is a Hypervisor situation, then the same principle applies. vSwitch "non-CIP" doesn't have a traffic path that includes vSwitch "CIP" and the Guests cannot communicate with each other. If you create an EAP between them, then you apply the inbound and outbound access controls at that point.

And lest we forget, L2 VLAN is not the only way to have a shared switch infrastructure. Software Defined Networks and Network Overlays exist on a shared hardware infrastructure, but provide network isolation between the security zones as defined in configuration. This isolation, like that of the VLAN example above, is provided as a baseline function of the device, implemented in the control plane by code base. Access to modify configuration and code base is confined to a management plane interface. Traffic transiting the device doesn't have access to this function, only administrators accessing the device in the management plane do.

"Anything else, such as a VLAN that isn’t an ESP, is completely out of his or her purview; the auditor just has to assume these are completely insecure, and thus shouldn’t be found on the same switch as ESP VLANs."

Auditors aren't forced to assume anything.  The configuration of the switch is in scope, because it provides isolation to the "CIP" VLAN. (It doesn't provide ESP under the current model, because it doesn't provide EAP with in-bound/out-bound ACLS. It simply provides isolation, which is one flaw in CIP-005's prescriptive approach. CIP-005 would benefit from being changed to a security objective of isolating more critical assets and traffic from less critical.)The configuration of devices in the "non-CIP" VLAN are out of scope. If something is out of scope, they are not required to render an opinion on it.

"the auditors can’t look at what isn’t in an ESP, which limits the kinds of evidence an entity can show them. "

Currently, auditors can examine EACMS that may exist outside the ESP. They can examine Intermediate Systems which are required to be outside the ESP. I don't think this is a valid assertion. It's an auditing approach preference with deliberate blinders on.

"most entities, if told they could force CIP auditors to consider security controls they have implemented for non-ESP networks, but only in return for having at least some of the cyber assets on those non-ESP networks fall into scope for CIP, would say “Thanks but no thanks. We’ll leave things as they are.”

Perhaps the entity might choose that. But the issue isn't "security controls implemented for non-ESP networks." It is "security controls that exist outside the ESP, that are implemented to protect CIP Assets". ESP is a fundamentally limited construct, and must be considered fatally flawed when used without other other layered defenses.

Cyber Assets vs Cyber Systems Confusion in the Virtual Environment

I'm concerned to still see discussion in the Nerc-isphere about how to categorize a VM in the most simple of clear-cut examples: the virtual server/hypervisor combination. Some folks still disagree that it is necessary to treat each virtual machine and hypervisor as separate Cyber Assets. They think:

"The hypervisor (parent) is the device or software which runs the virtual machine (child). The virtual machine (VM) cannot operate without the hypervisor. This shared relationship means that neither can be separate Cyber Assets. For example, if a VM has been identified as a BES Cyber Asset (BCA); the hypervisor that runs the VM is also a BCA; which also applies to PACS, EACMS, and PCA’s

Treating the VM and hypervisor as separate Cyber Assets can cause mixed-trust virtual environments; the hypervisor runs CIP and corporate VM’s. CIP controls are only being applied to the CIP VM and not the hypervisor; even though the hypervisor “if rendered unavailable, degraded, or misused” can impact the CIP and corporate VM’s."

 Allow me to counter:

The hypervisor (parent) is the device or software which runs the virtual machine (child).

The Hypervisor or Host merely provides a container or environment for the virtual machine. In this aspect the control plane of the Hypervisor operates certain control plane functions on behalf of the guest. However, the Hypervisor is not involved in the control plane¹ decisions made by the guest.  The Hypervisor does not need to (and should not be configured to) interact in the data plane of the Guests. They make computing decisions entirely independent of each other, including their reactions to inputs and malware. If an RE decided to treat Host and Guest as one Cyber Asset for compliance reasons, there is no logically consistent framework to require the long-standing and well-known security best practice of separating the management and data plane of the Hypervisor and Guests. The guest should be completely unaware of and unable to interact with the Host in a secure virtual environment. The Host and Guest more often than not run different operating systems, and it would be difficult to categorize that as one Cyber Asset for any practical purpose. 

