Introduction to Microsegmentation

Introduction to Microsegmentation

This blog begins an introductory series of moderately long blogs, covering key aspects of Microsegmentation and Zero Trust.

The key point to Microsegmentation and Zero Trust is that security is tied to device type, identity, trust status, and other attributes, as opposed to IP address or MAC address.


The series will use graphics from Elisity because:

  • Their product is totally focused on micro-segmentation — fewer distractions!
  • Elisity’s early sales and company growth depended on getting in at the customer and getting the job done. FWIW, one of their target markets is hospital and other environments with rapid growth of OT/IOT devices, and in general, companies needing rapid success at micro-segmentation. (“Time To Value”).
  • They’ve allowed re-use of their graphics, including screen captures of the Elisity administrator training course.
  • Elisity currently supports networks with selected Arista, Cisco, HPE Aruba, or Juniper switches, and selected Palo Alto firewalls.
  • Their product contains some interesting features and policy scaling innovations, ones that AFAIK Cisco (and perhaps others) lack. Perhaps due to customer experience and rapid feature development?
  • Elisity also has some good advice and supporting technology to manage micro-segmentation changes with minimal network disruption.


Cisco ISE and DNAC/CCC can do many similar things and a lot more. But:

  • Their extra functionality makes telling the story more complex.
  • I don’t currently have access to a copy of the Cisco products for screen captures.


Other vendors: I lack data, have no experience of their products. Feel free to add value and share your experience / knowledge by commenting on the blogs! What do other vendors products do especially well? What are they good or bad at?


I’ve included the following Elisity graphic to represent the goal of micro-segmentation. In this case: securing OT manufacturing line controls, user and IT devices, and IOT devices, across multiple switches in one or several sites or buildings.

Elisity image suggesting segmentation of OT and converged IT/OT traffic. The right side shows this as three different micro-segments, as lines with attached devices connected to a VEN switch.


Micro-Segmentation Topics

Here is a list of some of the things one needs to address when implementing micro-segmentation. I intend to cover these in a few blogs.

  • What Is Micro-Segmentation? Why do it?
  • Picking, Licensing, and Installing a Micro-Segmentation Product
  • Network Device Discovery. What’s attached to the network? Do I trust it?
  • Network Traffic Flows Discovery. What traffic exists, what traffic should be allowed, blocked?
  • How Scalable Modern Switch Security Works (Security Group Tags!)
  • Building and Testing Security Policies
  • Managing and Scaling Security Policies Across Many Sites
  • Managing Micro-Segmentation Updates


This first blog in the series covers the first two items above.


The blog series may combine some of these topics, and also will likely cover some of these in more detail than others. Experience also says that the topics will likely shift as writing the series progresses. Hey, my crystal ball is cloudy, and things evolve as the blog sequence gets written.

Anyway, let’s get started …


About Micro-Segmentation

Remember firewalls with un-manageable 70,000 or 100,000 entry access lists with seemingly random address ranges and ports? Long disorganized access lists (“ACL’s”) which nobody understood? I still recall the pain.

Those access lists had some real challenges, including:

  • Lack of organization (at most sites). It could be hard to tell if a given port was permitted or denied between two addresses. Deciphering the impact of the ACL could take a lot of time and effort, and be error-prone.
  • Not at all self-documenting.
  • Embedded address wildcards (talk about obscure!).
  • Embedded ports and ranges (with no mention of the app they support).

That made maintenance slow and error-prone. And easy-to-make errors then would cause inappropriate traffic blocking and outages, requiring scrambling to fix them.

Just say no to bloated, unreadable access lists!


The firewalls were also a problem. Stateful enforcement: good. But backhauling traffic to secure it also aggregated traffic, complicated the network topology and flows, and required buying one or several massive and very costly firewalls. Now with Cisco announcing DSP’s in switches, you can (for a price) distribute that firewalling workload. (“Every switch a firewall”?).


Another aspect prior to Microsegmentation was VLANs. You typically ended up mapping devices to VLANs, to separate traffic from groups of devices up to the router and/or firewall. That made the network more complex to deploy and manage. Having logical tags to separate traffic rather than on-the-wire tagging makes it a lot easier to deploy the network and add security incrementally.


Thank goodness all that has changed, as sites have moved on to better tools!

Doing Micro-Segmentation and switches that can enforce ACL’s based on Security Group Tags has moved the goalposts!

I’m going to assume if you’re reading this, you have some idea what micro-segmentation IS, and why you’re doing it. Having said that...


So What IS Microsegmentation?

For what it’s worth, my latest definition is that micro-segmentation is about grouping devices and device types and providing finely-grained control over flows between those groupings.

