Добавил:
Опубликованный материал нарушает ваши авторские права? Сообщите нам.
Вуз: Предмет: Файл:

Information protection in digital communication systems. Textbook

.pdf
Скачиваний:
0
Добавлен:
07.09.2026
Размер:
2 Мб
Скачать
301
• Solaris (Level-1);
• Windows 2000 (Level-1);
• CISCO IOS Router (Level-1/Level-2);
• Linux (Level-1);
• HP-UX (Level-1);
• AIX (Level-1);
• Check Point FW-1/VPN-1 (Level-2);
• Apache Web Server (Level-2);
• Windows NT (Level-1);
• Windows 2000 Bastion Host (Level-2);
• Windows 2000 Workstation (Level-2);
• Windows IIS5 Web Server (Level-2).
In this list, the Level-1 specifications define the basic (minimum) level of protection that is required to be provided for most systems with Internet connections. Level 2 specifications define an advanced level of protection required for systems that have high security requirements.
The listed specifications are the result of a generalization of global experience in ensuring information security.
To analyze the configuration of speaker components for compliance with these specifications, specialized testing software (CIS-certified scoring tools) is used.
As an example, consider the specification of the basic level of protection for the MS Windows OS and the corresponding software tools for analyzing the OS configuration.
Network scanners
The main factor determining the security of an AS from security threats is the presence of security vulnerabilities in the AS. Security vulnerabilities can be caused by both errors in the configuration of AS components and other reasons, which include errors and bugs in the software code, lack of security mechanisms, their misuse or inadequacy to existing risks, as well as vulnerabilities caused by human factor. The
302
presence of vulnerabilities in the AS security system ultimately leads to the successful implementation of attacks that exploit these vulnerabilities.
Network scanners are perhaps the most accessible and widely used security analysis tools. The main principle of their operation is to emulate the actions of a potential attacker to carry out network attacks. Searching for vulnerabilities by simulating possible attacks is one of the most effective ways to analyze the security of an AS, which complements the results of pattern-based configuration analysis performed locally using templates (check lists). A scanner is a necessary tool in the arsenal of any administrator or AS security auditor.
Modern scanners are capable of detecting hundreds of vulnerabilities in network resources that provide certain types of network services. Their predecessors are considered to be phone number scanners (war dialers), which were used since the early 80s and have not lost their relevance to this day. The first network scanners were simple Shell scripts that scanned various TCP ports. Today they have evolved into mature software products that implement many different scanning scenarios.
A modern network scanner performs four main tasks:
1) identification of available network resources;
2) identification of available network services;
3) identification of existing vulnerabilities of network services;
4) issuing recommendations to eliminate vulnerabilities.
The functionality of the network scanner does not include issuing recommendations on how to use found vulnerabilities to carry out attacks on network resources. The scanners ability to analyze vulnerabilities is limited by the information that available network services can provide it with.
The principle of operation of the scanner is to simulate the actions of an attacker who analyzes the network using standard network utilities, such as host, showmount, traceout, rusers, finger, ping, etc. In this case, known vulnerabilities of network services, network protocols and OS are used to carry out remote attacks on system resources and documenting successful attempts.
303
Currently, there are a large number of both commercial and freely distributed scanners, both universal and specialized, designed to detect only a certain class of vulnerabilities. Many of them can be found on the Internet. The number of vulnerabilities in modern scanner databases is slowly but surely approaching 1 000.
One of the most advanced commercial products of this class is the NetRecon network scanner from Symantec, whose database contains about 800 vulnerabilities of UNIX, Windows and NetWare systems and is constantly updated via the Web. Consideration of its properties will allow you to get an idea of all the products in this class.
NetRecon network scanner. The NetRecon network scanner is a security administrator tool designed to examine the structure of networks and network services and analyze the security of network environments. NetRecon allows you to search for vulnerabilities in network services, OS, firewalls, routers and other network components. For example, NetRecon allows you to find vulnerabilities in network services such as ftp, telnet, DNS, email, Web server, etc. This checks the versions and configurations of services, their security from network threats and resistance to penetration attempts. To search for vulnerabilities, both standard tools for testing and collecting information about the configuration and functioning of the network, and special tools that implement algorithms that emulate the actions of an attacker to carry out network attacks are used.
The program runs in the Windows NT OS environment and has a convenient graphical interface that allows you to define scanning parameters, monitor the scanning progress, generate and view reports on scanning results. The results are displayed in graphical and tabular form in real time.
The reports generated by NetRecon contain detailed information about the vulnerabilities found, including weak user passwords, exposure of certain services to denial of service threats, OS configurations vulnerable to network attacks, and many others. Along with messages about found vulnerabilities and their descriptions, recommendations for eliminating
304
them are provided. The report on the scan results allows you to outline an action plan to eliminate the identified deficiencies.
