A History and Survey of Network Firewalls KENNETH INGHAM Kenneth Ingham Consulting and STEPHANIE FORREST University of New Mexico Firewalls are network devices which enforce an organization’s security policy. Since their development, various methods have been used to implement ? rewalls. These methods ? lter network tra? c at one or more of the seven layers of the ISO network model, most commonly at the application, transport, and network, and data-link levels. In addition, researchers have developed some newer methods, such as protocol normalization and distributed ? ewalls, which have not yet been widely adopted. Firewalls involve more than the technology to implement them. Specifying a set of ? ltering rules, known as a policy, is typically complicated and error-prone. High-level languages have been developed to simplify the task of correctly de? ning a ? rewall’s policy. Once a policy has been speci? ed, the ? rewall needs to be tested to determine if it actually implements the policy correctly. Little work exists in the area of ? rewall theory; however, this article summarizes what exists. Because some data must be able to pass in and out of a ? ewall, in order for the protected network to be useful, not all attacks can be stopped by ? rewalls. Some emerging technologies, such as Virtual Private Networks (VPN) and peer-to-peer networking pose new challenges for ? rewalls. Categories and Subject Descriptors: C. 2. 0 [COMPUTER-COMMUNICATION NETWORKS]: General General Terms: security Additional Key Words and Phrases: Firewalls, Network Security The University of New Mexico Computer Science Department Technical Report 2002-37. Author’s addresses: K. Ingham, Kenneth Ingham Consulting, 1601 Rita Dr NE, Albuquerque, NM 87106-1127, ingham@i-pi. om. S. Forrest, Department of Computer Science, University of New Mexico, Albuquerque, NM 87131, forrest@cs. unm. edu. Permission to make digital/hard copy of all or part of this material without fee for personal or classroom use provided that the copies are not made or distributed for pro? t or commercial advantage, the ACM copyright/server notice, the title of the publication, and its date appear, and notice is given that copying is by permission of the ACM, Inc. To copy otherwise, to republish, to post on servers, or to redistribute to lists requires prior speci? permission and/or a fee. c 20YY ACM 0000-0000/20YY/0000-0001 $5. 00 ACM Journal Name, Vol. V, No. N, Month 20YY, Pages 1–42. 2 · K. Ingham and S. Forrest 1. INTRODUCTION The idea of a wall to keep out intruders dates back thousands of years. For example, over two thousand years ago, the Chinese built the Great Wall as protection from neighboring northern tribes. A second example is that of European kings who built castles with high walls and moats to protect themselves and their subjects, both from invading armies and from marauding bands intent on pillaging and looting. The term “? ewall” was in use by Lightoler as early as [1764] to describe walls which separated the parts of a building most likely to have a ? re (e. g. , a kitchen) from the rest of a structure. These physical barriers prevented or slowed a ? re’s spread throughout a building, saving both lives and property. A related use of the term arose in connection with steam trains, as described by Schneier [2000]: Coal-powered trains had a large furnace in the engine room, along with a pile of coal. The engineer would shovel coal into the engine. This process created coal dust, which was highly ? ammable.
Occasionally the coal dust would catch ? re, causing an engine ? re that sometimes spread into the passenger cars. Since dead passengers reduced revenue, train engines were built with iron walls right behind the engine compartment. This stopped ? res from spreading into the passenger cars, but didn’t protect the engineer between the coal pile and the furnace. In this paper we will be concerned with ? rewalls in a more modern setting— computer networks. The predecessors to ? rewalls for network security were the routers used in the late 1980s to separate networks from one another. A network miscon? uration which caused problems on one side of the router was largely isolated from the network on the other side. In a similar vein, so-called “chatty” protocols on one network (which used broadcasts for much of their con? guration) would not a? ect the other network’s bandwidth [Avolio 1999; Schneier 2000]. From these historical examples we can see how the term “? rewall” came to describe a device or collection of devices which separates its occupants from potentially dangerous external environments (e. g. , the Internet). A ? rewall is designed to prevent or slow the spread of dangerous events.
For the purposes of this paper, we de? ne a ? rewall as a machine or collection of machines between two networks, meeting the following criteria: —The ? rewall is at the boundary between the two networks; —All tra? c between the two networks must pass through the ? rewall; —The ? rewall has a mechanism to allow some tra? c to pass while blocking other tra? c. The rules describing what tra? c is allowed enforce the ? rewall’s policy. Additional desirable criteria include: ACM Journal Name, Vol. V, No. N, Month 20YY. A History and Survey of Network Firewalls · 3 Resistance to security compromise; —Auditing and accounting capabilities; —Resource monitoring; —No user accounts or direct user access; —Strong authentication for proxies (e. g. , smart cards rather than simple passwords); —Fail-safety. If it fails, the protected system(s) is(are) still secure because no tra? c is allowed to pass through the ? rewall. Our review of ? rewall technologies proceeds roughly in historical order. Interestingly, the historical development of ? rewalls parallels the descending levels of the protocol stack: ISO level Internet example Application File Transfer Protocol (FTP), Telnet Presentation e. g. Common Object Request Broker Architecture (CORBA) Session no directly corresponding protocol Transport Transport Control Protocol (TCP), User Datagram Protocol (UDP) Network Internet Protocol (IP) Data link Ethernet or Asynchronous Transfer Mode (ATM) Physical twisted pair or ? ber optic cable The protocols used on the Internet for these layers, as well as all other Internet standards are speci? ed by documents known as Requests for Comments (RFCs). RFCs often provide information beyond the bare speci? cations of the standard, and can be useful network administrators, and they appear frequently as citations in this article.
For more information about RFCs, read the RFC frequently-asked question list [The RFC Editor 2001]. Following the order of much of the historical development of ? rewalls, this paper is organized around the levels of the protocol stack, describing approaches which ? lter or are otherwise directed at each level. Section 2 sets the stage by describing the many reasons why organizations and individuals need ? rewalls. The next section, Section 3, describes other surveys of ? rewall technologies. Sections 4-7 descend through the protocol stack, discussing ? rewall technologies at each of the four relevant levels.
Figure 1 provides a graphical view of several of the architectures that we cover in remainder of the paper. Many of the ? rewalls we discuss use or have used proxies as a major part of how they protect networks. A proxy is a program that receives the tra? c destined for another computer. Proxies sometimes require user authentication; they then verify that the user is allowed to connect to ACM Journal Name, Vol. V, No. N, Month 20YY. 4 · K. Ingham and S. Forrest the destination, and then they connect to the destination service on behalf of the user. Section 8 considers ? rewall designs which do not ? cleanly into one of the various network protocol layers, for example, methods addressing multicast services, distributed ? rewalls, and protocol normalization. Firewalls are a large part of the commercial network security market. Many companies market products which ? lter network tra? c one of the ISO levels. Although not the primary focus of this paper, in Section 9 we consider important examples of these commercial products and show how they ? t into the existing literature. Specifying ? rewall policies can be a complex task, and consequently, some higher-level languages have been developed in which to express policies (Section 10).