The virtual machine (VM) cannot operate without the hypervisor.

This is not strictly correct. More precisely, the VM cannot operate without a hypervisor. It is not dependent upon any specific hypervisor (rather, it can exist on any Hypervisor in the cluster and this is one of its biggest advantages). Therefore treating them as one Cyber Asset is inappropriate because this approach would not require distinct vulnerability assessments, patching, baseline configuration etc.

This shared relationship means that neither can be separate Cyber Assets.

The assumption in this statement doesn’t work in both directions. Presuming that a Guest could not be a separate (meaning independent) Cyber Asset, does not preclude a Hypervisor from being an independent Cyber Asset. After all the Hypervisor is a complete hardware, operating system, (and potentially software) stack in itself. Depending on whether it is a Type I or Type II, it might even be the same operating system as the Guests with Hypervisor function software merely installed on top of a generic operating system. It exists with a hostname, an address, a particular set of open ports/APIs, network connection, and responds to network traffic. The Hypervisor can exist and operate without a single Guest inside.

For example, if a VM has been identified as a BES Cyber Asset (BCA); the hypervisor that runs the VM is also a BCA; which also applies to PACS, EACMS, and PCA’s

This may be true; it doesn’t negate the necessity of treating the Host and Guest as distinct Cyber Assets (for multiple reasons). At most it makes the entire assemblage a BCS.

Treating the VM and hypervisor as separate Cyber Assets can cause mixed-trust virtual environments; the hypervisor runs CIP and corporate VM’s.

This is known as argumentum ad consequentiam (appeal to consequences) and is a known logical fallacy. Whether or not shared infrastructure is allowed and whatever the consequences of doing so, it does not change the distinct character of the Guest vs the Host. It also presumes that sharing infrastructure between CIP-applicable systems and “Corporate” VMs is impossible to secure and must be prohibited. This remains to be proven.

CIP controls are only being applied to the CIP VM and not the hypervisor; even though the hypervisor “if rendered unavailable, degraded, or misused” can impact the CIP and corporate VM’s.

Again, this is an unproven assumption. Treating the Hypervisor and the Guest as separate Cyber Assets does not require a difference in controls. If both are BES Cyber Assets then the same requirements apply to both. Additionally, the technical configuration controls applied to a Hypervisor are generally different than those applied to a Guest (Mal-ware, for example). Again, The Host and Guest may run different operating systems with different open ports, APIs, software vulnerabilities, and baseline capabilities. 
The question is whether a BCA Hypervisor can host a PCA or non-CIP Applicable System safely. The advisability of this approach depends upon whether or not sufficient security controls can be put in place to render negligible the risk of unavailability, degradation, or misuse of the BCA Hypervisor and associated BCA Guests. Risk assessment and acceptance are highly subjective and specific questions that depend more upon the overall architecture and defense in depth posture, than upon a simplistic question of "to virtualize, or not to virtualize."

(1)Control plane functions are not directly accessible by users of the system, they are embedded in the logic of the code base, and are generally require modification to the code base to change them. Virtualizing a server involves abstracting the hardware interactions much like the Hardware Abstraction Layer (HAL) does in Windows. The difference being that the Guest operating system simply sends its hardware access requests to the Hypervisor rather than to firmware. This has a three-fold security benefit. 

1. The users and software accessing the Guest in the Data plane cannot substitute malware in place of authorized device drivers. Since device drivers are one of the worst vectors for malware after Phishng attacks, this narrows the attack surface of the Guest OS.
2. The Hypervisor can be a different operating system than the guest, which means attacks in the data plane against the guest firmware will be ineffectual against the Guest due to no device drivers present and no direct access to hardware, and will be ineffective against the Hypervisor because there is no Data plane access between the two, and the device drivers that actually do interact with hardware are for a different operating system than what is visible to the attacker.
3. Drivers that actually interact with the hardware are generally a smaller subset of better-vetted drivers approved by the Hypervisor vendor, as long as you make the smart choice and go with a Type I Bare Metal Hypervisor.

