In a recent post,
I pointed out that CIP-013 R1.1 is close to un-auditable. This is because it
requires the entity to develop (and implement) a supply chain cyber security
risk management plan, but it gives very little information about what should be
in the plan. Specifically, it should include a list of risks that should be
addressed in the plan, but it doesn’t do that – meaning the entity is on their
own to draw up that list[i].
However, R1.1 does provide some
high-level information on what the plan should cover; I discussed that in the
post.
Now I want
to turn to R1.2. What does this requirement tell us? As you know if you’ve read
CIP-013, R1.2 provides a list of six “things” that the entity needs to do. What
are these things, and how do they relate to the supply chain cyber security
risk management plan that’s mandated in R1.1?
Let’s look
at the first thing, numbered R1.2.1: “Notification by the vendor of
vendor-identified incidents related to the products or services provided to the
Responsible Entity that pose cyber security risk to the Responsible Entity”.
Let’s pull this apart. It’s essentially saying that a vendor needs to notify
you (i.e. the NERC entity responsible for compliance with CIP-013) when they
identify a security incident related to products or services you buy from them.
How does
this relate to the R1.1 plan? R1.1 says the Responsible Entity needs to
identify supply chain cyber security risks. It doesn’t also say you also have
to mitigate those risks, but that is clearly implied. What would be the purpose
of CIP-013 if you were just required to identify a bunch of big, hairy risks
you face, then you could simply declare your job done? It would obviously make
no sense to do that, and I predict that anyone who omits risk mitigation in
their actual CIP-013 R1.1 plan will not fare well at the hands of their
auditor.[ii]
So the six
things in R1.2 must be either risks or mitigations. They clearly aren’t risks,
so they must be mitigations. In other words, each one lists a high-level
mitigation for a particular supply chain cyber security risk. The entity is
required to implement all six of these mitigations. While the risks that are
being mitigated aren’t explicitly stated, it isn’t too hard to state them:
- R1.2.1 (quoted above) describes mitigation of the risk that a security breach at a vendor won’t be reported in a timely fashion to users of the vendor’s products, leading to cyber security incidents involving the vendor’s systems installed on user networks.
- R1.2.2 reads “Coordination of responses to vendor-identified incidents related to the products or services provided to the Responsible Entity that pose cyber security risk to the Responsible Entity”. This is mitigation of the risk that a vendor whose products are found to have a serious vulnerability won’t provide adequate assistance to their users in mitigating that vulnerability, leading again to security incidents at users, caused by the vendor’s products.
- R1.2.3 reads “Notification by vendors when remote or onsite access should no longer be granted to vendor representatives”. This is a mitigation of the risk that a vendor will terminate an employee who had electronic and/or physical access to users’ systems, and that ex-employee will take out their resentment for this action on those systems.
- The mitigation described by R1.2.4 is “Disclosure by vendors of known vulnerabilities related to the products or services provided to the Responsible Entity”. This mitigates the risk that users will experience cyber security incidents that could have been avoided had they been notified of the vulnerabilities.
- R1.2.5 reads “Verification of software integrity and authenticity of all software and patches provided by the vendor for use in the BES Cyber System”. Of course, this is really two mitigations, addressing two risks. Verification of software integrity mitigates the risk that a malicious actor will tamper with a vendor’s software or firmware during the process of delivery (electronic or on physical media) to the user. Verification of software authenticity mitigates the risk that a malicious actor will set up a fake web site or otherwise impersonate the vendor, so that the user obtains software or firmware that is damaging to their systems.
- Finally, R1.2.6 describes this mitigation: “Coordination of controls for (i) vendor-initiated Interactive Remote Access, and (ii) system-to-system remote access with a vendor(s)”. The risk here is the one behind the recent (but ongoing) Russian attempts to compromise grid control systems by first compromising vendors and then hijacking their ability to access their products remotely for maintenance and troubleshooting purposes.