Once a policy has been speci? ed, the next question is, “How can it be determined that a given ?rewall correctly implements the policy? ” Firewall testing is one way to address this question, Section 11 reviews the literature on this topic. Although little theory has been developed about ? rewalls, Section 12 describes the existing theoretical literature. Very few computer networks can a? ord to be completely isolated from external networks. Thus, an important job of a modern ? rewall is controlling what tra? c is allowed through the ? rewall. Detecting hostile data in this tra? is an important problem confronted by modern ? rewalls, one that we discuss in Section 13. This discussion leads to the related topics of expected future challenges for ? rewalls (Section 14) and some speculations about how ? rewalls have in? uenced recent protocols and how they are likely to change in the future (Section 15). 2. THE NEED FOR FIREWALLS In the early years, the Internet supported a relatively small community of compatible users who valued openness for sharing and collaboration. This view was challenged by the Morris Worm [Spa? ord 1988; Eichin and Rochlis 1989; Denning 1989; Spa? rd 1991]. However, even without the Morris worm, the end of the open, trusting community would have come soon through growth and diversi? cation. Examples of successful or attempted intrusions around the same time include: Cli? ord Stoll’s discovery of German spies tampering with his system [Stoll 1988; 1989], and Bill Cheswick’s “Evening with Berferd” [1992] in which he set up a simple electronic “jail” for an attacker. In this jail, the attacker was unable to a? ect the real system but was left with the impression that he or she had successfully broken in.
Cheswick was able to observe everything the attacker did, learning from these actions, and alerting system administrators of the networks from where the attacks were originating. Such incidents clearly signaled the end of an open and benign Internet. By [1992] Steve Bellovin described a collection of attacks that he had noticed while monitoring the AT ? rewall and the networks around it. The ACM Journal Name, Vol. V, No. N, Month 20YY. A History and Survey of Network Firewalls · 5 gatekeeper Application Proxy Server gate Packet Filtering Router mailgate Email Server A router B router Application or Transport Proxy Server R
Outside Network Inside Network C Packet Filtering Router DMZ Network Packet Filtering Router D Bridging ? rewall Fig. 1. Common ? rewall architectures, referred to throughout this paper. The blue portion of the picture represents the trusted parts of the network, and the red portion indicates machines or networks which are not trusted; these colors came from the original DEC documentation for their ? rewall described in Section 4. 1. A) The DEC SEAL from Section 4. 1. B) A ? rewall using only application- or transport-level proxies, a more common architecture than the DEC SEAL architecture due to the reduced cost of hardware.
The box containing just the “R” is an optional router in this architecture. This architecture with the transport-level proxy server is similar to the AT ? rewall described in Section 5. 1. C) A ? rewall using two packet ? ltering routers (described in Section 6. 1) and illustrating one way of creating a DMZ, sometimes also known as a screened network. A DMZ network consists of a network of machines o? ering services to the Internet. If a machine here is compromised, the inside network remains safe. D) A bridging ? rewall similar to that described in Section 7. ACM Journal Name, Vol. V, No.
N, Month 20YY. 6 · K. Ingham and S. Forrest result was clear—there were many untrustworthy and even malicious users on the Internet. Because not everybody can be trusted, when networks are connected together, a di? erent level of trust often exists on the di? erent sides of the connection. “Trust” in this sense means that an organization believes that the both the software and the users on its computers are not malicious. Firewalls can be used to enforce trust boundaries, which are imposed for a variety of reasons: Security problems in operating systems:. Operating systems have a history of insecure con? urations. For example, Windows 95 and Windows 98 were widely distributed with windows ? le sharing turned on by default; a collection of viruses exploited this vulnerability (e. g. , [Computer Emergency Response Team (CERT) 2000b; 2000c]). A second example is Red Hat Linux versions 6. 2 and 7. 0, which were vulnerable to three remote exploits when installed using default options [Computer Emergency Response Team (CERT) 2001a]. It is an on-going and expensive process to secure every user’s machine, and many organizations make a conscious decision not to secure the machines inside their ? rewall.
If a machine on the inside is ever compromised, the remaining machines are likely as vulnerable [Mu? ett 1995], a situation that has been described by Cheswick as “a sort of crunchy shell around a soft, chewy center” [1990]. Individuals can protect a single machine connected to the Internet by purchasing a personal ? rewall. Rather than trying to secure the underlying operating system, these ? rewalls simply prevent some types of communication. Such ? rewalls are often used in homes and on laptops when they are outside their normal ? rewall. In this case, the trust boundary is the network interface of the machine.
Preventing access to information:. A second example of protecting a network is the use of national ? rewalls, for example, China [McKay 1998]. This ? rewall exists not to protect them from attack, but instead to (attempt to) limit the activities of their users on the Internet. A similar idea is the use of ? ltering mandated in the US by the Children’s Internet Protection Act (CHIPA). This law requires that schools and libraries which receive federal funding put certain ? lters on all web content. Preventing Information Leaks. Because all tra? c leaving a network must pass through the ? ewall, it can be used to reduce information leaks, as in [Ranum 1992]: The key criterion for success for the Digital corporate gateways is preventing an unauthorized or unnoticed leak of data to the outside. Enforcing Policy. Firewalls are one part of an overall security policy; they enforce the policy of network tra? c allowed to enter or leave a network. These policies may limit the applications used, remote machines which may be contacted, and/or the ACM Journal Name, Vol. V, No. N, Month 20YY. A History and Survey of Network Firewalls · 7 bandwidth allowed [Treese and Wolman 1993; Hambridge and Sedayao 1993].
Auditing. If a security breach (which does not include the ? rewall) occurs, audit trails can be used to help determine what happened. Audit trails have also been used to take employment actions against employees using work network resources for non-work purposes. 3. OTHER SURVEYS Firewalls have existed since about 1987, and several surveys and histories have already been written. However, none of them provide both the depth and breadth of this survey, nor do they focus on the peer-reviewed literature describing ? rewall technology. In [1994], Alec Mu? ett wrote a paper which provided an excellent review of the ? ewall policies and architectures of the time. This paper was aimed at people considering implementing a ? rewall, describing the technologies which they might select, their tradeo? s, and how to maintain a ? rewall. One section of the Internet standards document RFC 1636 [Braden et al. 1994] is about the status of ? rewalls as of February, 1994. In this section, they discuss the problem of false security that a ? rewall often provides to an organization behind one. They also review the concepts of application- and transport-level proxies, as well as simple packet ? ltering.
