Jeff Crume, known as "the Security Guy" at IBM and an adjunct professor at NC State University, opens IBM Technology's cybersecurity architecture series, based on the 400-level enterprise security architecture course he teaches. He starts with the fundamentals: five principles every security design should follow, and one it never should.
This article is our own summary, with short quotes, from the video's automatic transcript. Watch the full video above for everything that was said.
The short version
- Defense in depth: never rely on a single security mechanism. Layer them so that if one fails, the others still hold.
- Least privilege: give people only the access they need, only for as long as they need it, and harden systems by removing what isn't used.
- Separation of duties: no single person should be able to compromise a system alone; force collusion.
- Secure by design: build security in from requirements to production, not as a bolt-on at the end.
- Keep it simple (KISS): "complexity is the enemy of security." If the right thing is harder than the wrong thing, people will do the wrong thing.
- Never: security by obscurity. "Secrecy and security are not the same thing." The only secret should be the key.
The five principles to follow
1. Defense in depth
Crume starts with the castle: thick walls, then a door (a weak point), then a moat, a drawbridge and maybe an angry dog. The modern version is a user on a workstation reaching a web server, an app server and a database, with a layer at each step:
- Multifactor authentication (MFA) to confirm the user is who they claim to be.
- Mobile device management (MDM) to enforce patches and password policy on devices, and EDR (endpoint detection and response), "a next generation antivirus."
- Firewalls around the web server and the sensitive zones behind it.
- Vulnerability testing of the web and app servers.
- Encryption and access controls on the data.
The goal is no single point of failure, and a system that "fails safe."
2. Least privilege
Only people who can justify a business need get an access right, and "the clock is ticking": access is reviewed and removed when no longer needed. Two related practices:
- Hardening: turn off default services you don't use (such as FTP or SSH on a web server), since each one expands the attack surface; remove unneeded IDs, rename default ones such as "admin," and change every default password.
- Stopping privilege creep: when someone changes jobs, administrators often add new rights "just in case" and leave the old ones. He says just-in-case "is just the opposite of principle of least privilege." The fix is a recertification campaign at least once a year to confirm every right is still needed.
3. Separation of duties
There should be no single point of control. Like a door with two locks and two keyholders, a compromise should require two or more people to collude, which is hard because it is hard for many people to keep a secret. In IT, the person who requests access, a funds transfer or a delivery can never be the person who approves it.
4. Secure by design
You don't build in an earthquake zone and make the building earthquake-proof afterwards. Across the lifecycle (requirements, design, coding, installation, testing and production), security must be present at every stage, not added once the system is live. Designers, administrators and users all share responsibility, but it begins with the designer. The goal is a system that is "secure out of the box."
5. Keep it simple, stupid
Security teams often build a maze that only the good guys have to walk through, and people find ways around it. His example is password rules: complex formats, very long passwords, a different one on every system and frequent forced changes. Users respond by writing one password down and reusing it everywhere. Make the system just complex enough to keep attackers out, but not so complex that it gets in the way of legitimate users. The obstacle course of defense in depth is meant for the bad guy, not the good guy.
The one to avoid: security by obscurity
Relying on secret knowledge to keep a system safe doesn't work. Crume cites Kerckhoffs's principle: a cryptographic system should be secure even if everything about it is known except the key. He says to "run and not walk" away from anyone selling a proprietary, black-box encryption algorithm that "can't be broken." That only means its inventor couldn't break it; given access and time, the world will. Trusted algorithms like AES and RSA are published for anyone to study. He calls this "glass box security," and the same applies to operating systems and applications.
What it means for your business
- Turn on MFA everywhere and add endpoint protection, firewalls and encryption so no single control is your only defense.
- Review access at least once a year. Remove rights people no longer need, and stop granting access "just in case."
- Split request and approval for payments, data access and admin changes.
- Ask about security at the requirements stage of every project, not before go-live.
- Simplify the safe path. A password manager and single sign-on beat complex password rules people will work around.
- Be wary of secret, proprietary security. Prefer published, widely reviewed standards.
Video details
- Title: Cybersecurity Architecture: Five Principles to Follow (and One to Avoid)
- Presenter: Jeff Crume, IBM ("the Security Guy"), adjunct professor at NC State University
- Published by: IBM Technology on YouTube
- Watch: youtube.com/watch?v=jq_LZ1RFPfU
- Note: quotes come from the video's automatic transcript and are lightly cleaned. Check the video for exact wording.
Want to put this to work?
We help owners turn AI trends into a plan, a product and a launch. Book a free call or see our services.