Electronic Security Perimeter

Over at his blog, Tom Alrich commented to me:

"objectives-based requirements are the only way not to have the situation you discussed (drawing from Lew Folkerth's recent article in RF's Newsletter)- where an entity can be doing great things for cyber security that the auditors can't even consider because they're not in their "zone of authority" (in this case, the ESP). 

Objectives-based requirements would have to be built on top of a framework of concepts that goes beyond just BCS and ESP's, to include all of the entity's computing infrastructure (including "IT" networks)."

I definitely agree that the ESP construct is inherently limiting. For one thing, it is established by arbitrarily drawing a line, identifying where the access point (with controls applied) is, and it only exists at OSI Model Layer 3: routable protocols. Lest we forget, not all security takes place at Layer 3, it's simply the easiest place to measure.

Right now, guidance from NERC and the regions tells us that we cannot use a Layer 2 switch with some ports inside the ESP and some outside. Some would argue you can't use any switch, even a Layer 3 or 4 switch this way. However, there is no requirements language that says this is the rule. It's not a clearly established and absolute security best practice. The reason there can't be any NERC CIP requirements around this is because the ESP is defined strictly at Layer 3, and it was done that way on purpose to exclude Layer 2 connections from needing ACLs (access control lists) and firewalls. Here's the relevant reasoning from CIP-005  Guidelines & Technical Basis: 

"This requirement applies only to communications for which access lists and ‘deny by default’ type requirements can be universally applied, which today are those that employ routable protocols. Direct serial, non-routable connections are not included as there is no perimeter or firewall type security that should be universally mandated across all entities and all serial communication situations. There is no firewall or perimeter capability for an RS232 cable run between two Cyber Assets. Without a clear ‘perimeter type’ security control that can be applied in practically every circumstance, such a requirement would mostly generate technical feasibility exceptions (“TFEs”) rather than increased security."

It's not easy to measure security being applied to layer 2, so we define this type of connectivity out of relevance. Good security would be to apply controls upstream to those devices which can participate, but the device-centric focus of early CIP standards language leads people away from thinking in these terms. Security doesn't have to be "perimeter" based, and generally in a modern layered approach security functions are distributed across multiple devices/systems within your network. They're not all "on the perimeter" so the current NERC CIP standards apply poorly if at all to this strategy. We need some changes to the dominant paradigm. At the same time, we don't need auditors crawling into every ancillary system in our IT department.

Some concepts that add to the discussion:

Accessing a device means the ability to view/modify it's configuration settings. This can be because the device interacts only in the data plane and doesn't have a separate management plane, i.e. a Windows Server. It could be access to a dedicated management port.

Traffic transiting a device means that the packets are forwarded via that device. It doesn't necessarily mean that the user originating the traffic has the ability to view or modify the device in question, in most cases they may not even be aware that the device exists in the traffic stream. Packets crossing a router, switch, or firewall in the data plane would be examples.

Management plane is the logical function which allows a user to interact with the configuration settings of a system. It includes any dedicated management ports, the user interface (terminal emulation of graphical users interface), and the network connectivity to these if accessing remotely (really remote access, which means from anywhere not directly connected, not NERC CIP remote access which only means from outside the ESP).

Control Plane is the logic embedded in the code base of the system. Admins can modify the control plane logic, but usually only by replacing the code base (upgrades, patching, etc.)

Data plane is just packet switching. traffic is generated by users or devices, and sent to a destination IP address. It is switched, routed, forwarded to devices inline along the way. This traffic has no access to the Control Plane or Management Plane on a properly configured network appliance (including hypervisors).

Determining Metrics & Requirements

On Thursday, I wrote a bit about defining BES Cyber Assets, in response to some older discussion on the WICF Forum that is still an undecided topic of discussion today.

