Staying Current without Going Insane
Security is a moving target and chasing it can be positively dizzying. One of the most unrelenting and tedious security tasks on powerful systems like Linux is keeping its myriad applications, commands and libraries current. As anyone who subscribes to the BugTraq mailing list can attest, new software vulnerabilities are constantly being discovered, exploited and patched against. But given the hundreds of software packages installed by the typical Linux distribution, how can you ever hope to keep up?
The bad news is, you can't. Even if you had a way of instantly patching against every new vulnerability published on every vulnerability and incident report mailing list, sooner or later you would become the subject of an incident report yourself; some vulnerabilities don't become common knowledge until they've been exploited.
The good news is that since you can't possibly keep up, any progress you do make is a win. If you get cracked via a three-week-old software vulnerability, at least you're more elite than someone who gets cracked via a three-year-old 'sploit. (On the other hand, either way you're cracked.)
In all seriousness, though, from a purely statistical standpoint, fewer unpatched bugs means fewer exploitable vulnerabilities. Despite how thankless and endless the task seems, it is worth trying to keep your Linux software current. And luckily, recent versions of popular Linux distributions include new tools for automating much of this task, including SuSE's YaST2, Red Hat's up2date and Debian's apt-get. (Some of these tools are even secure!)
I'll start with a key piece of advice that I've arrived at after years of skepticism: wherever possible, stick to your distribution's supported packages. To many people, this probably sounds obvious; why build from source if you don't have to? But to some of us old-school types, building from source is as much a habit as a skill. Maybe this is because way back in the early 1990s (you know, back before we had Slashdot or powered flight), there was a lot less software for Linux than there is now, which meant that many of the things we ran we had to build from sources developed on other platforms. Additionally, the whole concept of packaging Linux into distributions was much less mature: there were fewer distributions, and they didn't change nearly as frequently.
(Have you ever had to compile ps from source because your brand-new kernel wasn't backward-compatible with the older version of ps from your distribution? Some of us have. We also had to walk ten miles to school each day through snake-filled swamps, etc.)
But things are different now. Don't get me wrong: I don't mean that we should all be slaves to binary packages. Sometimes you need features that are available in the very latest version of an application but not in your distribution's version of it; sometimes you want a leaner-and-meaner build than the one-size-fits-all juggernaut that your distribution provides. However, there are some important advantages to sticking with binary packages most of the time.
The first is convenience: downloading a large group of applications from a single site is faster and easier than downloading each from its developer's site (which is why we have distributions in the first place), and installing a binary package is much faster and less prone to mishaps than compiling from scratch. Convenience is not something we *nix bigots admit to valuing, but there it is.
The second advantage of packages is stability: the major distribution packagers put a good deal of testing and research into deciding whether to include a given application in their distribution, and if so, which version is the most stable and which compile-time options best suit their distribution. There have been notable exceptions to this, but most distributions nowadays do quite well with quality assurance.
Stability is unquestionably a big factor in security: where there are bugs there are vulnerabilities. Even bugs that don't have obvious security ramifications often can be exploited in, for example, denial-of-service attacks (the object of which is to crash a system, which some bugs do effectively).
This leads us to one of the more frustrating paradoxes in application security: although some of the most widespread vulnerabilities on the Internet stem from poorly maintained applications (i.e., obsolete and/or known-vulnerable versions), newer is not always better. The security community has rightfully lambasted Microsoft over the years for being slow to acknowledge and provide patches against security vulnerabilities in their products. But we complain equally bitterly when Microsoft does provide a speedy patch that affects stability because this potentially mitigates any benefit derived from fixing the original bug.
Ignoring for a moment stability's desirability in its own right, ask yourself this: supposing an application has an obscure buffer-overflow condition that is theoretically exploitable for root privileges (but only by a skilled assembly programmer with a working knowledge of the RC4 stream cipher), and you patch it with code that introduces a denial-of-service opportunity that can be exploited even by attention-span-deficit script kiddies—how much security have you really gained? The answer to this will vary depending on the precise circumstances and on who you ask, but the important point is that software upgrades often have ramifications of their own.
I'm beating this point to death, but it's an important point because it debunks the notion of instant software updates being some sort of panacea. (Plus it's been on my mind for a long time; it's taken me years to fully understand why, for example, the very latest version of OpenBSD still ships with a hack of BIND v.4.9.8, which in computer/dog years is really ancient.)
It's also important because, getting back to my original topic, there's something to be said for letting your Linux distributor make the difficult patching decisions, and thus for waiting to patch a vulnerable application until your distribution releases an official (and hopefully tested) patch, which of course is part of the deal when you rely on binary packages rather than cold hard source code. It's difficult enough to keep up with Linux distributors' application updates; I can only imagine how futile it would be to patch and recompile all my critical applications myself and waiting anxiously to see whether the patching broke anything else.
So to summarize these pearls of wisdom (at least I hope they're pearls, instead of some other small spherical secretion): keeping current is a virtue, using binary packages is a virtuous form of laziness (props to Larry Wall) and relying on secondhand security updates from your Linux distributor rather than “going cowboy” isn't laziness at all, it's prudence.
|Non-Linux FOSS: libnotify, OS X Style||Jun 18, 2013|
|Containers—Not Virtual Machines—Are the Future Cloud||Jun 17, 2013|
|Lock-Free Multi-Producer Multi-Consumer Queue on Ring Buffer||Jun 12, 2013|
|Weechat, Irssi's Little Brother||Jun 11, 2013|
|One Tail Just Isn't Enough||Jun 07, 2013|
|Introduction to MapReduce with Hadoop on Linux||Jun 05, 2013|
- Containers—Not Virtual Machines—Are the Future Cloud
- Non-Linux FOSS: libnotify, OS X Style
- Linux Systems Administrator
- Validate an E-Mail Address with PHP, the Right Way
- Lock-Free Multi-Producer Multi-Consumer Queue on Ring Buffer
- Senior Perl Developer
- Technical Support Rep
- UX Designer
- Introduction to MapReduce with Hadoop on Linux
- RSS Feeds
Free Webinar: Hadoop
How to Build an Optimal Hadoop Cluster to Store and Maintain Unlimited Amounts of Data Using Microservers
Realizing the promise of Apache® Hadoop® requires the effective deployment of compute, memory, storage and networking to achieve optimal results. With its flexibility and multitude of options, it is easy to over or under provision the server infrastructure, resulting in poor performance and high TCO. Join us for an in depth, technical discussion with industry experts from leading Hadoop and server companies who will provide insights into the key considerations for designing and deploying an optimal Hadoop cluster.
Some of key questions to be discussed are:
- What is the “typical” Hadoop cluster and what should be installed on the different machine types?
- Why should you consider the typical workload patterns when making your hardware decisions?
- Are all microservers created equal for Hadoop deployments?
- How do I plan for expansion if I require more compute, memory, storage or networking?