NetRecon independently determines the network configuration and allows you to select network resources for scanning. Parallel scanning of all network resources, scanning across a range of network addresses, scanning of individual systems or subnets can be carried out. A scanning session can include all types of scans or individual scans at the user's discretion. The scanning depth is determined by the duration of the scanning session, which is specified by the user. For example, checks associated with guessing user passwords using a dictionary are time-consuming and cannot be completed during a short scanning session.
NetRecon allows the user to track the vulnerability search path, which is the sequence of checks performed by NetRecon that led to the identification of a given vulnerability. The vulnerability search path allows you to trace the actions of a possible intruder carrying out an attack on network resources.
The database used by NetRecon contains descriptions of known vulnerabilities and attack scenarios. It is regularly updated with new data. This database is updated automatically through the Symantec Web site using the LiveUpdate mechanism.
Network scanner NESSUS. The Nessus network scanner can be considered as a worthy alternative to commercial scanners. Nessus is a freely distributed and constantly updated software product. A convenient graphical interface allows you to define the parameters of a scanning session, monitor the scanning progress, create and view reports.
In terms of its functionality, the Nessus security scanner is on par with, and in some respects even surpasses, such well-known commercial scanners as NetRecon from Symantec and Internet Scanner from ISS.
Version 0.99 of the server part of the Nessus scanner was certified by the State Technical Commission of Russia (Certificate No. 361 dated September 18, 2000).
305
Nessus provides very broad capabilities for searching for vulnerabilities in corporate networks and examining the structure of network services. In addition to using standard TCP and UDP port scanning methods, Nessus allows you to search for vulnerabilities in implementations of the ICMP and SNMP network management protocols.
High scanning speed is achieved through the use of a multi-threaded programming architecture when implementing the Nessus scanner, which allows simultaneous parallel scanning of network hosts. To scan each host, the nessusd server creates a separate thread of execution.
When implementing Nessus, a client/server architecture atypical for network scanners was used. Interaction between the client and the server is carried out using a secure client-server protocol, which provides for the use of a reliable authentication scheme and encryption of transmitted data.
All scanning scenarios are divided into groups according to the types of network attacks they implement, the vulnerabilities they detect, as well as the types of network services being tested. So, there are special groups of scenarios:
• Backdoors to detect Trojan programs;
Gain Shell Remotely to implement attacks on obtaining user
privileges on a remote UNIX system;
• Firewalls — for firewall testing;
• FTP — for testing FTP servers;
• Windows — to search for vulnerabilities in Windows systems, etc.
A special group of Denail of Service scanning scenarios are denial of service (DoS) attacks. The only way to be sure that the system being scanned is susceptible to one or another DoS is to perform this attack and see how the system reacts. This group of scripts, however, is potentially dangerous, since their launch can lead to unpredictable consequences for the scanned network, including failures in the operation of servers and workstations, data
loss and “complete paralysis” of the corporate network. Therefore, most
DoS in this group is disabled by default.
306
7.2.5. Architecture of audit systems [14]
For active audit systems, it is advisable to distinguish between local and global architecture.
Within the local architecture, elementary components are implemented, which can then be combined to serve corporate systems.
The main elements of the local architecture and the connections between them are shown in Fig. 7.3. The primary data collection is carried out by agents, also called sensors. Registration information can be extracted from system or application logs (technically it is not difficult to obtain it directly from the OS kernel) or obtained from the network using the appropriate mechanisms of active network equipment or by intercepting packets using a network card set to monitoring mode.
Сетевой сенсор
Системный
сенсор
Прикладной
сенсор
Сенсор сервиса
безопасности
Другие сенсоры
Унификация данных,
фильтрация, сохранение
Хранилище регистрационной
информации
Выявление
злоумышленной
активности
Выявление
аномальной
активности
Возбуждение
тревог,
автоматическое
реагирование,
сохранение данных
о подозрительной
активности
Хранилище данных
о подозрительной
активности
Анализ ситуации,
принятие решений,
контроль за
работой подсистем
безопасности
Сенсоры
Центр распределения
регистрационной
информации
Анализ
регистрационной
информации,
выявление
подозрительной
активности
Реагирование
Администратор
безопасности
Fig. 7.3. Basic elements of local architecture
of active audit systems
307
At the agent (sensor) level, data can be filtered to reduce its volume. This requires some intelligence from the agents, but it relieves the rest of the system components.
Agents transmit information to the distribution center, which reduces it to a unified (standard for a specific active audit system) format, possibly carries out further filtering (reduction), stores it in a database and sends it for analysis to statistical and expert components. One distribution center can serve several sensors.
A meaningful active audit begins with statistical and expert components (for example, because for single-host systems, registration information does not need to be retrieved and transmitted in any special way). Well look at them in detail in the next two sections.
If suspicious activity is detected through statistical or forensic analysis, a report is sent to a decision maker who determines whether the alarm is warranted and decides how to respond.
