The Security Predicament

The more I follow network security issues, the bigger the job of security appears. Online business activity has been with us since the 1990‘s and many managerial structures still don’t have Chief Information Officers, much less Chief Information Security Officers. Even if these positions exist, authority is limited. As long as InfoSec isn’t represented in the board room (and is considered a second-tier matter - a cost to be minimized), as users we’re on our own. Other than compliance with policy in order to satisfy the boss, there is a responsibility to ourselves to handle security better than do our organizations.

My latest formula with which to observe the broad security situation is this: there are three zones of concern, namely You, Them and the Connection between You and Them.

You - this includes any account or device under your direct control or ownership.

Your Connection - this is the Internet Service Provider, or more properly the server(s) through which you get to the outside world, operated by their administrator or administrators.

Them - this is the other end of the conversation, including businesses, government entities or other individuals or groups served by their Connections.

Problems are possible in any of the zones, but we should consider mitigation possible in only two of them. Let’s say you’ve locked down your own devices and are using best practices. Let’s also say the server admin between you and the world is also locked down and behaving cautiously. What happens next? You connect with someone or something outside the control of either you or the admin.

One can affect or at least observe problems at the position of oneself or one’s own connection to others. The situation on the other side of the connection is anyone’s guess. So:

1. Maintain your own material and situation as best you can, using my “go to the PUB” advice (pay attention to Passwords, Updates and Backups).
2. Use whatever influence you have to assure the quality of your Connection, and attempt to give out as little information as possible through this connection to the other end of the communication.
3. Monitor the situation with the people and organizations with which you communicate, and regarding the information you’ve given to these organizations and which you therefore no longer control. For example, watch your credit card statement and get with that institution in case of bogus charges. Challenge inaccurate information about yourself.

... It looks like maintaining your stuff, giving out as little info as possible and monitoring the other end would be the only three courses of action capable of improving network security. At least relative to the user. In the larger world, it’s still the wild west and could become even more so. There are APT’s (state-sponsored bad actors), organized crime elements, software bugs, hardware bugs, poor decisions within software development projects, political protest hacks, social media abuse, natural disasters, program or code incompatibilities and more things that won’t fix themselves.

We’re on our own.

FREE RANSOMWARE MITIGATION TOOLS

Here's a site that helps if one of your machines has been struck with a ransom program that encrypts your information - or claims to have done so:

Malware Hunt Team

In order to identify the problem you can upload code snippets or other associated info, like ransom notes or email addresses. Keys to decrypt your machine may be available immediately.

Firewall Adventures

So my - ok, one of my - bottom tier computers which was running the pfsense firewall will now get a new and different firewall for me to play with. Apparently pfsense no longer supports architecture below 64 bit, or I can't find the link to the older-box versions. So I blew it away and installed ipfire, with which I have no experience and know nobody with any (though I haven't yet brought it up at the Linux club). Figured I'd wreck it a little first.

The box is of 2005 vintage, a Dell of the WinXP era. Maybe the installer didn't expect that, since although the root and admin passwords were accepted by it during the setup routine, neither of them worked on reboot. On reinstall, root worked but not admin - and passwd informed me that such a user was not present, as in not even created. But CLI works and blinky lights are happening on the downstream router. I'll update within this post below as I discover how this new-to-me product works; it's evidently maintained by a company that has both supported and free versions.

UPDATE #1, 9/8
After one or two reboots and/or reinstalls, the installer kept putting in a root password that the system wouldn't remember. A forum suggested entering the password during installation without using the "Return" to get to the next line, but using "Tab" instead. The theory was that "Return" might insert screwy control characters but that "Tab" would not. This seems to have been true, fixing the root PW creation issue. I had to add the "admin" user manually (via useradd, not the web interface) but it seems to be successfully sticking as well. Now running through as many simple commands as I can think of, and it's ok after a half dozen reboots. I have an iso of Smoothwall in case something craters me later.

UPDATE #2, 9/12
And it forgot the root password again some more repeatedly... so I'm giving up on the old Dell (a GX520 Optiplex) and am now trying ipFire on a Dimension 2400 of equally ancient vintage but better specs (faster proc at 2.6ghz Celeron D and more memory at 2g). The ncurses based installer worked after the 4th or 5th attempt and the root PW has been remembered for days now. No routing to switch yet but pinging out and GUI adds/removes rules; will do a new post when most things work and there's news to report.

