CTIO Road Warriors: Providing Secure Remote Access
James Babcock Hughes CISSP, GCIA, GSEC
Updated August 8, 2008
Introduction
CTIO folks who attend conferences, visit other observatories, or travel to meet with scientific collaborators or vendors still need to access CTIO's information infrastructure from a possibly unsafe environment. Even those who don't travel that frequently still wish to log in from home or other places to read their email, work on an important project or just check that things are OK.
The security risk to CTIO due to outside users is that their computer, laptop or even PDA, besides being lost or stolen along with its valuable data (like passwords), could also become a vector through which viruses and other malware can breach CTIO's defenses, either from home, during the trip or later when mobile devices are reattached to the CTIO network.
This document looks into the security issues mostly specific to CTIO's traveling users with a great deal of overlap with remote home users: the topics are access methods, hardening techniques, suggestions, and some specific solutions which CISS uses to address these issues. At risk of spoiling the punchline, you should make sure your computer is properly protected, and in particular have CISS check out your traveling laptop and make sure it is up to snuff!
Motivation
An interesting 2004 article on www.slashdot.com states that a brand new computer running Microsoft software attached to the internet from an unprotected network (just attached -- worms for example don't need email to infect a computer) without firewall or anti-virus is infected within 20 minutes on average (!). A few years earlier, this average was 40 minutes, and today you will typically be able to see the first probes from malware within 5 minutes of attaching a computer to a network for the first time.Fairly recently, a CTIO Windows computer without any users was inadvertently left running outside the firewall. Four hours later, the entire CTIO commodity connection to the Internet was completely saturated with traffic -- emanating from the now infected machine!
In order to shelter users from the internet jungle and the wild woolly web, and to guarantee the confidentiality, integrity and availability of a user's data and applications, best industry practice requires the architecture of a computer infrastructure to provide defense-in-depth against hackers and their software.
Defense-in-depth means that security is layered: protection is afforded by a Firewall, by a network Intrusion Detection System, by a host based Intrusion Detection System, by encrypted data transfers, by non-trivial passwords, by virus and spyware scanning software, etc. If one layer fails, the evil hacker must still break through more layers before actually taking over a machine or stealing confidential information like your credit card number. The hacker may have an exploit that breaks through one layer, but then must analyze the defenses and find another exploit to break through the next layer, and so on.
Even if the hacker happens to have exploits for ALL the layers, applying them all takes time, hopefully enough time to detect the hacker and shut him down (which is what an Intrusion Detection System does), or enough time to make the hack moot (the targeted user disconnects and moves on),or enough time to frustrate the hacker so he/she will just give up and seek less defended victims.
Unfortunately, remote users, particularly mobile ones, simply will not have the same defenses at their disposal as the main institutional network. For CTIO and most organizations, traveling users are simply more vulnerable while they are traveling.
That which is desired is to cover CTIO's remote and mobile users with as effective a layered defense as possible which still adequately protects them and the organization against security incidents.
Access to CTIO
Services by CTIO for remote users includes public access provided to everyone with an internet connection (in other words, access to the CTIO public ftp site to extract files, access to CTIO's external WWW site, and sending email), along with vectored access exclusive to CTIO staff and visitors such as remote login to the internal CTIO network via SSHv2 or transferring email from CTIO's email server via IMAP.Secure, encrypted remote logins can be provided using SSH Secure Shell. SSH packages usually come with a secure FTP based on the same SSL libraries. At CTIO we generally use freeware versions of SSH such as puTTy for Microsoft based systems. See the CTIO WWW security pages.
Remote IMAP connection to our email server ctiosy.ctio.noao.edu is possible through VPN or the older but quickly obsolescing setmail authorization granting utility.
The industry 'best practice' for allowing remote access to an organization is the one also preferred at CTIO: network access via VPN. While not a security panacea, VPN offers the best combination of transparency, flexibility and protection for the organization and the remote user.
Virtual Private Networks
VPN, Virtual Private Network connections, are a simple and safe means to remotely connect up to the CTIO network. They can provide authentication (protection against spoofing and session hijacking), and confidentiality (protection against stolen information), between the remote user's machine and CTIO: via VPN a remote computer actually becomes a virtual part of the organization network and its traffic "appears" to be emitting from within CTIO. As a result, a laptop connected to CTIO's internal network will function almost identically to one connected via VPN from far away: most applications will work transparently in the VPN environment.A nice side benefit for computers connecting to CTIO via VPN is that they inherit many of the resources available to other CTIO computers, for example access to intranets open to CTIO, IP number based subscription services, journals, etc.