They discuss the problems of authentication and enforcing policy, and provide some solutions from the time. One of the most important parts of the ? rewall section is a discussion of how to design ? rewall-friendly protocols in the future. A review of ? rewalls and their technology appeared in Spectrum [Lodin and Schuba 1998]. This paper is an excellent description of ? rewalls and their technology at the time it was written. However, it has no references to peer-reviewed literature. Several books have been written which describe how to build a ? rewall (e. g. , [Cheswick and Bellovin 1994; Zwicky et al. 2000]).
These books are excellent for people wanting to either evaluate a commercial ? rewall or who are implementing their own ? rewall. However, neither spends much time on ? rewall history, nor do they provide references to peer-reviewed literature. In [1997], John Schimmel wrote a historical review of ? rewall technologies aimed at technical people working in the ? eld of system administration. This review contains good history about early packet ? lters and a brief overview of proxies. Schimmel also mentions limitations of ? rewalls, many of which remain to this day and are discussed in this paper in Section 13.
Unfortunately, this paper has no references to the original sources of the works described. A survey of existing ? rewall technology appeared in Schuba’s Ph. D. dissertation, [Schuba 1997]. In this dissertation, Schuba cites many key papers, but he reviews ACM Journal Name, Vol. V, No. N, Month 20YY. 8 · K. Ingham and S. Forrest the literature as it relates to technology, and does not provide as comprehensive a collection of ? rewall-related references as we do in this paper. Steph: is this too harsh? His work is not bad, and his entire thesis has 148 references. However, his review of ? ewall literature is only 23 references. His review is really weak compared to this paper. Also in [1998], Rik Farrow wrote a ? rewall product analysis which was related to the CSI ? rewall comparison for that year. This analysis is aimed at management and people just arriving at ? rewalls, and provides them with the background information they would need to talk with a ? rewall vendor intelligently. More recent publications include Frederic Avolio’s history of ? rewalls published in the Cisco Internet Protocol Journal [1999]. Avolio is well-quali? ed to write such a document, as he was involved with some of the ? st ? rewalls. His history describes some of the beginnings of ? rewalls, from a technical point of view, and aimed at technical people and technically-oriented management. He provides a short description of ? rewalls which use proxies at the application or transport levels, as well as packet ? ltering and ? rewalls which may be a mix of technologies. Rather than providing details, he refers the reader to Cheswick and Bellovin’s [1994] book on ? rewalls. As a contrast with Avolio’s history, this paper places emphasis on the academic literature and as a result has substantially more references than Avolio’s history.
Habtamu Abie wrote an overview of current ? rewall technology options and emerging trends in [2000]. He discusses the technology, but does not cite the papers by the people who originally developed this technology. Also, Yakomba Yavwa wrote a similar but less in-depth overview [2000]. Like Abie, Yavwa cites none of the original papers where the technology was ? rst described. Finally, several trade and consumer magazines have reviewed ? rewalls; [Markus 2001] is a web site with links to many of them; the Computer Security Institute (CSI) in San Francisco, California has a search engine for comparing di? rent commercial ? rewall products [Computer Security Institute 2001]. Several related topics which we do not address thoroughly in our review include intrusion detection systems [Northcutt and Novak 2001; Axelsson 1998], honeypots [Cheswick 1992; Honeynet Project 2001], IPsec implementations [Doraswamy and Harkins 1999], and commercial products. 4. THE APPLICATION LEVEL The application layer contains programs that interact directly with a user. Many application-level protocols exist, including FTP, the Simple Mail Transport Protocol (SMTP), the HyperText Transport Protocol (HTTP), Telnet, etc.
The ? rst published paper to describe ? ltering network tra? c for access control, by Je? rey ACM Journal Name, Vol. V, No. N, Month 20YY. A History and Survey of Network Firewalls · 9 Mogul in [1989], described a system which worked at the application level. 4. 1 The DEC Securing External Access Link Mogul described a user-level1 solution to deciding whether or not to pass packets though a router running the Unix operating system [Mogul 1989]. The packet ? lter monitored the protocol, source and destination addresses and ports to decide which packets were allowed to continue.
Mogul’s implementation did not keep track of the state of TCP connections, nor the pseudo-state that can be applied to some UDP requests and responses. However, his implementation could have been expanded to handle these; performance limitations of the computers of the day might have prevented this. The technology described by Mogul was applied to Digital Equipment Corporation’s (DEC) ? rewall by Paul Vixie [Ranum 1992] and further extended by Marcus Ranum when it became the Securing External Access Link (SEAL)2 , which was the ? rst commercial ? rewall [Ranum 2001; Avolio 1999], delivered on June 13, 1991 [Avolio 1999].
The DEC SEAL consisted of three components (pictured in Figure 1, part A): Gatekeeper. The application proxy3 server for users who were allowed to access external services. Gatekeeper was also an externally-accessible machine for uses such as anonymous FTP, the Domain Name System (DNS), etc. Gate. A packet ? ltering router limiting what tra? c was allowed to pass between the local and external network. This router was con? gured so that all tra? c to/from the inside went to a proxy on gatekeeper. screend, as described in Mogul’s paper [Mogul 1989], performed the packet ? ltering. Mailgate.
The internal mail server/mail router; this machine was not accessible from the outside. Instead, mail from the outside is delivered to gatekeeper, which passes it to mailgate. External connections are permitted only to gatekeeper, and it may (depending on the con? guration) also be the only way internal users are allowed to access services external to the network. Gatekeeper ran application-level proxies. These proxies also ensured that the proper protocol was used (e. g. , that FTP tra? c is what went to port 20, and not a Telnet session). The proxies on gatekeeper were [Ranum 1992; Treese and Wolman 1993]: 1 As opposed to a kernel-level. ocumentation referred to it as the Screening External Access Link. 3 Proxies were described in detail on page 3. 2 Some ACM Journal Name, Vol. V, No. N, Month 20YY. 10 · K. Ingham and S. Forrest —email and USENET news (since these are store-and-forward protocols, no special handling was needed) —Telnet —FTP —WHOIS (a protocol for looking up network contact information) —sup (a software distribution system developed at Carnegie Mellon University) —the X window system. X is an especially hard protocol to secure, since the client and server are “backwards”: the server is on the desktop and the client is a remote machine.
