How to Cheat at Managing Information Security

Ben Rothke writes "Mark Osborne doesn't like auditors. In fact, after reading this book, one gets the feeling he despises them. Perhaps he should have titled this book 'How I learned to stop worrying and hate auditors'. Of course, that is not the main theme of How to Cheat at Managing Information Security, but Osborne never hides his feeling about auditors, which is not necessarily a bad thing. In fact, the auditor jokes start in the preface, and continue throughout the book." Read the rest of Ben's review. How to Cheat at Managing Information Security author Mark Osborne pages 302 publisher Syngres rating 8 reviewer Ben Rothke ISBN 1597491101 summary The adventures of an information security professional and his efforts to secure corporate networks

The subtitle of the book is 'Straight talk from the loud-fat-bloke who protected Buckingham Palace and ran KPMG's security practice'. Essentially, the book is Osborne's reminiscence of his years in information security; including the good, the bad, and more often then not, the ugly.

The book is written for someone looking to develop an information security program, or strengthen an existing program, to ensure that all of the critical technology areas are covered.

The thirteen chapters of the book cover the main topics that an information security manager needs to know to do their job. The author candidly notes that this book is not the most comprehensive security book ever written, but contains most of the things a security manager needs to get their job done. The author also observes that information security is different from other disciplines in that there are many good books about disconnected subjects. The challenge is getting the breadth of knowledge across these many areas, which is quite difficult. The challenge of information security is to effectively operate across these many areas.

Chapters 1 and 2 deal with the information security organization as a whole, and the need for information security policy. Chapter 1 details the various areas where a security group should be placed, and describes the pros and cons of each scenario. As one of the scenarios which place information security below the head of audit, Osborne notes that 'if you have any sort of life, you don't want to spend it with the auditors, I promise you'.

Wherever the security group is placed in an organization, its ultimate success or failure is likely to be determined by its level of autonomy and independence. Unfortunately, in far too many organizations, information security is not given that liberty. It is often placed in a subservient role to groups with opposing interests. Any security group or security manager placed in such a situation should likely start working on their resume.

The scenario is described in 'Practical Unix and Internet Security' where author Professor Gene Spafford spells out Spaf's first principle of security administration. This principle states that 'if you have responsibility for security but have no authority to set rules or punish violators, your own role in the organization is to take the blame when something big goes wrong'. Spaf's principle is a cruel reality faced by many of those responsible for information security.

Between those chapters and a few more auditor jokes, Osborne makes the blatently obvious observation that wherever possible, one should eradicate single points of failure. As a corollary to this, Osborne notes that while trying to eliminate these failure points, companies will often build redundant systems. Part of their admiration for these redundant systems is the hope that this will simultaneously reduce performance bottlenecks. But these companies do not realize that the routers, firewalls and switches are not the bottleneck, rather it is the software application which is the bottleneck.

Osborne plays the role of contrarian in chapter 8 when he asks why we need firewalls. He notes that if every database maker, operating system programmer and CRM/ERM vendor put as much effort into security as the firewall vendors do, then there would be no need for firewalls. Furthermore, if each system administrator worked as hard on security as the typical firewall administrator did, and devoted as much time to hardening their servers and laptops as they did; then centralized firewalls would likely not be needed. Given that the firewall-free reality is not happening any time soon, chapter 8 provides a lot of good information on everything you need to know about firewalls.

Chapter 9 is about one of the most maligned security tools, the IDS. After providing an anecdote about a network manager who did not understand the fundamentals of how DHCP operates, and how he used Snort to debug the problem; Osborne provides a meaningful piece of security wisdom when he notes that IDS can help any network or security person understand network traffic. These devices can even give you information on new attacks and how they can be mitigated. But for an IDS (or any security hardware or software device for that matter) to be truly useful, a security professional needs to understand their IT infrastructure, the mechanics of networks and applications and the risks involved. Those who don't understand those three things will only be able to use these security technologies with minimal benefit.