If you think of a segment as a piece of network with devices attached, micro-segmentation is about logically taking that piece of network and logically sub-dividing it into micro-segments. The old way of doing this may have used VLANs, or security group tags embedded in headers (Cisco). More recently, doing identity (MAC or current IP) to logical group tag is what I suspect vendors are doing.


Micro-segmentation shifts the focus from IP addresses to device types and other information. By tying in Identity-based tools, micro-segmentation can be based on user or device identity, and even on a dynamic behavioral trust score for that identity.

That (in principle) enables you to segment without creating VLANs and related configuration for different device types. I.e. just have all devices on one VLAN and the switches enforce micro-segmentation rules as to which device groupings can talk to each other.

Thus microsegmentation provides security enforcement on traffic going between the different micro-segments. It may or may not be stateful enforcement, but how critical is that, really? (Philosophical debate lurking there, I’m going to duck it.)

Most vendors have put out good sales documents about what Microsegmentation is, and its advantages. I’ve included some such links below for your further reading, in the first of two Links sections.


What Does Microsegmentation Look Like?

Elisity starts by classifying devices into approximately 15 dynamic global Policy Groups. That’s enough to start putting a fairly robust initial Microsegmentation policy in place.

The following Elisity graphic suggests what those representative groups might be:

Elisity graphic showing device types such as Raspberry, IP Camera, Printer, HVAC, Media Player. A call-out shows device information such as IP, hostname, MAC address and more.

IdentityGraph is Elisity’s term for what the controller knows about a device. That can include lots of information imported dynamically from other security products, especially Identity and Trust related products.

The following image shows a deny-all policy for traffic between two device groups.

Elisity image showing the source/destination matrix at the left. A callout from one of the grid boxes in the matrix is labelled Medical Imaging Policy, and provides some high-level details.

The graphic suggests what I call “The Matrix”, common to both Elisity and Cisco SDA/DNAC/CCC. The matrix shows source vs. destination security group, and each box in the matrix grid links to the security policy for traffic from that source to that destination. In the above graphic, the policy controls traffic from CT Scanners to Streaming Media Players. There is of course a related matrix cell (box) with policy for the reverse flow.

Part of micro-segmentation ideally includes replacing ACL’s with sanity and readability, organizing security policies in structured form around named groups of devices and users, address blocks, ports and port groups, rather than raw IP addresses or address blocks, TCP/UDP ports, etc. That is, achieving a degree of self-documentation!

That definition may be a bit optimistic, as far as named ports/port groups. After all, which ports each app uses may change over time, is likely poorly documented, and probably overlaps with the port ranges many other apps use.


For comparison with my list of topics above, Elisity has a good document summarizing how to implement: Implementing Microsegmentation: A Practical Guide. Here is their list of steps:

  • Assessing your network and identifying assets
  • Defining and refining segmentation policies and rules
  • Implementing the enforcement technologies (PJW note: ideally, the technology helps with the above tasks too!)
  • Enforcing and monitoring Microsegmentation policies


That list of tasks is aligned with working with their product. Whereas I’m trying to cover some of the broader project scope. Hence my earlier list of topics is a bit broader.


The Elisity blog also notes some Best Practices:

  • Establishing clear and granular segmentation policies
  • Regularly reviewing and updating policies as needed
  • Ensuring proper communication between stakeholders
  • Balancing degree of security with user experience and performance

There are some key points there that sites often have to learn by experience.

Segmentation and especially Microsegmentation are not going to be one-and-done. So you need to think ongoing updates, manageability, and not going off the deep end. One sign you’ve over-done it is ongoing outages due to complexity and human error, difficulty updating the policy to accomodate new needs, etc.


Getting Started Doing Micro-Segmentation

So if you’re getting started, you presumably have a mandate or approval from management, funding, etc.

It is good to pause for a moment, ask yourself, and document: Why are you doing Micro-Segmentation?

Hopefully not because you’re a masochist!

Key questions:

  • What’s the driving management directive/support/funding, or it is network/security-originated on the tech side? Perhaps even replacing incomprehensible long ACL’s with an approach that can actually be maintained and improved over time?
  • Perhaps sold to management as “we keep causing outages because the ACL’s are so gnarly, it’s time to? do it a better way”.
  • What are the deadlines? What is the “minimal viable product”, i.e. what MUST absolutely get done.
  • What’s the funding? For product, but also staffing budget? Does that include training and long-term full- or part-time commitment for two or more engineers?