Treese and Wolman at DEC Cambridge succeeded in developing a proxy which needed no changes to the X client or server code at all [Treese and Wolman 1993]. All access across the ? rewall was through these application proxies. The proxies provided several services [Digital Equipment Corporation 1992]: —They validated that the source and destination machines were allowed to communicate with each other. —They ensured that the user was allowed to access the service and remote machine speci? ed. —They maintained an audit trail—all successful and unsuccessful connections were logged. They enforced that the protocol behaved as expected (e. g. , that telnet was not used to connect to an interactive access terminal server which happened to be running on the FTP port). When a user wanted to send tra? c outside of the local network, the tra? c was passed through gate to the proxy on gatekeeper. If the tra? c was allowed, gatekeeper then sent the tra? c on to the destination machine. No tra? c was allowed to pass directly from the local machine to the remote machine (and correspondingly, no tra? c was allowed to pass directly from the remote machine to the local machine). The DEC ? ewall did not allow any UDP tra? c to pass through it at all. This prohibition was probably for two reasons: —Some of the UDP protocols at the time (e. g. , NTP, DNS) could have a server running on gatekeeper to which the clients connected. —Other UDP-based protocols, such as the Network File System (NFS), would have security risks if they were allowed through the ? rewall. The DEC SEAL is a classic example of a ? rewall which uses application-level proxies. Proxies at this level provide excellent security and auditing. Also, because ACM Journal Name, Vol. V, No. N, Month 20YY. A History and Survey of Network Firewalls 11 they are themselves generating the connection to the remote machine, a proxy has no problems knowing which connections are real and which are being spoofed; this is in contrast to some packet ? ltering ? rewalls (Section 6). One drawback of an application proxy is that it must be designed for the speci? c protocol. New protocols are developed frequently, requiring the development of new proxies; if there is no proxy, there is no access. Often this time lag creates friction with the users, who may be demanding access to the latest applications, an example of the classic tradeo? etween security and user convenience. Application proxies have other drawbacks. In order to use them, the client program must be changed to accommodate the proxy, telling it about the actual packet destination and authenticating directly to the proxy. Because source code is not publicly available for some applications, in many cases these changes can be made only by the application’s vendor, a signi? cant bottleneck. A ? nal drawback of application proxies is performance related, because each packet requires two trips through the complete network protocol stack. None of the papers about the DEC ? ewall discussed performance directly. Over time, the DEC SEAL evolved into the AltaVista ? rewall. The architecture appears to have remained similar, although it was collapsed onto one machine, probably to stay cost-competitive with its competition [Mariet 1997; Smith et al. 1997], resulting in an architecture hybrid of Figure 1, parts B and D. Today, the AltaVista ? rewall appears to no longer exist. 5. THE TRANSPORT LEVEL In the Internet, the transport level consists of only two protocols, TCP and UDP. This small number of protocols makes writing a proxy much easier—one proxy su? es for all protocols which use TCP. Contrast this with the application-level proxies, where a separate proxy is required for each service, e. g. , Telnet, FTP, HTTP, SMTP, etc. Transport-level proxies retain the advantage that a machine outside of the ? rewall cannot send packets through the ? rewall which claim to be a part of an established connection (some of the packet ? lters described in Section 6 have this problem). Because the state of the TCP connection is known by the ? rewall, only packets which are a legitimate part of a communication are allowed inside of the ? rewall. 5. AT&T’s Transport-Level Firewall While DEC was building SEAL, Dave Presotto wrote and Bill Cheswick re-wrote a ? rewall for AT [Cheswick 1990]. AT’s ? rewall also relied on proxies, a term ? rst used in the context of ? rewalls by Bill Cheswick in [Cheswick 1990]. The resulting ? rewall was similar in e? ect to that of Figure 1, part B, although it was ACM Journal Name, Vol. V, No. N, Month 20YY. 12 · K. Ingham and S. Forrest composed of multiple machines. In contrast to the application-level proxies in the DEC SEAL, the AT&T proxies worked at the transport level [Cheswick 2001]. This di? rence arose from the different policies at DEC and AT&T. The DEC policy limited outbound connections, with the goal of preventing unauthorized leaks of information to the outside. AT&T did not limit outbound access other than requiring it to pass through the proxy. As with application-level proxies, users paid a performance penalty because each datagram had to traverse the network protocol stack twice. Cheswick noted that ? le transfers without the proxy ran at 60-90 Kb/sec, while through the proxy they ranged from 17-44 Kb/sec. Like application-level proxies, transport-level proxies require that the client program be modi? d to use the proxy (however, modifying a program to use a transport-level proxy is easier than modifying it to use an application-level proxy; see SOCKS in Section 5. 2). 5. 2 Later Proxies Kolbas and Kolbas wrote SOCKS [Koblas and Koblas 1992; Leech et al. 1996; Leech 1996; McMahon 1996] with the goal of simplifying the use of proxies. A SOCKS call replaced a normal socket call, which resulted in all outbound tra? c using the proxy. This approach was a nice, clean solution, and worked well if one had the source code for the relevant operating system utilities.
Some commercial applications (e. g. , Netscape) also were written to accommodate SOCKS. A system using SOCKS and TCP connections could be transparent to the user (assuming the proxy allowed them access to their destination host). In [2000], Fung and Chang describe an enhancement to SOCKS to allow it to handle UDP streams, such as that used by RealNetworks’ RealPlayer. Marcus Ranum and Frederick Avolio worked on the Trusted Information Systems (TIS) Firewall Toolkit (FWTK), a collection of proxies for building ? rewalls [Ranum and Avolio 1994; Avolio and Ranum 1994].
This freely-available toolkit provided SMTP, the Network News Transport Protocol (NNTP), FTP and Telnet application proxies as well as a generic circuit-level proxy. To improve security, the proxies used the UNIX system call chroot to limit how much of the system became visible; this way if a proxy were compromised, the rest of the ? rewall was more likely to remain trustworthy. The TIS FWTK had no proxies for UDP services; instead, the ? rewall machine ran DNS and the Network Time Protocol (NTP). The inside machines used the ? rewall for those services. When Trusted Information Systems and Network Associates, Inc. NAI) merged in February 1998, the TIS ? rewall became NAI’s Gauntlet Internet Firewall. ACM Journal Name, Vol. V, No. N, Month 20YY. A History and Survey of Network Firewalls · 13 6. THE NETWORK LEVEL If a limited number of protocols at the transport level is an advantage, then it would seem that the single protocol, IP, at the network level would be even better. In some ways, it is. Packet ? ltering is fast, and it does not require the knowledge or cooperation of the users. However, a packet ? lter working only at the network level means that it cannot determine whether a packet belongs to an established communication or not.
And, similarly to transport-level proxies, it cannot enforce which application-level protocol is used. For organizations trying to prevent an information leak, this is a serious limitation. It is relatively straightforward to use allowed protocols to send arbitrary data; for an example, see [daemon9 and Alhambra 1996] which describes communication using the Internet Control Message Protocol (ICMP) echo (“ping”) and ICMP echo reply (“pong”) packets. In spite of these impediments, packet ? ltering at the IP level has been successful when state information about the connection is used (Section 6. 2). 6. 1 Packet Filtering Packet ? tering for network security began with Mogul’s paper about screend [1989]. Whereas application- and transport-level proxies require each datagram to make a trip up and down the whole protocol stack, packet ? ltering can be much faster. Most of the early papers on packet ? ltering for security stressed the performance bene? ts [Corbridge et al. 1991; Chapman 1992; Bailey et al. 1994; Molitor 1995]; later papers continued this trend [Suri and Varghese 1999; Lyu and Lau 2000]. Besides the performance bene? ts, packet ? ltering does not require the cooperation of the users, nor does it require any special action on their part [Weber 1999].