Overall, How to Cheat at Managing Information Security, is an informative and often entertaining introduction to information security. For those that want to get a good overview of the core elements of information security, or strengthen their existing knowledge base, they will find this book to be an informative and valuable read."

You can purchase How to Cheat at Managing Information Security from bn.com. Slashdot welcomes readers' book reviews -- to see your own review here, read the book review guidelines, then visit the submission page.

Comments

You can never do away with a firewall. Ever.

[growse]

I'm not sure the the comments about firewalls are accurate. Sure, if every software maker paid attention well to security, then we'd be in a lot better position than we are now, but I'm not necessarily sure that building firewall-level security into every application is a good thing.

For example, if I want to restrict the access to a particular service to a few ip addresses, I'm more likely to do this on my firewall than on the service myself. Sure, the people who make the service could include that functionality, but I like the separation of security out away from the application. I like the fact that I control all my access in one place, and not across hundreds of application-specific config files. I believe modern filewalls can do all sorts of clever things such as coping with DoS attacks, stateful examination of network traffic etc etc etc. Can you imagine what it would be like if every single service had that functionality built in, but implemented slightly differently and with slightly different types of weakness in each one? Think of the duplicated functionality and bloat!

There's no such thing as software which is immune to malicious attack, but I like to keep my security weaknesses all in one place, and minimize them buy buying my firewalls from a company that has track record and experience in security issues, rather than a company that makes an ftp server for a living.

auditor jokes thread.... :)

[advocate_one]

those who can, do... those who can't, teach... and those who can't teach, audit...

Ain't that the truth

[Billosaur]

The scenario is described in 'Practical Unix and Internet Security' where author Professor Gene Spafford spells out Spaf's first principle of security administration. This principle states that 'if you have responsibility for security but have no authority to set rules or punish violators, your own role in the organization is to take the blame when something big goes wrong'. Spaf's principle is a cruel reality faced by many of those responsible for information security.

This same principal applies to a great number of jobs in IT. If it's your job to create content for display on the Internet/Intranet and you aren't given the proper access and tools to get the job done, it is often your fault for the failure, even though you're at the mercy of others. Same goes for bad project management; if a project is slow or fails, it's not because the project manager was an ignorant troll, but was in fact due to the "inability of programmers to meet their goals," even though the goals and timelines were unreasonable and ultimately futile.

Re:Ain't that the truth

[Karellen]

"Authority and responsibility must be equal - else a balancing takes place as surely as current flows between points of unequal potential. To permit irresponsible authority is to sow disaster; to hold a man responsible for anything he does not control is to behave with blind idiocy."

-- Robert A. Heinlen, _Starship Troopers_

Defense in depth.

[khasim]

The only reason I'd like to see decent firewalls on the workstations is for more "depth" to my security model.

With a firewall, it is a single point that can be cracked. If that is your only security point, you'll be wide open if it is every cracked. And "cracked" also includes "someone brings in an infected laptop".

Now, on the workstation level ...
#1. No services running that aren't absolutely necessary.

#2. No open ports that aren't absolutely necessary.

#3. Any open ports/running services will ONLY accept connections from my servers / admin workstations. Anything else is logged and I am alerted.

Most of this can be accomplished with an IDS. I'd like the workstation firewalls AND the IDS. Having multiple checks is good. (and the firewall, you need the firewall)

Memories

[Aqua_boy17]

Osborne never hides his feeling about auditors

I had to smile when I read this as it took me back to my first financial audit years back. I nervously awaited our internal auditor who had a reputation for being completely ruthless in his approach and did not give a fig if heads rolled as a result of his findings. When he first met with me, he began with a story: "You know, we auditors are often compared to soldiers, and your brothers-in-arms in the field. The only difference with us is that we fix bayonets to our rifles, and go around stabbing our own troops while they lay wounded on the ground. Now, with that out of the way let's begin your audit." I suppose that would have made most people nervous, but I was charmed by his candor, and actually wound up getting along with him quite well in the end.