It's funny how often the definition of BES Cyber Asset has to be discussed. Tom touches on it again here in talking about the difficulty of an auditor determining whether something is a BCA based upon legal semantics. The BES CA definition is:

"A Cyber Asset that if rendered unavailable, degraded, or misused would, within 15 minutes of its required operation, misoperation, or non-operation, adversely impact one or more Facilities, systems, or equipment, which, if destroyed, degraded, or otherwise rendered unavailable when needed, would affect the reliable operation of the Bulk Electric System. Redundancy of affected Facilities, systems, and equipment shall not be  considered when determining adverse impact. Each BES Cyber Asset is included in one or more BES Cyber Systems."

In NERC CIP Standards, definitions are often contrary to plain language or commonly-understood terminology, and all-too-often include definition by exclusion, even when it tortures reading comprehension. In the definition above, they are intending to close a loophole that one might exploit to get out of compliance scope. And to the point that an identically-configured system might be just as vulnerable to a flaw or cyber attack, it's a valid point. But there are many scenarios where redundancy alone makes the idea of a functional impact into a very remote possibility.

The same article from Tom crosses over to touch upon the Streetlight Effect I wrote about here and here, and is particularly apt with the tagline I ended the second article with: "When your auditor insists upon a simple and clean way to measure compliance at the device level, they may be doing us all a disservice."

It boils down to a common management problem. I can't manage what I can't measure, because to determine the effectiveness of processes I must determine whether they change an outcome and whether that change is positive or negative. So determining what to measure and how it is related to a specific process is necessary, but the problem is that this is not always easy. Some measurements are meaningless. They either don't relate to the process, or the relate to the wrong (or misunderstood) process and therefore indicate results unrelated to the change I'm trying to make. If I don't choose the right test points and interpret the test results correctly, I am not gaining anything other than a thin cover story for why it's not my fault when things go drastically wrong. We have lots of people measuring lots of things and producing tons of documents to prove that they're testing them.

But are we looking at the right things or just the easy ones?

VoIP Phones, BES Cyber Asset or not?

Over at the WICF forum, there's an older discussion on inclusion of VoIP phones as a NERC CIP-applicable Cyber Assets.

Folks keep trying to differentiate between VoIP and POTS (Plain Old Telephone System) analog lines. Here's a newsflash:

A PBX (Private Branch Exchange) system is certainly a programmable electronic device, whether it is VoIP or Analog. Even the majority of Analog PBXs have the ability to be administered remotely via Telnet or similar protocols. Unless the Responsible Entity has analog phone lines that are directly-supplied local loops from the Telco, they do have a programmable device in their voice system. Even then, the Telco switch is programmable, but would fall under a communications system exemption due to the RE not having control over it.

Here's the problem, a VoIP phone does have an OS/firmware that can be updated, including with a hacked copy. It's a simple TFTP operation. It has a configuration that can be modified by the administrator and to a certain extent by the end user. So do smart cell phones (I've used a jail-break to hack my own phone in the past) and even some relatively dumb cell phones.

What it probably doesn't have is the ability to directly affect the BES as long as it is not included inside an ESP with zero internal technical controls (legacy philosophy alert: hard crunchy shell around soft gooey center) that is the minimum acceptable solution at the moment. This could really be a fine example of a situation where short-sighted compliance actually reduces security by forcing you to include VoIP systems inside your trusted perimeter rather than keeping them properly segregated as they should be.

The current comment form on Virtualization is trying to address similar issues. The definition of Cyber Asset is up for discussion. It's also asking specifically whether CIP-005-5 ESP Requirements are adequate to address isolation in a virtualized world, so the parallel may not be obvious, but many of the risks identified for virtualization are unaddressed for nearly-identical risks that apply to physical Cyber Assets.

Bringing in the concept of security zones and granular internal technical controls applied to Applicable Cyber Systems by grouped risk or impact rating has a lot more implications than just for virtual Cyber Assets.

More on Streetlight Effect, and: "The Way We've Always Done It."