Packet ? lters use one or more of the following pieces of information to make their decision on whether or not to forward the packet [Reed 2002a]: —source address —destination address —options in the network header —transport-level protocol (i. e. , TCP, UDP, ICMP, etc. ) —? ags in the transport header —options in the transport header —source port or equivalent if the protocol has such a construct —destination port or equivalent if the protocol has such a construct —the interface on which the packet was received or will be sent —whether the packet is inbound or outbound ACM Journal Name, Vol.
V, No. N, Month 20YY. 14 · K. Ingham and S. Forrest Although packet ? ltering is fast, it does have drawbacks. The most important limitation is the di? culty of writing correct ? lters. Several researchers have pointed out this problem and attempted to solve it. Mogul in [1991] discusses the di? culty of setting up packet ? ltering rules and provides correct examples as illustrations. In [1992], Chapman makes the point that most packet ? lter languages are similar to assembly language. In [1995], Molitor describes an improved commercial ? lter language. Another drawback of packet ? tering is that it cannot determine which user is causing which network tra? c. It can inspect the IP address of the host where the tra? c originates, but a host is not the same as a user. If an organization with a packet-? ltering ? rewall is trying to limit the services some users can access, it must either implement an additional, separate protocol for authentication or use the IP address of the user’s primary machine as a weak replacement for true user authentication. Also, because IP addresses can be spoofed, using them for authentication can lead to other problems.
If the router is running a properly con? gured ? lter, remote attackers should not be able to spoof local addresses, but they could spoof other remote addresses. Local machines can spoof other local machines easily. In spite of these problems, many organizations are using IP addresses or DNS names for access control. When a proxy is used, the connection to the remote machine comes from the machine running the proxy. With packet ? lters, the local machine directly initiates the connection to the remote machine. A result of this di? rence is that the entire internal network is potentially reachable from external connections; otherwise the reply packets from the remote host would not be delivered properly. As a consequence, hostile remote computers may be able to exploit weaknesses in the protocol implementation of the local machine (e. g. , [Securityfocus. com 1997]). Protocols such as FTP [Postel and Reynolds 1985] also cause problems for packet ? lters. FTP uses a control channel opened from the client to the server for commands. However, when getting a ? le, one method of using FTP (active FTP) has the server open a connection back to the client.
This connection is initiated from the FTP data port (TCP port 20) to a random client port. Since malicious programs could use the FTP data port as their source port, packet ? lter writers cannot assume that a connection from TCP port 20 is a FTP transfer. One solution is to use passive FTP, where the client opens the connection for the data transfer to the server. However, not all FTP clients support passive transfers. RFC 1579 [Bellovin 1994] suggests that FTP client writers include support for the passive method of ? le transfer, but this solution has yet to be universally adopted. ACM Journal Name, Vol.
V, No. N, Month 20YY. A History and Survey of Network Firewalls · 15 6. 2 Packet Filtering with State Originally, packet ? lters did not keep track of the state of a connection. This means that a remote host could send in packets which appeared to be part of an established TCP connection (with the TCP ACK ? ag set), but which in reality were not. This allowed a remote attacker to map the local network as if the ? rewall did not even exist. Attacks against bugs in the TCP/IP protocol stack (e. g. , [Securityfocus. com 1997]) can pass through the ? rewall by claiming to be part of an established TCP session.
The solution to this problem is to keep track of the state of a connection, a property referred to variously as stateful ? rewalls, adaptive ? rewalling and packet inspection. In other words, the packet ? lter looks at both the network level and the transport level data. The router monitors the initial TCP packets with the SYN ? ag set and allows the packets to return through only until the FIN packet is sent and acknowledged. A similar pseudo-state can be kept for most UDP (e. g. , DNS, NTP) and some ICMP communication (e. g. , ping)—a request sent out opens a hole for the reply, but only for a short time.
In [1992], Chapman was one of the ? rst to point out the problem of the lack of state in packet ? ltering ? rewalls. In [1995], Molitor identi? ed the problem of implementing state in his packet ? ltering architecture as a “future direction. ” The ? rst peer-reviewed paper to describe adding state to packet ? lters was by Julkunen and Chow in [1998], where they describe a dynamic packet ? lter for Linux. This paper was proceeded by a technical report in [1997]. Although Julkunen and Chow were the ? rst to publish a peer-reviewed paper about keeping state in packet-? ltering ? ewalls, they were not the ? rst to keep track of connection state. IP Filter (IPF) version 3. 0 in 1996 by Darren Reed predated their work [Reed 2002b; 2002c]. The ? rst peer-reviewed description of this work appeared much later in a discussion of state in IPF [van Rooij 2001]. 6. 3 Improving Packet Filter Speci? cation In [1992], Chapman makes the point that writing packet ? lter rules is similar to writing in assembly language. Some researchers have therefore developed higherlevel languages for specifying packet ? lters. These improved packet ? lter languages are simpler than the policy speci? ations mentioned in Section 10, and they do somewhat ease the writing of ? lter rules. Speci? c examples include: —In [2000], Hazelhurst proposed binary decision diagrams (BDDs) for visualizing router rule sets. Since BDDs represent boolean expressions, they are ideal for representing the block/pass rules which occur in packet ? lters. BDDs also make automated analysis of packet ? lter rules easier, as well as providing better performance than the table lookups used in many routers. ACM Journal Name, Vol. V, No. N, Month 20YY. 16 · K. Ingham and S. Forrest —The ? lter language compiler, ? [Reed 19], allows the use of the C preprocessor, speci? cation of a default block or pass policy for various directions of tra? c ? ow, and provides a simple if-then-else facility. ?c also generates rules for several di? erent packet ? lters (IPF, ipfw, ipfwadm, ip? rewall, Cisco extended access lists, and screend ). 6. 4 Network Address Translation (NAT) Because the Internet is short of IPv4 addresses, many people use Network Address Translation (NAT) to gain more mileage out of a single IP address [Egevang and Francis 1994; Srisuresh and Egevang 2001; Hain 2000].
When a router uses NAT, it changes the source address of outbound packets to its own address (or one from a pool of addresses which it controls). It chooses a local, unused port for the upper layer protocol (TCP or UDP), and stores in a table the association between the new address and port and the real sender’s address and port. When the reply arrives, it looks up the real destination in this table, rewrites the packet, and passes it to the client. When the connection is ? nished (or after a timeout period for UDP packets), the entry is removed from the table.