However, since the remote machine is still connected up from the outside internet even while VPN'ed into CTIO, if taken over by malware it can be used as a trojan horse used to bypass CTIO's normal defenses. Even worse, since much of the VPN traffic is encrypted via IPsec, security appliances such as IDS are unable to scan remote machine traffic for malware.
Fortunately, VPN these days is handled by firewall devices, such as CTIO's CISCO PIX, that impose the same or tighter security requirements on a VPN client as they do to other internal organization computers. In addition, we require both a VPN group password and a valid username and password before our RADIUS authentication server will allow the VPN connection to establish.
All this helps, but does not prevent a machine that has been taken over by malware from causing problems via VPN.
The upshot is that the remote computer or laptop needs to have its own perimeter defenses in place, such as personal IDS or firewalls, anti-virus and anti-spyware software installed. A CISCO VPN client is typically required for running VPN into CTIO and making best use of the IPsec encryption options - for Linux, the open source vpnc software is even better. Check with CISS to make sure the proper VPN client is installed on your laptop.
Remote Hardening Procedures
Ideally, all computers, permanent or mobile which are ever connected to the CTIO network are already configured to be as secure as possible. This means having the latest security patches and locking out un-needed service servers such as WWW (particularly Microsoft IIS!), telnet, FTP, etc. However, the intrinsic security of the average machine in the CTIO network is still not sufficient if that machine were magically (or otherwise) transported outside the safety perimeter of the CTIO computer network.
Traveling devices are worse off even than the public CTIO servers that live outside the firewall in what is known as the "DMZ", which must operate in an security environment similar to those of remote user's computers. These DMZ servers, although exposed to the entire Internet, are hardened more than the typical CTIO computer, and a firewall still protects them from many types of probes and attacks. Moreover, the CTIO network based Intrusion Detection System is scanning ALL incoming and outgoing packets before these even reach the firewall or DMZ. Even home computers may be safer, since cable modems or switches often have a rudimentary firewall, and some Internet Service Providers often run at least some firewall protection.
Applying the following extra steps will go a long way in restoring some of the protection not available to traveling computers while on the road, and need to be applied to home computers as well:
Note that some of the packages refered to below are available from our anonymous FTP server ftp.ctio.noao.edu, under pub/software, but there is no firm guarantee that the latest version is there. For the latest, and for software that is licensed, you will have to ask CISS explicitly.Also note that Linux or MacOS systems are typically immune to many, but by no means all, of the dangers referred to below: only some of the references would apply...
If the laptop runs
a form of UNIX such as Linux or MacOS, the ipchains or better variants such as iptables or ipfilters
firewall package can substitute for a personal IDS. This is the most
important defense on a Linux system and can help keep hackers from breaking in by shutting down services and ports that are not actively used.
Out in the Cruel World
This section is germane mostly to remote users travelling with laptops, but may be of interest to home users as well.
The healthiest concept for the traveling road warrior is - extreme paranoia. Common sense is the most useful defense. Clearly, don't do things like type in a password to your account here at CTIO, or your American Express card number into a random computer screen of some random airport "cybercafe" - the computer could easily be running special key capture software, or even a hardware (!) device like the KeyKatcher or similar which gathers what you type *before* it is ever encrypted. Running one's own laptop using a wireless connection is only a bit safer, but wireless security is shaky at best and whatever you type in *may* be easily sniffed by someone. Hopefully what was in the packet is already encrypted (which is what VPN brings to the security table) or not of value.
The traveling user needs to keep the following risks in mind while on the road:
Conclusion
The guidelines presented here for road warriors are just a mixture of security common sense, practical advice for travelers, and the application of technologies which did not exist at all a few years ago (personal IDS software first appeared in 1999). Many of the points discussed apply equally well to other situations: users working from home, turnkey computers provided as part of a larger system, or even visitors hooking their laptops up to CTIO's network. Each of these situations provides a way for hackers to get into our network.
These guidelines will clearly change over time and the nature of the security threats change along with computer technology. Again, please keep security in mind whenever you use a computer remotely to get back into the CTIO network, and check with CISS so we can make sure your laptop has all of the latest patches, personal IDS, anti-virus and encrypted applications so you can at least rest a little easier when you are on the road...