Usually, when they write about response methods, they list sending a
message to the administrator’s pager, sending an email to him, etc., that is, they mean “manually” taking action after receiving a signal about suspicious
activity. Unfortunately, many modern attacks last seconds or even fractions of a second, so including a human in the response process introduces an unacceptably long delay. Responses must be automated to the greatest extent possible, otherwise audit activity becomes largely meaningless.
Automation is also needed for the simple reason that not all organizations have system administrators with sufficient qualifications to adequately respond to incidents. A good active audit system should be able to clearly explain why it raised an alarm, how serious the situation is, and what the recommended course of action is. If the choice must be left to the individual, then let it come down to a few menu items rather than solving conceptual problems.
Global architecture implies the organization of single- and multi­rank connections between local active audit systems (see Fig. 7.4). At one level of the hierarchy there are components that analyze suspicious activity from different points of view. For example, the host may host user and
308
application behavior analysis subsystems. They can be supplemented by a network activity analysis subsystem. When one component detects something suspicious, it is often advisable to report it to neighbors, either to take action or to increase attention to certain aspects of the systems behavior.
Комплексный
анализ в
рамках сегмента
сети №1
Анализ сетевой активности
Анализ системной активности
Анализ активности приложений
Анализ активности пользователей
Анализ активности
сервисов безопасности
Анализ других
аспектов активности
Комплексный
анализ в
рамках сегмента
сети №2
….…
Системы анализа более высоких уровней
Fig. 7.4. Global Active Audit System Architecture
Multi-rank connections are used to summarize the results of the analysis and obtain a holistic picture of what is happening. Sometimes the local component does not have enough grounds to raise an alarm, but
“collectively” suspicious situations can be combined and jointly analyzed,
after which the suspicion threshold will be exceeded. A holistic picture may make it possible to identify coordinated attacks on different parts of the information system and assess the damage throughout the organization.
Obviously, the formation of a hierarchy of active audit components is also necessary to solve scalability problems, but this aspect is standard for management systems and we will not dwell on it.
Critical architectural decisions include what information to collect and analyze at what scale. The first active audit systems were single-host. Then multi-host configurations appeared. We owe the breakthrough in the field of commercial products to network systems that analyzed exclusively network packets.
309
Today, we can see a convergence of architectures, resulting in complex systems that monitor and analyze both computer and network registration information.
The traditional question is: where to place the sensors of active audit systems?
The equally traditional answer is: “wherever possible.” Only an
analysis of all available sources of information will make it possible to reliably detect attacks and abuses of authority and get to the bottom of their root causes. If we return to the interpretation of the information system as a set of services, then the means of detecting attacks should be located in front of the protected resources (bearing in mind the direction of movement of requests to the services), and the means of detecting abuse of authority should be located on the services themselves. Detection of anomalous activity is useful at all points mentioned. Only with such placement of sensors will the most important principle of the impossibility of bypassing protective equipment be met. In addition, the number of sensors will be minimized, which also turns out to be a problem in conditions of network segmentation and the use of switching technologies.
In order for an active audit system, especially a distributed one, to be practically useful, it is necessary to ensure the integrity of the analyzed and transmitted information, as well as the integrity of the software system itself and its survivability in conditions of failure or compromise of individual components (often an attack is directed first at security means, and only then for application components). It is clear that this is a problem for all distributed systems, and mutual authentication and integrity control services (including data source authentication) are used to solve it.
7.2.6. Requirements for active audit systems
In this paragraph, we will consider the requirements for active audit
systems that are significant from the customers’ point of view. The
requirement of completeness should come first. This is a very comprehensive concept that includes the following aspects:
310
1. Completeness of tracking information flows to services. Active auditing should cover all flows of all services. This means that an active audit system must contain network and system sensors and analyze information at all levels from network to application. Obviously, from the considered aspect of completeness, the requirement of extensibility follows, since no software product can be initially configured for all services.
2. Complete range of detected attacks and abuses of authority. This requirement not only means that the system must have a sufficiently powerful language for describing suspicious activity (both attacks and abuse of authority). This language must be simple so that customers can configure the system in accordance with their security policies. The supplier of an active audit system must transmit signatures of new attacks to the customer as soon as possible (about 24 hours). The system must be able to detect anomalous activity in order to cope with previously unknown modes of disruption.
3. Sufficient performance. An active audit system must cope with peak loads of protected services.
Missing even one network packet can give an attacker a chance for a successful attack. If an active auditing system is known to be underperforming, it may become the target of an availability attack that will lead to other attacks. For local networks, speeds of 100 Mbit/s have become standard. This requires the active audit system to have a very high quality implementation and powerful hardware support. If we consider that the protected services are in constant development, it becomes clear that the performance requirement is also a scalability requirement.
In addition to completeness, active audit systems must satisfy the following requirements:
1. Minimum of false alarms. In absolute terms, no more than one false alarm per hour is permissible (it is better if there are an order of magnitude fewer of them). With intensive data flows between services and their clients, such a requirement turns out to be very stringent. For example, suppose that 1 000 packets pass through the controlled channel per second. In an hour