If you are doing micro-segmentation, you will need to know the goal, the definition of success, and how soon results are needed. You will have talked to peers in networking, security, operations, or across other organization boundaries, to make sure they’re aware of your efforts, approve, won’t react angrily over control, fear, or surprise reaction issues, etc.

Put differently, who is the team that will be working to gather the necessary information, work out the security rules and policies, and prevent as well as deal with any adverse impacts. Usually the working group with include some combination of network and security staff, depending staffing availability and interests, team structures, and management choices.

You’ll also need to know priorities. Are there any glaring security problems?


Plan An Incremental Approach

It’s important to recognize that you cannot achieve full control all at once. Also that Micro-Segmentation is not just technical: it requires coordinating deployment and policy planning with staff across the organization. And taking judicious incremental steps to minimize breaking things, and so as to have good reversibility of changes should there be surprise problems.

Does management understand you will be taking an incremental approach, in order to minimize outages? I.e. get started at small scale, get experience with the tools and new approach, provide some level of critical segmentation, then broaden the network scope? Then perhaps start a second pass, increasing the policy granularity? If management doesn’t understand this, ask them how they feel about outages caused by going too fast?

Startups talk about Minimum Viable Product. What’s yours? What’s the simplest micro-segmentation you can deliver that provides solid value, and a basis for more finely-grained security controls in follow-on phaes?

You will have to discover devices, come up with a draft policy, validate it, deploy it, fix problems and broaden scope. Sometimes all at the same time, e.g. if the company failed a security audit and management is in a hurry.


This is a big task, for at least three reasons:

  • Finding all the weird and wonderful things connected to the network. Expect surprises!
  • It takes time to identify devices and critical app flows. Their owners often are non-technical and may not know tech details. Assuming you can even identify who the owner is.
  • Determining what the policy needs to be takes time (talking to security, mgmt, legal,? …, and getting consensus). If you’re lucky, most non-technical groups will have broad goals but let you define the specifics.
  • Your policy will likely evolve over time (and get more complex as nuances appear).
  • It takes time to schedule change windows, monitoring for problems, etc.


The only way to succeed is to:

  • Recognize this won’t be “one-and-done”. Make sure sure management and tech staff understand that. A permanent “tech owner” or two will be needed in various teams.
  • Given that, don’t bite off too much! Start causing outages and the Microsegmentation project will lose support and fail. (E.g. getting one change window per year likely equates to failure.)
  • Everyone needs to understand that there is a trade-off between time and accuracy. That is, you can only monitor traffic or a pilot policy for some time period before activating enforcement. But in doing so, the policy may not be perfect and may need adjusting as later issues occur, e.g. traffic flows that only occur at year-end.
  • You’ll need to think in terms of: doing something, scale it up, refine it, increase key granularity, iterate.
  • Make sure you have the staffing that will be needed. Tools are great, but they do need skilled people driving them.
  • And if you’re doing the micro-segmentation with new tool(s), allow time to thoroughly learn to work with the tools, preferably on a small scope or lab where mistakes will have limited adverse impact.
  • Building trust is important. If you keep causing outages, it will become hard to progress. Thorough prep and communication, minimal outages, and outages fixed quickly are key components to success!


If the project is super-rushed, you can perhaps do “widening circles”. Do a less-critical site, learn, make the inevitable learning mistake or two, refine, get it to a good point. Then task a large staff group to do more sites, perhaps in order of less-critical to more-sensitive. Meanwhile the original team can be deepening the policy, providing finer granularity where needed. But always think: no more complexity than absolutely needed, or your job will become very painful!


So does one start with:

  • Requirements (note many online docs focus on the various security standards, but you can spend a lot of time just reading them, let alone trying to prioritize)
  • What criteria are available (vendor ID, authentication, etc.)
  • Priorities and growing a ruleset (vs. “I need to fix my security NOW”? This latter probably sells better?)

Short answer: yes. You need to find out what’s out on the network, device-wise, but at the same time it would be a good idea to start having security rule discussions.


The key point here is to keep making progress on multiple fronts, and to start providing results while recognizing there may be some unknowns and ambiguities that will take longer to sort out.

At one site I worked with, it took over a year to discover that beside some encrypted credit card transaction traffic, there were cash transactions being carried unencrypted over the network. Someone had a vendor add something without network/security being aware of it. And my guess is they told the vendor the network was secure, or the vendor didn’t care (“not their problem”).


Picking and Deploying a Micro-Segmentation Product

According to my list far above, picking and deploying a micro-segmentation product comes next.

I intend to NOT blog about product selection.

In part because I’m in no position to critique all the products that are out there. But also because I want this blog series to cut to the chase, and focus on what you do, rather than on product per se.