NAT provides a similar form of protection to that of proxies. In NAT, all connections originate from the router performing the address translation. As a result, someone outside the local network cannot gain access to the protected local machines unless the proper entry exists in the table on the router. The network administrator can manually install such an entry, causing all tra? c destined for a speci? c port to be forwarded to a server for that service (in e? ect, providing an Internet-accessible service on an inside machine. 4 ). RFC 2663 [Srisuresh and Holdrege 1999] notes that NAT is not without its problems.
For example, NAT may prevent IPsec from working correctly. One feature of IPsec is the ability to know that a packet was not modi? ed during transit. However, one of the purposes of NAT is to modify the packet—the source address and possibly the source port must be modi? ed for NAT to work. DNS problems can also occur. A machine behind a router using NAT has a name and an IP address. However, most networks using NAT also use RFC1918 [Rekhter et al. 1996] private IP addresses, which are not globally unique. Therefore, DNS inside the network is not meaningful outside the network. 4 Setting p such an entry is usually a bad idea from a security standpoint. Maintaining a server inside a ? rewall is risky because if it is compromised, the attacker then has access inside the network, which as noted in Section 2 is likely to be insecure. ACM Journal Name, Vol. V, No. N, Month 20YY. A History and Survey of Network Firewalls · 17 7. THE DATA-LINK LAYER A bridge is a network device which works at the ISO data-link layer. Because it operates at this level, it does not need to access to routing information. A bridging ? rewall uses the information listed in Section 6. 1 to decide whether or not to block a packet.
As a result, a bridging ? rewall can look at data in several other levels of the Internet Protocol suite, including the network and transport layers. Because a ? ltering bridge is still a ? lter, the disadvantages of packet ? ltering still apply to it. What makes a bridging ? rewall di? erent from a packet ? ltering router is that it can be placed anywhere because it is transparent at the network level (see Figure 1, part D). It can be used to protect a single machine or a small collection of machines that would normally not warrant a separate network required when using a router.
As it does not need its own IP address, the bridge itself can be immune to any attack which makes use of IP (or any of the protocols on top of IP). And, no con? guration changes are needed in the protected hosts when the ? rewall is installed. Installation times can be minimal (for example, Limoncelli claims three second install times [Limoncelli 1999]), so users are minimally disrupted when the bridge is installed. The ? rst bridging ? rewalls were described by Kahn et al. in [1997] and developed for computers running DOS. They refer to earlier unpublished research concerning packet ? tering ? ltering on a bridge. Later work which also explores the idea of a bridging ? rewall includes Jianbing Liu and Yan Ma in [Liu and Ma 1999]. Keromytis and Wright discuss using IPsec on a bridging ? rewall to set up secure, virtual LANs [2000]. 8. OTHER APPROACHES TO FIREWALLS Some ? rewall designs do not ? t cleanly into a single network protocol layer. Some of them (e. g. Normalization, discussed in Section 8. 2) are ideas that can be implemented in a ? rewall using one of the technologies already discussed. These proposals are just now being incorporated in ? rewall implementations.
Others, such as the distributed ? rewalls discussed in Section 8. 5 are more revolutionary, changing the entire architecture of the ? rewall. Although a few organizations may be using ? rewalls based on these more revolutionary ideas, they are not yet in widespread use. 8. 1 Transparent Proxies As mentioned in Sections 4. 1 and 5. 1, the primary problem with proxies has been that the client software must be modi? ed and/or the user must work di? erently when using the. Transparent proxies address this limitation. With a transparent proxy the client sends packets to the destination in a normal fashion.
However, when the packets reach the ? rewall, access control checks and logging are performed as in ACM Journal Name, Vol. V, No. N, Month 20YY. 18 · K. Ingham and S. Forrest a classical proxy system. The “magic” is implemented by the ? rewall, which notes the destination address and port, opens up a connection to it and then replies to the client, as if the proxy were the remote machine. This relaying can take place at either the transport level or the application level. RFC 1919 [Chatel 1996] compares classical proxies with transparent proxies.
Bonachea and McPeak use transparent proxies to improve the security of FTP in [2001]. Rodriguez et al. describe what they call translucent proxying of TCP in [2001]. Transparent proxies are demanding because the ? rewall must operate both at the network and application levels. Therefore, performance has remained a concern. One solution proposed by Spatscheck et al. [2000] and Maltz and Bhagwat [1998] is that of “splicing. ” In splicing, after the proxy veri? es that communication is allowed to proceed, the ? rewall converts to a network-level packet ? ltering ? rewall for that communication.
Splicing provides the extra control of proxies but maintains performance closer to that of packet ? lters. 8. 2 Normalization Attackers often abuse protocol speci? cations, e. g. , by sending overlapping IP fragments or out-of-order TCP byte sequences. In [2001], Handley et al. stressed that a ? rewall is a good location for enforcing tight interpretation of a protocol. Besides protecting the computers behind the ? rewall from attacks based on protocol abuses, this so-called “normalization” also makes signature-based intrusion detection systems more reliable because they see a consistent data stream.
Handley et al. provide a list of possible normalizations, ranging from those guaranteed to be safe to others that are potentially too strict in their interpretation of the standard. They were not the ? rst to suggest normalization, however. In [2000], Malan et al. describe “transport scrubbing,” and more recently the idea is elaborated in [Watson et al. 2001]. At about the same time, Strother [Strother 2000] proposed a similar idea. Her solution involved di? erent rings of trust, in which a network packet must pass through one ring before proceeding to the next. Many of her rings achieve the same e? ct as normalization. 8. 3 Signature-based Firewalls Malan et al. discuss “application scrubbing” in [2000]. In this approach, a user-level program is established as a transparent proxy (see Section 8. 1) which monitors the data stream for strings known to be hazardous (and presumably to prevent these strings from reaching the client). Watson et al. refer to the same concept as a “? ngerprint scrubber” in [2001]. Snort [Roesch 1999] is a common intrusion detection system. Hogwash [Larsen 2001] is a ? rewall that blocks packets matching the snort rules. It runs on a bridging ACM Journal Name, Vol.