UPDATE #3, 10/8
Holy ruleset, Batman: this is all a bit more complex than I figgered. Routing out to the world is fine but the intrusion detection engine isn't turning over. Simple fw rules are operational through the GUI AFAIK and can test, so all seems ok with this package and box. BUUUT I was really interested in trying out its built-in IDS/IPS (intrusion and detection) which both uses signatures and behavior methods of detection/blocking (this would provide another perspective in the ongoing architecture debate between Snort and its later rivals). See, they used to use the venerable Snort but changed to Suricata a few releases back, so I dug up the command line commands for that. I think Suricata's config file was mangled at installation; the good ol' linux locate(updatedb) command verifies that the config file is where it's expected to reside and simple things like versioning display with -V happen right. I even catted the whole thing and it looks complete to my untrained eye. IPS won't run, though.

In my defense, I'd still be running pfSense if they had continued with the 32 bit version (with which all services ran when I asked them to); I just may get some raspberry Pi or other and throw pfSense on there - unless I reinstall IPFire and get it to run. Wish me luck! Until a better theory comes along, I'll blame Dell.

More Good Than Bad

Here's an article from Foreign Policy magazine that's better than the usual take on computer security problems. The writer commits the common error of tracing computer warfare oversight mistakes primarily to recent US Presidential administrations (when protocol vulnerabilities in TCP/IP date from about the time Stevie Wonder released "Livin' For The City" - RFC 675, Specification of Internet Transmission Control Program, December 1974). There's enough blame to go around. But what's properly covered by the article is considerable: the author points out the need for a "digital Geneva Convention" agreement so we'll have a guideline that identifies hostile action in a network context. That would help protect penetration testers and other cybersecurity researchers from unfair prosecution.

She even apparently coins a new and useful definition: "Cyberwar is the continuation of kinetic war by plausibly deniable means". Tarah Wheeler's article also confirms some of my own gripes of the past few years.

I'd prefer that it didn't.

Let's Get Specific

The previous post pitched the idea that a shifty operator might craft a customized attack on each of the citizens individually in a target area, region, city or even country that had a digital presence or litter trail. These days, that could be most people. Once every citizen in the target zone had a file built, a script could run a series of attacks on citizens using the file information on say, person 349384. In sequence, that script could try password cracks on email accounts, simple google searches on the uncontrolled net, requests for city services changes or disconnections, stop payments on checks or debit card uses, cancellation of credit cards, intentional violations introduced so as to trigger fraud alerts interrupting service... and so on.

Ever see the film Pacific Heights (the one I mentioned two posts ago under "Prediction")? It's from 1990. Released prior to internet growth, it shows how a dishonest renter goes after the property owner with legal and municipal trickery. All very low-tech. And as I also indicate earlier, a personalized digital Pearl Harbor sneak attack could just be the equivalent of misplacing your digital "wallet" - except that it could happen to 70% of the population at the same time. Add to that some assorted nationwide traffic signal and industrial control misconfigurations and you've at least got a rough commute home.

Anyway, to pull off the personalized attack you'd need the information on the marks before starting. So what would an assembled, finished citizen file have in it? Oh, maybe something like this:

name
address
city
state
country
previous addresses
phone number(s)
phone provider
previous phone providers
smart phone install application list
known email addresses
birthdate
tax number or SSN
DMV license
lawsuit case numbers
previous names
previous phone numbers
voter registration info
previous voter registration info
party primary affiliation from voter registration
employer
previous employers
declared income
religious or PAC memberships or affiliations
military service
hospital record control numbers
bank customer info or profile
insurance company customer info or profile
brokerage company customer info or profile
union membership info or profile
degrees or certifications from posted resumes/CV's
entire content of posted resumes if available connections from known email to social media accounts
connections from known email to conventional HTML posting boards
internet service provider
previous internet service providers
social media screen names if different from account name
charitable contributions or affiliations
any previous web activity recorded by archive.org

--- What? A world in which information on someone or something from many sources could be combined in one place? Not possible, you say. Larry Ellison says we got there five or six years ago.

Broadly, the issue is that much of our personal information is stored by businesses or public entities that have poor security practices. We should clean up our security act, of course, but as to information compromise, that ship has probably already sailed. We should assume compromise and proceed from there.

[What can one do? My earlier post here keeps it simple.]

Prediction

Ok, here's a problem situation of which I feel comfortable in predicting the rise, since I'm sure scads of persons have already thought of it.

Suppose a government military agency of a country not friendly with us were to develop a data mining capability extensive in nature and well-funded. Suppose that they were to understand that the mining itself could be weaponized and that one need not develop a cloud or botnet to distributed-denial-of-service (DDOS) backbone servers of a country's infrastructure "off the air", so to speak... but to attack in a way that avoids the network completely.