I suppose my point in telling that is that you can look at auditors two different ways. Either they're there to help you, or they are there to get in your way and point fingers. I believe that most genuinely good auditors try to be more like the former and less like the latter. And you can learn a lot from them if you remain objective and cooperative. God help you if you get the other kind though as they are usually just nothing but self-promoting tattle-telling toadies.

Re:Memories

[Darth Muffin]

That reminds me of my experience a few months ago. We were in for our Sarbanes-Oxley (SOX) audit. One of the policies to comply with SOX is not to allow any non-company machines on the network (finally! Been wanting that for years.).

Of course the first thing the auditors want to do is plug into our network so they can get their email. I said no, because if we do they it violates SOX and we fail their audit. They asked how they're supposed to audit us then if they can't use their e-mail? Not my problem, refer up to management :)

I actually won this round. We ended up isolating a portion of the network so they could have access straight to the Internet.

Re:Memories

[gosand]

Of course the first thing the auditors want to do is plug into our network so they can get their email. I said no, because if we do they it violates SOX and we fail their audit. They asked how they're supposed to audit us then if they can't use their e-mail? Not my problem, refer up to management :)

We had a similar discussion with our auditors. It wasn't SOX, it was SAS70, but still a process audit. What I thought was hilarious was when I walked past the conference room that the 3 auditors had occupied, and there sat their 3 laptops, screens unlocked and nobody in the room. The urge to set their background image to goatse was almost overwhelming, but I thought better of it.

Re:Be like Slashdot

[Linux_ho]

Make dupes, grammer errors and spelling errors

Pot, I'd like to introduce you to my good friend kettle. Kettle, this is pot. I believe you two have a lot in common, so play nice, OK?

Re:Comparison To Security Engineering?

[whizistic]

Yes. How to Cheat at Managing Information Security is to Security Engineering as reading about morse code is to designing a fiber optic network. Hope that helps.

Re:You are missing the point

[asdavis]

Ok, lets assume that there is a huge datacenter behind the firewall. What does the firewall do to protect the datacenter? Generally, you do not allow direct inward access from the Internet into a DC proper. Rather, you use a DMZ to host exposed nodes. So in the end, for the DC, the firewall is just a router. It allows traffic from select DMZ nodes to access hosts inside the network. That's really the function of a router. However, we often filter as well to ensure that only the minimum ports and services that are required are passed. Why do we do this? Because we are concerned that the DMZ nodes might get compromised and be used as a gateway into the environment to compromise nodes on the internal network. Why are we concerned about this? Because we have come to accept that the vendors of server platforms, operating systems, middleware, databases, etc ship fundamentally flawed products. They are buggy, exploitable and are not carefully coded to prevent compromise. We trust firewalls, because they are very carefully coded and great pains are taken to ensure that they cannot (generally) be compromised. That is the author's point. Let's spend the time ensuring that products are as well coded as the firewalls and we do not need a firewall. Is this likely to happen? Probably not, but it is a valid point.

Re:The Necessity of Auditors

[bzipitidoo]

To the contrary. Necessary structure is good, and IT workers know that. A lot of managers are bad. They try to impose methodologies that do not fit the problems, and demand excessive structure and paperwork. They demand schedules, and then superficially alter them until what might have been a reasonable best guess is now a death march. They think things should be done "their" way, and get in the way. That is extremely irritating when they demonstrably don't know what the heck they're doing, and can't or won't see that. IT workers don't want to hand those guys any more rope than they must; they know it's only going to be used to hang everyone. What you see as rebelling against structure itself, I see as rebelling against the abuse of structure, and against those who think the "art" of management means they don't need to know anything about the technology or science, let alone the boring technical details. They only need to know how to make engineers be productive, avoid being blamed for problems, and get the credit.

Of course a classic way to avoid blame is to make stupid rules and then point to the engineering geeks' supposed lack of discipline for not following those rules. Really great when there's a handy stereotype available. Most people aren't going to provide accurate status reports with useful content if the main use of it will be against them. I'd say not giving a status report at all is more honest than giving a status report that's nothing but evasions, fluff, misdirection, boilerplate, and such garbage.