V, No. N, Month 20YY. A History and Survey of Network Firewalls · 19 ?rewall (Section 7) and the authors claim it can handle network speeds of up to 100Mbps on hardware which is not state-of-the-art. 8. 4 Firewalls at Other ISO Network Levels The Common Object Request Broker (CORBA) allows applications written in one language to make requests of objects which may be written in a di? erent language and which may be available on a di? erent machine. The CORBAgate by Dotti and Rees in [1999a; 1999b] is a presentation-level proxy. When a request is made to an object which is on the other side of the ? ewall, the proxy transparently changes the references. The result is that objects on either side of the ? rewall end up referring to an object on the ? rewall. 8. 5 Distributed Firewalls There are several limitations to the ? rewall technology that we have presented so far. One common assumptions is that all the hosts inside a ? rewall are trustworthy. This assumption is not always valid—for example, see the discussion about the problems with virtual private networks (VPNs) in Section 14. 1. A related problem is that ? rewalls ignore internal tra? c, which may violate security policy. Because ? ewalls are typically centralized in one location, they can become performance bottlenecks as well as providing a single point of failure. A further limitation of conventional ? rewalls arises because in some cases the local machines know the context of the communication not available to the ? rewall. For example, a ? le transfer may be allowed or denied based on what ? le is being transferred and by whom. The ? rewall does not have this local, contextual knowledge. One solution to these problems, proposed by Bellovin [1999], is a distributed ? rewall. This was implemented by Ioannidis et al. in [2000] and by Markham and Payne in [2001].
In this ? rewall, the network administrator has a single policy speci? cation, loaded onto all machines. Each host runs its own local ? rewall implementing the policy. Machines are identi? ed by cryptographic certi? cates, a stronger form of authentication than IP addresses. With a distributed ? rewall, the common concept of a DMZ or screened network, where servers accessible to the outside world are located, is no longer necessary (for an example of a DMZ or screened network, see Figure 1, part C). Hwang and Gangadharan [Hwang and Gangadharan 2001; Gangadharan and Hwang 2001] propose using ? ewalls on all devices attached to the network, where they can be combined with an intrusion detection system (IDS). When the IDS detects an anomalous event, it modi? es the ? rewall to react to the threat. Lower overhead can be achieved with this approach than that reported for the distributed ? rewall developed by Ioannidis [2000]. However, the architecture still uses a conACM Journal Name, Vol. V, No. N, Month 20YY. 20 · K. Ingham and S. Forrest ventional ? rewall. Distributed ? rewalls have a di? erent set of problems than centralized ones. The most signi? cant is that a distributed ? ewall relies on its users (with physical access to the machine) not to override or replace the policy. In other words, the network administrator must trust his or her users. Additionally, if the ? rewall is running as a part of the operating system, then the operating system must protect the ? rewall software. However, the local ? rewall is protecting the operating system, creating a circular set of dependencies. In [2001], Markham and Payne propose implementing the distributed ? rewall on a programmable network interface card (NIC) to reduce reliance on the operating system for protection.
Around the same time that Bellovin proposed the distributed ? rewall, Ganger and Nagle at Carnegie Mellon University also proposed a distributed approach to security in [2000], in which each device was responsible for its part of the security policy. Ganger and Nagle argue that if each device were more secure, then an attacker who succeeds in passing the outer defenses (the ? rewall) would not ? nd vulnerable targets inside. They propose installing security devices on many parts of a network, such as NICs, storage devices, display devices, and networking hardware such as routers and switches.
The idea is that the devices would dynamically adjust their approach to security based on the overall network defense level. As with Bellovin’s proposal, programmable NICs are an important part of the overall strategy. 8. 6 Protection against denial of service attacks In a denial of service attack, the attacker’s goal is to reduce or eliminate an authorized user’s access to a service, machine or network by disabling the service or machine, or by overloading some aspect of the system with voluminous but legal activities. Because all tra? c for the network travels through the ? ewall, some denial-of-service attacks can be stopped at the ? rewall. For example, protocolbased attacks (e. g. , [Computer Emergency Response Team (CERT) 1997b] and [Computer Emergency Response Team (CERT) 1998]), which cause the remote machine to crash, can be stopped by protocol normalization (Section 8. 2). When attackers are attempting to saturate available bandwidth, Ioannidis and Bellovin proposed the idea of pushback—the ability for a router to detect and preferentially drop packets which are part of an attack, and also to notify upstream routers of the attack [Ioannidis and Bellovin 2002].
In [2000], Balasubramanian described an architecture in which network- and hostbased intrusion detection systems are combined with a policy-driven coordinator to respond to denial of service attacks. This response is a job that traditionally would be given to the ? rewall. As presented, the response could be anything a process ACM Journal Name, Vol. V, No. N, Month 20YY. A History and Survey of Network Firewalls · 21 can do; it is up to the network administrator to determine the proper response to the attacks.
Balasubramanian recognizes but does not address the problem of false positives or spoofed network packets being used to cause a reaction desired by the attacker. In [1997], Schuba et al. review various approaches to SYN-? ood denial-of-service attacks and present a new approach (not limited only to ? rewalls). They suggest using a program to monitor all tra? c on the LAN. They describe a program which categorizes SYN packets into four classes: never seen, belonging to correctly behaving hosts, potentially spoofed, and certainly spoofed. In addition to the program’s classi? ation, the administrator can provide a collection of addresses known to be good or bad. When the program sees SYN packets from bad hosts (dynamically characterized by their behavior), it sends a RST packet to kill the half-open connection. 8. 7 Multicast On the Internet, multicast is often used for various forms of multimedia. In contrast with traditional unicast communication, in multicast the sender does not necessarily know the identities of all the participants in the session. And, this is also true for the recipients, who do not know in advance all the possible people who might be sending to them.
This di? erence also makes proxies such as SOCKS di? cult to implement unless they change the multicast into a collection of unicasts, a change that defeats the bene? ts of multicast—with multicast, once one client inside of the ? rewall has joined a group, others may join without needing to authenticate. Additionally, the multicast routing protocol, the Internet Group Management Protocol (IGMP), speci? es only multicast groups and not UDP ports; in a default con? guration, a multicast source has access to the complete set of UDP ports on client machines.
If a source has access to all UDP ports, then it could potentially attack other services, (e. g. Microsoft networking) which are unrelated to the service it is providing. The classic paper on multicast and ? rewalls was published by Djahandari and Sterne in [1997]. In this paper they describe an application proxy for the TIS Firewall Toolkit. The proxy has the following features: —It allows authentication and auditing; —It prevents multicast tra? c from reaching hosts which did not request it; —It allows the multicast tra? c to be sent only to “safe” ports. The proxy converts multicast tra? into unicast tra? c. Unfortunately, this approach also means that it does not scale well, as a collection of N users all receiving the same multicast stream increases the tra? c inside the ? rewall by a factor of N over what it would have been if multicast had been retained. On the other hand, ACM Journal Name, Vol. V, No. N, Month 20YY. 22 · K. Ingham and S. Forrest they do solve all of the security problems mentioned in the previous paragraph and later in this section. RFC 2588 [Finlayson 1999] suggests several possible solutions to the problem of multicast and ? rewalls.