So the other day I wrote a bit about looking for things in the wrong place simply because it is easier. It's a pretty common problem in a variety of human endeavours. But it's not just laziness that causes people to look where the light shines. Often there simply is no perfect means of examining a problem. In some cases we have to rely on proxy metrics because we can't directly measure the goal. It might be a quality we're trying to determine rather than quantity, so we choose something that seems to track along with our desired outcome, and measure that.

Another reason we might do something like this is because we don't have a good understanding of the underlying mechanisms that produce a given result. We don't understand, but we desire a particular outcome, so we try to screw the inscrutable by going through the forms that we've seen other people doing, but without understanding why they did. It's as though, controlling something that looks like something else will influence it (sympathetic magic). It's like a cargo cult ritual where the form is the only thing happening; there's no technical substance to the action.

A long time ago in a galaxy not so far away, a paradigm was born; of a hardened perimeter that would keep attackers out of our networks. As security techniques go, it was primitive but had some significant positive effect. Indeed perimeter or boundary defense is still part of a solid security strategy. But it's only a part. Although we've long since passed the point where a simple OSI model layer 3-4 access control list based upon source IP, destination IP, and TCP port is sufficient for security, people are still doing it "The Way We've Always Done It™". Attacks are far more sophisticated now than when the "hard crunchy shell" first surrounded the "soft gooey center" of a network. (Now I want a candy bar!)

We need a layered defense that includes controls internal to our network. We also need security mechanisms that are not based on integrity of the function they are trying to protect. This means that not all security mechanisms can or should be all-in-one. While NERC CIP uses "systems" language at the Requirements level, many of the Measures, the RSAWs and the VRF/VSLs implicitly require a given device to provide the control. This is yet another example of wanting the auditing approach to be simple and easy (looking only where the light is shining).

Unfortunately that sort of simple-to-audit approach precludes, or at least discourages, layered defense mechanisms where components of the security mechanism are provided by network-based functions that integrate multiple types of devices. For example, AAA or Authentication, Authorization, and Accounting functions (which in the NERC CIP world for some bizarre reason is called "Electronic Access Control & Monitoring System or EACMS) generally should not reside entirely upon the single device being accessed, if for no other reason than users requiring access to this device means that attacks on the security mechanisms embedded in it are possible. Far better to have this device act as a AAA client and receive Authentication and Authorization from that network-based function, while sending Accounting information to it. This isolates the code base of the security appliance(s) from the code base of the device that users require access to, and limits opportunity for privilege escalation.

When your auditor insists upon a simple and clean way to measure compliance at the device level, they may be doing us all a disservice.

Virtualization Comment Form- Q&A Session

NERC has comments open right now on Virtualization and CIP Standards. Registered Entities, hie thee hence and make known your will. After all, you get the governance you deserve, not the one you want. Especially if you don't participate.

Here's more or less what we came up with in my neck of the woods:

Q1. Version 5 introduced the BES Cyber System concept, and requirements reference applicability at the BES Cyber System level. However, language in the measures shows that, implicitly, many controls are expected to be implemented at the BES Cyber Asset or device level. The SDT assumes that most auditors expect entities to demonstrate compliance at the device level. Do you agree with the SDT’s assumption?

A1. It has been our experience that guidance provided to auditors leads them to expect and look for controls to be applied to the Cyber Asset. Also, they seem sceptical of implementations where a given device performs a portion of the control function and additional components of the security strategy are implemented across multiple devices on the network. Auditors might consider only the device portion of an overall control and evaluate it outside of the network-based defense-in-depth strategy.

One way to address this inconsistency would be to normalize the use of the term “system” across the example measures rather than “device” wherever applicable. The SDT should add Guidance in the Technical Basis sections to clarify that defense in depth strategies are desirable. The EACMS paradigm should be revised in line with standard IT Security practice and terminology as performing Authentication, Authorization, and Accounting. It is also important to explicitly allow for distributed systems to perform this AAA function for a security zone rather than the legacy concept of hardened perimeter.