It would be
very hard to argue that the six risks are all that need to be included in the
plan, since R1.1 calls on the entity to “identify and assess cyber security
risk(s) to the Bulk Electric System from vendor products or services resulting
from: (i) procuring and installing vendor equipment and software; and (ii)
transitions from one vendor(s) to another vendor(s)”. It certainly doesn’t just
say “Develop a plan to implement the six risk mitigations in R1.2.”
Yet if you
look at almost any of the guidance on, and discussion of, CIP-013 that has come
out from NERC
or the Regions as well as other organizations like NATF,
you’ll see close to no discussion of anything but these six things[iii].
I haven’t seen any discussion from NERC or anyone else (other than myself)
about what should go into a CIP-013 R1.1 supply chain cyber security risk
management plan. Is this because the six
risks (and their mitigations) that I’ve just enumerated are far more important
than all other supply chain cyber security risks? Is that why the drafting team
specifically called them out in R1.2?
I certainly
won’t deny that these are important risks. However, they are called out in R1.2
because FERC specifically required that these six mitigations be included in
the new standard, when they ordered
that it be developed in July 2016[iv].
But there
are certainly lots of other supply chain cyber security risks that should be
addressed as well. Last week, I was fortunate to be able to attend the Software
and Supply Chain Assurance Fall Forum, held at the offices of the MITRE Corporation
in McLean, Virginia. This is a quarterly event that has been going on for a
couple of years, although I just recently heard about it. It is a gathering of
cyber security professionals working at various U.S. Departments (including
DoD) and the “Three-letter Agencies”, who are very concerned about supply chain
security. I will write a post on some things I learned at this meeting in the
near future.
The group
now includes attendees from private industry, and this meeting’s theme was the
energy industry. There was some discussion of CIP-013, and Howard Gugel of NERC
gave a very good presentation on the standard and all the activities going on
around it. However, Howard unsurprisingly didn’t dwell on the idea of the R1.1
plan (frankly, I’m not sure he even mentioned it) and spent more time
discussing the six mitigations in R1.2, and especially the question of the role
of contract language in obtaining vendor participation for addressing those six
mitigations (since they are all vendor risks).[v]
In the
Q&A period, two audience members brought up risks they thought were very
important, that they were surprised weren’t mentioned by Howard. One questioner
brought up[vi] the
risk of malicious code and back doors being embedded in chips purchased by a
vendor in Asian markets because they are low cost – which has been discussed a
lot in the past few years.
The other
was the risk of unpatched vulnerabilities in software that was developed by the
NERC entity itself, or was custom-developed for it. Both CIP-007 R2 and CIP-013
R1.2.5 only deal with patches released by vendors for their software. The
gentleman who asked the question (who was the main organizer of the conference)
was incredulous that this risk was nowhere addressed in CIP. I pointed out that
this was the whole idea of the R1.1 plan – that the entity needed to consider
all of the important risks in their plan, not just the six from R1.2. Of
course, there are lots of other supply chain risks we could discuss as well –
some of which you’ll find in the NERC/EPRI Supply
Chain Risk Assessment report from September.
And folks,
here’s the point of this post: Supply chain is by far the biggest cyber threat
area for most industries now, and without doubt it’s the most serious for the
power industry – just look at the Russian attacks. If the industry treats
CIP-013 as just being about the six risks in R1.2 and nothing else, CIP-013 won’t
mark a watershed in grid security (as well as supply chain security in general)
– which it could be if implemented properly. Even more importantly, you will be
leaving your OT network completely open to a lot of serious supply chain security
risks, just because they didn’t happen to be one of the six listed in R1.2. I
certainly hope NERC will wake up and realize they have to make an effort to
help the entities understand the need for developing a comprehensive supply
chain cyber security risk management plan. And maybe FERC can nudge them a
little in that direction when they approve CIP-013.
Before I go,
I want to bring up one concern that a lot of people may have – especially when
they hear me bringing up the need to “comprehensively” treat all supply chain
security risks. Does this mean that, instead of just worrying about six risks,
you’ll have to worry about say five times that number? The six risks are
already looking like a good amount of work. Do you really have the bandwidth
and the resources to address all
important supply chain security risks in your plan?