For example, communication between external and internal machines could be tunneled through the ? rewall using the UDP Multicast Tunneling Protocol (UMTP). This protocol was designed to connect clients to the Multicast Backbone (the MBone), but would work for tunneling through multicast-unaware ? rewalls. RFC 2588 also mentions the possibility of dynamic ? rewall rules, and in [1999], Oria describes in further detail how they can be implemented. A program runs on the router, which monitors multicast session announcements. The program reads the announcements, and if the speci? d group and UDP port are allowed by the policy, it generates the necessary rules permitting the data to pass through the ? rewall. When a client informs the router that it wishes to join a multicast group, it sends an IGMP join message to the router. The dynamically generated rules permit or deny this access. This approach to multicast on the ? rewall assumes that session announcements can be trusted. Unfortunately, this is not a valid assumption because they can be spoofed. 8. 8 Transient Addressing Many protocols, such as FTP, RealAudio, and H. 23 (a protocol used for programs such as Microsoft’s NetMeeting), open secondary channels for additional communication. These additional channels are a problem for ? rewalls unless the ? rewall makes use of a stateful proxy. In [2001] Gleitz and Bellovin propose a solution to this problem by taking advantage of version 6 of the Internet Protocol (IPv6), which has 128 bits of address space. This is large enough for each host to have multiple addresses. A client initiating a connection to a FTP server uses an address which includes the process group ID of the FTP client process.
The ? rewall sees a connection from a speci? c IPv6 address going to a FTP server at a remote site, and then allows all communication from the server back to the client’s address. On the client side, this address is only used for the FTP process; connections from the FTP server to other ports on the client will not be accepted, because only the FTP client is using that speci? c address. 9. COMMERCIAL FIREWALL PRODUCTS It is di? cult to separate completely advances in ? rewall technology from the commercial products that implement them. There is a large market for commercial ? ewall products, which has driven many important recent developments. At the same time, without direct inspection of the source code, it can be quite di? cult ACM Journal Name, Vol. V, No. N, Month 20YY. A History and Survey of Network Firewalls · 23 to ascertain exactly how or why a particular product operates. Further, the success or failure of commercial products depends as much on business issues as on technical merit. In spite of these problems, no review of ? rewall technology would be complete without some discussion of commercial products.
In the following, we highlight a few of the many products currently on the market, emphasizing those that we believe are interesting intellectually, have had unusual impact on the development of ? rewalls, or are particularly clear examples of a particular methodology. 9. 1 Network Associates, Inc. (NAI) As mentioned in Section 5. 2, the TIS ? rewall toolkit was an early ? rewall based on proxies. After the merger with NAI, it became the basis for the Gauntlet ? rewall. This ? rewall continued to evolve and now incorporates the following features [Network Associates Technology, Inc. 2001b; 2001c; 2001d]: —Packet ? tering with state; —Transport-level proxies; —Transparent application-level proxies for common protocols such as RealAudio, QuickTime, HTTP, SMTP, and FTP; —Groups to ease the packet ? lter speci? cation; —Splicing like was described in Section 8. 1 to improve performance; —Content scanning for known viruses and hostile applications. 9. 2 Cisco Cisco has two ? rewall products (IOS ? rewall and Private Internet Exchange (PIX)) which share many features, but also have some di? erences. Both are based on packet ? ltering with state. They also perform what is known as “inspection,” in which the packet ? ter inspects the data inside the application layer and modi? es the packet ? ltering rules to adapt to what the application is expected to do. For example, the inspection algorithm could determine that a FTP client will need a port opened, changing the ? ltering rules appropriately. The result of such packet inspection is similar to that provided by a transparent application proxy, but with less overhead. It also su? ers from the same problems—as protocols evolve, the inspection code must also evolve in order to take advantage of the new features in the protocol. Both ? ewalls can require authentication against Terminal Access Controller Access Control System (TACACS)+ or Remote Authentication Dial-In User Service (RADIUS) before passing packets from or to speci? ed hosts and/or services. The IOS ? rewall can also use Kerberos for authentication [Cisco Systems, Inc. 2001]. ACM Journal Name, Vol. V, No. N, Month 20YY. 24 · K. Ingham and S. Forrest The PIX is especially notable because it was the ? rst commercially available implementation of NAT [Cisco Systems, Inc. 2000]. It also regenerates TCP sequence numbers, using strong random numbers.
This is important because machines which have weak TCP sequence number generation algorithms are vulnerable to session hijacking and spoofed one-way connections. The IOS ? rewall protects against pre-identi? ed malicious Java applets, presumably by matching against a static signature ? le. The IOS ? rewall defends and protects against SYN ? oods and other closely related TCP attacks. As explained in [Cisco Systems, Inc. 2001], SYN-? oods are detected by comparing the rate of requests for new connections and the number of half-open connections to a con? gurable threshold level to detect SYN ? ooding.
When the router detects unusually high rates of new connections, it issues an alert message and takes a precon? gured action The precon? gured action either instructs the protected host to drop half-open connections (presumably by sending it a TCP reset packet), or temporarily ? lters all TCP SYN packets destined for the protected host. Unfortunately, this action disables all inbound connections to the protected host. The IOS ? rewall also monitors TCP sequence numbers and drops all packets with suspicious numbers. 9. 3 Lucent’s ?rewall family Lucent’s ? rewall family [Lucent Technologies 2001] is based on packet ? ltering with state.
Similar to Cisco, packets are inspected, and the packet ? lter examines at the higher-level data (including some protocols at the application layer). Unlike the Cisco products, the ? rewall can be a bridging ? rewall. 9. 4 Sun’s Sunscreen Secure Net Sun’s Sunscreen Secure Net 3. 1 [Sun Microsystems 2000] is based on packet ? ltering with state and a collection of application proxies. Its HTTP proxy can be con? gured to pass or drop Java applets. It can also ? lter: —Java applets based on the cryptographic signature included in the applet, —Java applets based on a hash of the applet, —Cookie requests, and —ActiveX content.
The FTP proxy is a true application proxy and therefore can limit access to certain ? le transfer commands such as put or get based on source and destination addresses and/or user authentication. Sunscreen Secure Net can also be con? gured as a bridging ? rewall. ACM Journal Name, Vol. V, No. N, Month 20YY. A History and Survey of Network Firewalls · 25 10. FIREWALL POLICY SPECIFICATION Firewalls were originally built and con? gured by experts. Now, ? rewalls are commodity products which are sold with the intent that nearly anyone can be responsible for their network’s security.
The network administrator uses a graphical user interface (GUI) to con? gure packet ? ltering rules. Unfortunately, this GUI still requires the user to understand the complexities of packet ? lters. These complexities were originally pointed out in [1992] by Chapman. In many cases, the only di? erence between then and now is that the user interacts with the packet ? lter rules through a GUI. The common use of transparent proxies only increases the complexity of the administrator’s task. Some researchers have tried to improve the way that ? rewalls are managed.