There may also be a need to revisit the RSAWs in light of system vs device to provide better guidance to auditors attempting to apply the questions in the RSAW to an entity’s particular security posture.

Q2. The SDT proposes that each virtual machine and hypervisor are separate Cyber Assets. 2. Do you agree with this position?

A2. The proposed definition of Cyber Asset must include virtual machines. Both “virtual machine” and “hypervisor” are  well-understood terms with formal definitions and broad IT Industry acceptance, thus do not need further definition in the NERC Glossary. We agree that each Hypervisor and Virtual Machine is a distinct Cyber Asset. Controls and strategies for securing virtual machines across a variety of industries have been published by agencies such as NIST and SANS.

The key issue the SDT appears to address in this revised definition is clarifying the scope or boundaries of a given virtual cyber asset in order to apply requirements and controls to each. Clarifying the definition is only necessary to address gaps in current requirements language that allow for mis-applying the requirement, not because Industry doesn't understand how to do it.

Q3. Do you agree that the proposed Cyber Asset definition clarifies the term programmable? Please provide a rationale to support your position.

A3: The SDT’s proposed definition clarifies programmable means that such a device is subject to configuration and software changes by the end user. This clarification and scope limitation is useful.

However, the device does not "consist of" all the data stored in the device. Data inside the device is peripheral and irrelevant to the operation of the device. It may be necessary to operation of the BES, but unless you want the BES to be one, singular BCS, it is ridiculous to scope it that way.

Much data inside the device is peripheral and irrelevant to the operation of the device. Even data which is required to perform the function of the device is generally not part of the device. This “data in the device” portion of the definition apparently is a mis-interpretation of wording from Section 215 of the Energy Policy Act of 2005 which actually reads: “programmable electronic devices and communication networks including hardware, software and data that are essential to the reliable operation of the bulk power system.” Even this is problematic in that standard IT Security best practice separates the concept of protecting a system (better known as Information Assurance, Source: NIST SP 800-50, CNSSI-4009) from the mechanisms of protecting data transiting or resident on that system (the latter being Information Security, Source: NIST SP 800-59; SP 800-53; SP800-53A; SP 800-60; CNSSI-4009; FIPS 199; 44 U.S.C., Sec. 3542). For example; a SCADA server can be deemed perfectly functional with no real SCADA data present. A newly-deployed SCADA server functions at the instant of being commissioned before real SCADA data is input.

The introduction of the concepts of management plane and data plane which are referenced in question 8 is a useful addition to the NERC CIP discussion because it enables appropriate controls to specifically protect systems or data.

Q4. Such (shared infrastructure) configurations are not addressed explicitly in CIP-005-5. Are modifications required to address the issue? 

A4. CIP-005-5 would benefit from being modified to a security objective-oriented standard rather than a requirements-based standard. The security objective in this case would be isolation of CIP-applicable Cyber Assets from Cyber Assets that are out of scope of CIP controls. The mechanisms of that protection are primarily Boundary Protection and Control of Network Ports, Protocols, and Services (SANS 20 Critical Security Controls).

CIP v5 narrowly focuses on routable protocols and Layer 3 controls and does not address the other layers of the OSI model. For example, under CIP-005-5 Layer 2 protocols are not addressed and can convey malware as well as allow information exfiltration and cyber-attacks even if no routable IP communications are present. Controls for these protocols should not be limited to or defined by Layer 3/4 ACLs on a firewall or router as the only or even the best means of achieving in-bound and out-bound access control. Entities need the opportunity to provide technical controls at whatever conceptual layer is appropriate to meet the security objective.

I recommend retiring the ESP construct in favor of security zones. When framed in terms of Boundary Protection, a security zone is more inclusive and granular because it is not limited to routable protocols at the OSI Model’s Layer 3. A security zone construct does not force any particular interpretation or control onto serial or other non-routable means of transporting data or accessing the management plane of any systems. Security zones apply to physical and logical separation equally. An example of logical isolation provided by other than ACLs would be when a hypervisor provides isolation between Guest VMs or between Virtualized Network Functions. This isolation is implemented in the control plane by means of logic embedded in code base (NIST SP 800-125A – Draft, Section 1.2: Hypervisor Baseline Functions) and not at a conceptual network layer.