A lot of the Elisity materials lead with the deployment aspect, in part because they need to introduce some terminology relating to their solution.

I also intend to minimize the product installation and setup aspect of micro-segmentation. That’s product-specific. This blog series intends to focus more on the logic and process of implementing micro-segmentation.


In the next blog, I will very briefly describe the Elisity components, because that will be needed to understand some of the subsequent discussion and graphics. I’m going to assume you’ve licensed and installed some product that will provide information similar to what Elisity does. Or you have some other way to get such information.


Conclusion

Micro-segmentation provides finely-grained security enforcement comparable to deploying devices on different network segments (VLANs) and applying security controls on traffic between the device groupings.

Micro-segmentation enforcement is generally done in the network switches for scalability. It may or may not be stateful, depending on the switch capabilities. Along those lines, I’ll note in passing that Cisco recently announced switches with DPU’s built-in, enabling line-rate stateful enforcement.

Follow this blog series as I delve into some of the details of micro-segmentation. My overall objective is to leave you with a better idea of what micro-segmentation entails and what it can do. While touching on some interesting nuances but not getting bogged down in detail!


Links re Microsegmentation

Here are some links covering WHAT Microsegmentation is, and WHY do it. Pre-sales / decision-making.


Other Microsegmentation Links

And here are more technical inks that assume you’re doing Microsegmentation, and describe various aspects of implementation.


Miscellany

Reminder: you may want to check back on my articles on LinkedIn to review any comments or comment threads. They can be a quick way to have a discussion, correct me, or share you perspectives on technology.

Hashtags: #PeterWelcher #CCIE1773 #Microsegmentation #Elisity #Cisco

FTC disclosure statement: https://www.dhirubhai.net/pulse/ftc-disclosure-statement-peter-welcher-y8wle/

Twitter: @pjwelcher

LinkedIn: Peter Welcher

Mastodon: @[email protected]

BlueSky: https://bsky.app/profile/pjwelcher.bsky.social

IMAGE with three badges: Cisco Insider Champion, CCIE Lifetime Emeritus, Networking Field Day


HPE Aruba has the CX 10000 for a few years already, based on the same Pensando switch as Cisco uses. These are ToR/EoR switches not as far as I know for Campus-based stateful enforcement. In the Campus similar to SDA/ISE they create role based policies in Aruba central through Netconductor. The roles are assigned and monitored through Clearpass and the switches use SGTs for the microsegmentation.

回复

I am going to read this! Very relevant ;-) thanks Pete

要查看或添加评论,请登录

Peter Welcher的更多文章

  • Pete’s Take: Catchpoint at Cloud Field Day 22

    Pete’s Take: Catchpoint at Cloud Field Day 22

    Tech Field Day always produces such great technical content! However, it can be a challenge keeping up with it due to…

  • AI Ate My Blog on RoCEv2

    AI Ate My Blog on RoCEv2

    I acknowledge I’ve been a blog technology summarizer for quite a while. It served to help me broaden/solidify my skills…

  • AI Datacenter Switch Math

    AI Datacenter Switch Math

    Author: Pete Welcher, Coauthor: Brad Gregory This is blog #3 in a small series about Networking for AI Datacenters…

  • AI Requirements for Datacenter Networking

    AI Requirements for Datacenter Networking

    Author: Pete Welcher. Coauthor: Brad Gregory.

  • Quick Takes #2, February 2025

    Quick Takes #2, February 2025

    I’m working on some longer blogs that I hope to be able post in the next week or two. In the meantime, lots of exciting…

  • Quick Takes: February 2025

    Quick Takes: February 2025

    I’ve got some longer technical blogs in the works. For this week, it’s time again for some of my “Quick Takes”:…

  • Pete’s Take: Pain Points in Networking and IT

    Pete’s Take: Pain Points in Networking and IT

    It’s a new year, so time to look at how Networking and IT have been evolving. Ignoring the AI elephant in the room.

    1 条评论
  • Pete’s Take: Pondering NetOps/AIOps Strategy

    Pete’s Take: Pondering NetOps/AIOps Strategy

    What’s new in NetOps, including AIOps, and where are things heading? Some thoughts ..

    1 条评论
  • Pete's Take: AI/ML and Error

    Pete's Take: AI/ML and Error

    Artificial Intelligence (AI) has certainly received a lot of press lately. And achieved new levels of hype.

  • Book Review: Machine Learning for Network (etc.) by Javier Antich

    Book Review: Machine Learning for Network (etc.) by Javier Antich

    Welcome to 2025. I’m easing back into blogging for 2025 after fun and (sort of) relaxing holidays with visits by 3…

    1 条评论