Suppose that said agency were to realize that the number of citizens in Country X (like ours) was not a terribly large number given the scale upon which a computer can operate. Suppose it were to build a file (using known data breaches and publicly available information) on every citizen in this country. With those files as targets, one simply could then prepare a rudimentary "identity attack" script to mess with every citizen's digital presence. Some citizens would not have a presence yet (being toddlers) but many at even a young age now have smart phones, laptops, smart TVs and so forth - lots of things with IP addresses. And the parents, who function as their own NETWORK ADMINISTRATORS, are possibly not current on their firewall rule skills... or any others...

The file on every person could contain email addresses, IMEI device numbers for cell phones, MAC addresses for desktop or laptop computers, social security numbers and birth dates (particularly if intimate processes like tax filing had been done via smart phone) and if a real name and public viewability were part of a Facebook or other social media presence then that would be interesting to include as well. There was a movie years ago called "Pacific Heights" (Melanie Griffith, Michael Keaton, Tippi Hedren) which told of a renter that tormented his landlord by using every low-tech form of harassment imaginable. This was in 1990 so the 'net as we know it had not arrived (pre dub dub dub popularity). The tormentor used such things as junk (snail) mail, the city complaint process and physical blocking of nearby driveways to create a buildup of problems in the lives of the property owners. The same sort of mischief could be leveled at the lives of citizens person-by-person with scripted attacks to change account passwords, delete address books and perhaps close credit card accounts.

Consider how many people have had the same email account for years and use the same simple password for many different things. Consider how many people do banking, insurance and other bill payment over computers or phones. Consider how insecure government and medical records no doubt are. Banks are probably the most secure overall as electronic entities in society but you hear of a breach at least a few times a year at some financial institution.

I'll call it, not DDOS - distributed denial of service but PDOS - personal denial of service. Simple scripted attack directed at only one could become like losing one's purse or wallet.

And everybody in the target country could lose their stuff all at the same time...

One Wishes One Didn't Have To Repeat This...

...but just as in the days around five years ago or ten, when virtual computing was this new thing promising cheap, automatic solutions for everybody's issues, we STILL see three pesky security problems which virtualization adds to ALL THE OTHER SECURITY PROBLEMS that were present in computing in the first place. Here they are again.

THE USER INTERFACE

The place you go to remotely administer a rented server at some unknown location (maybe they'll tell you which data center the virtual machine's in; maybe it's not an issue) is a web page, with all the inherent vulnerabilities those things have. SQL injections, buffer overflows - the usual suspects. Does your infrastructure provider use secure coding practices? Maybe, maybe not.

THE HOST HYPERVISOR

The big program inside which the virtual computers live is called the hypervisor. It's a very big program. Programs - particularly those that talk to networks - are vulnerable to all kinds of nasties. Updates to those programs are needed to counteract the nasties. Is the hypervisor program on the hardware host at your provider that runs your VM updated? Was it written by the virtualization product maintainer using secure coding practices? Maybe, maybe not.

THE PROVIDER EMPLOYEES

Did you hire the people that run the hosts that your VM lives on? No? Then you can't fire them. And don't even know who most or any of them are. Do you trust them? Maybe, maybe not.

Other Concerns

Do you maintain your VMs via an automatic payment from a credit card? What happens if a payment fails to go through? POOF - all your VMs could disappear, along with settings, data, transactions, virtual switches and so on. That's a nasty surprise, particularly if your backup locations are also VMs at the same provider that are paid for in the same way. And unless you're paying through the nose for managed hosting (where designated admins do stuff for you), there's the nasty surprise of lack of administration, meaning that you're behind on updates, disk space checks, disabling of unneeded services, log rotation and all that stuff.

Not to bug you, but these things haven't changed since years ago, when first I griped about them. Happy networking!

UPDATE: Another issue I remember from my tech support days is where a certain major infrastructure provider only a few years ago was discovered to be recycling VM's! OS loads at the provider where I used to work took about as long as a real computer (30-45 min.). A way to improve that - which they may have implemented by now - would be to image a new machine quickly from a master image. But one of our competitors was providing (apparently) already-built builds of Windows Server that had been previously used by an earlier customer. To recycle, I suppose they just ran some command that reset to defaults. Perhaps they came up with a powershell script. It would have provided a "new" computer (virtual machine image) to a new customer who was no doubt impressed by the speedy delivery! But that same customer may have been confused by the data, configurations and users left in the build from some previous owner. Hilarious.