Current CIP standards take a broad stroke approach by requiring the protection of all cyber assets within an ESP in a singular manner at the highest impact level of controls ("High Water Mark"). Further, High Water Mark is applied to the impact rating of the BES Asset or Facility, and not to the impact rating or risk level of the Cyber System itself. This is not cost effective or flexible enough for individual entities’ needs. Security zones provide a scalable means of appropriately protecting Cyber Systems of differing security risks by further isolation within the zone.

Q5. The SDT asserts that VLANs providing logical isolation are not addressed explicitly in CIP-005-5, and controls may be necessary to isolate BES Cyber Systems. Are the current requirements of CIP-005-5 sufficient to address logical isolation using VLANs? Please provide your rationale.

A5. The same modifications necessary to make CIP-005-5 adequate to question 4 would apply to addressing the specific question of 802.1 Q VLANs providing adequate logical isolation in question 5. The security objective should be to provide isolation by means of Boundary Protection as well as Control of Network Ports, Protocols, and Services. Legacy software vulnerabilities that have long since been patched or known exploits that are mitigated through proper configuration should not require the blanket rejection of VLANs as a component of a particular entity’s specific security scheme. Newly discovered exploits and vulnerabilities are addressed through CIP-007 testing and patching, CIP-010 baseline configuration, change management, vulnerability scanning and assessment.

Q6. Do you agree with the proposed definition of CMS (Centralized Management Systems)? If not, please provide alternative language for the definition and your rationale.


A6. The SDT’s proposed definition of CMS captures most types of systems that support automation with a large span of control and privileged access. A similar span of control risk exists in that EACMS is a type of CMS for electronic access but the term EACMS is specific to NERC CIP and used nowhere else in IT Security or Information Assurance in any other industry. This inherently limits the amount of expertise, guidance, and documentation available for solving the root problem of controlling access to CIP-applicable systems.

The SDT should retire the NERC CIP defined term Electronic Access Control & Monitoring System from the NERC Glossary and adopt the industry solution Authentication, Authorization, and Accounting System (AAA System). Non-standard jargon should be avoided when adequate terms and concepts exist already.

Further, the SDT should clarify in Guidelines and Technical Basis, that:

• AAA clients that subscribe to AAA services (e.g. via a protocol such as LDAP, RADIUS or TACACS+) but do not maintain any account information are not AAA Systems in themselves
• Remote access clients or terminal emulators that are used to connect to a CMS, are not a CMS in themselves

Q7. Do you agree with the SDT’s approach to reference the CMS specifically as a type of applicable system in the CIP standards? Please provide your rationale.

A7. The inherent risk in such (centralized management) systems’ span of control and privileged access to CIP-applicable Cyber Systems should be addressed in support of the security objective of protecting BES Cyber Systems from threats in the data plane and isolation of the management plane (out of band management).

Q8. Do you agree with the SDT’s approach to require the isolation between the data plane and the management plane?

A8. Cautiously agree with the SDT’s proposal to require isolation between Data Plane and Management Plane for centralized management systems when system capability allows and risk justifies it (i.e. Hgih and Medium Risk Control Centers). We caution the SDT against overly rigid prescriptions for providing isolation. Combinations of other controls may afford the same or better protection in a particular circumstance. When the use of automated tools can improve security and manageability, it is important to avoid discouraging automation with overly burdensome compliance requirements.

Q9. Do you agree with limiting the applicability to high and medium impact Control Centers?

A9. Limiting applicability only to those facilities such as High and Medium Control Centers with the highest level of risk is reasonable, and there may be exceptions to those as well. Combinations of other controls may afford the same or better protection in a particular circumstance.

Streetlight Effect Apparent in NERC CIP Requirements

This comment is in response to an article in the ReliabilityFirst Newsletter - "Virtual Systems and Zones of Authority".