I want to
remind you that this isn’t a supply chain cyber security plan; it’s a supply
chain cyber security risk management
plan. The essence of risk management is that you consider the degree of risk
posed by each threat you face, and then allocate your resources toward
mitigating the most serious threats. For threats that pose less risk, you
allocate fewer resources or perhaps none at all.
Whatever the amount of dollars or hours you have available to spend on risk mitigation, you will always mitigate the most risk – i.e. get the most bang for your buck – if you allocate those toward the greatest risks. With prescriptive requirements – like most of the “legacy” CIP requirements, and like CIP-013 R1.2 – you don’t get any choice on how you allocate your resources. You have to spend whatever it takes to comply with the prescriptive requirements, and if you have a few nickels left over you might try to mitigate a couple other risks to a small degree – even though you may think these other risks are much greater than the ones addressed by the prescriptive requirements. When it is up to you to identify threats, rank them, and mitigate them according to the risk each one poses, your entity will be more secure – and the BES will be, too. What’s to lose by doing this?
Any opinions expressed in this blog post are strictly mine
and are not necessarily shared by any of the clients of Tom Alrich LLC.
If you would like to comment on what you have read here, I
would love to hear from you. Please email me at tom@tomalrich.com. Please keep in mind that
if you’re a NERC entity, Tom Alrich LLC can help you with NERC CIP issues or
challenges like what is discussed in this post – especially on compliance with
CIP-013. And if you’re a security vendor to the power industry, TALLC can help
you by developing marketing materials, delivering webinars, etc. To discuss any
of this, you can email me at the same address.
[i]
To speak more accurately, the problem isn’t exactly that R1.1 doesn’t provide
information about what should be in the plan. The real problem is that NERC’s
very prescriptive enforcement format (based on auditing, of course), simply isn’t
suited for plan-based requirements like CIP-013 R1. If there were a
collaborative format in which the Regions could freely work with an entity to
make sure they developed a good plan, then worked with them to make sure the
plan was implemented well, the fact that R1.1 doesn’t provide much guidance on
what should be in the plan wouldn’t matter so much.
But because NERC only knows how to do strict
auditing-based enforcement, it simply isn’t possible for them to officially
work with the entity to understand what they need to do to develop a good plan –
because of auditor independence concerns, of course. To some degree, all of the
Regions actually do work with their
member entities as they come into compliance – and some more than others. But
it is always a kind of don’t ask/don’t tell sort of enterprise. I hope someday
NERC can have a different enforcement model for the CIP standards, since
cybersecurity is completely different from electrical engineering – which is
ultimately what the 693 standards are based on. Prescriptive requirements and
audits just don’t work for cyber.
[ii]
However, this is a pretty amazing omission. It would be like an ordinance for
obeying stop signs spending a lot of time describing the different types of
stop signs, but never actually requiring that you stop for one! As I said, I
don’t think there’s any possibility that this omission will invalidate
CIP-013-1, but I certainly hope FERC adds this to their to-do list for NERC for
CIP-013-2, as they develop their Order approving CIP-013-1. I can think of a
few other things I would like to see on that to-do list as well, so if one of
the Commissioners wants to call me or invite me over for beers (I would come,
even though I live in Chicago! – although I wish they’d invited me when I was
in DC and McLean, VA last week), I’ll be glad to describe my ideas to them.
These have all been stated in my
posts at various times, but who has time to read through all of those things? I
certainly don’t.
[iii]
The one exception to this rule that I know of is the recent NERC Supply
Chain Risk Assessment report, which was prepared by EPRI. It is a good
document, and does list some different supply chain risks that have nothing to
do with the six things. However, it also doesn’t even pretend to provide
guidance for CIP-013 compliance.
[iv]
The six mitigations appear at different points in the order, but they all were
directly ordered by FERC. The drafting team had the good idea to gather them
all in one place in the new standard, rather than have six separate
prescriptive requirements for them.
[vi]
I admit this person brought up his question during a different discussion of
CIP-013 on the first day of the conference (Howard spoke on the second day). My
response discussed below was also given on the first day.
No comments:
Post a Comment