I'm going to quote a bit liberally, since the original on page 7 is difficult to link to directly. But first, read this:

https://en.wikipedia.org/wiki/Streetlight_effect

"Responsible Entity personnel see their entire network as a whole, with the parts of that network subject to CIP compliance as part of a larger security picture. They can see the protections afforded to all systems, and can see how the protections applied to non-CIP assets increase the security of CIP assets as well.

CEA personnel, on the other hand, only have CIP assets within their purview. They cannot consider non-CIP assets as adding to the entity's security posture, as those assets are not under the CEA's regulatory authority. Non-CIP assets are not subject to audit by the CEA and may change at any time with no notification to the CEA.

This difference in viewpoint can lead to conflicting views of virtual systems such as virtual networks. Responsible Entity personnel see the protections applied to the non-CIP networks that might share, for example, a physical switch with CIP networks. They see the multiple layers of protection and the controls surrounding the security of these non-CIP networks.

CEA personnel, on the other hand, do not have the authority to review the security level of non-CIP assets or networks. The CEA personnel must therefore assume that any non-CIP assets or networks could be compromised and used in attacks on the in-scope CIP assets and networks. The resulting differences in the perception of risk can be a source of misunderstanding between Responsible Entity personnel and CEA personnel.

I call this difference in perspective “mixed zones of authority.” The Responsible Entity's zone of authority is all of its owned assets, both CIP and non-CIP. The CEA's zone of authority is limited to assets that are in scope for CIP.

For this reason, and others that I don't have the space to go into here, I strongly recommend that

Responsible Entities refrain from implementing Cyber Assets or networks that mix CIP in-scope and out-of-scope assets, network traffic, or data. The reason for not mixing in-scope and out-of-scope is not, as is commonly discussed, that “untrusted” configurations are implemented.

The biggest issue, in my view, is that without being able to view all aspects of the systems used for BES reliability, there is no way for the CEA to ensure that weak or high-risk configurations are not implemented."

{emphasis added in the last two paragraphs.}

That’s an interesting perspective. Once again, the crux of the argument is not whether or not virtualization adds to or subtracts from security. It’s the difficulty of auditing that drives the recommendation.

This, my friends, is a terrible basis for driving standards. It is what is known as a perverse incentive. It drives one to make decisions that, on the basis of achieving the objective, one would not otherwise make. It's a common problem in security, that one does what is visible, and easy to measure rather than closing the worst (invisible) gap. In fact, this is a problem far beyond security, it happens in all kinds of production environments with easily gathered statistics, in management, etc.

Standards should not be about making the network easy to audit, it should be about making it difficult for a bad actor to compromise reliability and security. Industry knows how (or individual entities can quickly learn how) to secure a virtual environment in shared infrastructure mode. Industry properly securing the virtual environment isn’t the sticking point. Even the Federal Government's more security-conscious entities have processes for contracting compliance in Cloud Computing. Surely an entity is capable of adequate control in its own networks.

Take a look in the PCI standard and its glossary. The concept of trusted and untrusted network is defined there as:

Trusted Network	Network of an organization that is within the organization’s ability to control or manage.
Untrusted Network	Network that is external to the networks belonging to an organization and which is out of the organization’s ability to control or manage.

There is no logical basis to assume that an entity capable of providing a compliant solution inside their ESP is simultaneously unable to or unwilling to provide security to Cyber Assets under their own control outside the ESP, simply because the auditing agency doesn’t have explicit control over this latter subset of Cyber assets. Requirements-based language may be the problem, where an objective-based standard wouldn’t have quite the same problem. For example, proof that you meet the objective of “isolation” would tend to consist of controls and measurements applied inside, outside, and on the perimeter. It doesn’t require the bright line of layer 3 “ESP” as the sole measure of compliance and it works well with security zones as a concept.

As long as Critical Infrastructure Protection is driven by compliance rather than security, the Grid will be at unnecessarily elevated risk.