Plan of Action

In the following setup and configuration procedures, we have made certain assumptions and policies. These were designed keeping in mind the requirements that we found in typical organizations. To start off, we assumed at least a two-server setup. This is radically different from the kind of setup we have described in previous issues, and is primarily aimed at security and manageability of the whole setup.

The first server referred to as the ‘gateway’ is the server that is connected to the Internet. Its function is simple–connect to the Internet and provide Internet connectivity to users on the LAN via the second server, while at the same time protecting the LAN from attacks and compromises from the Internet. The gateway server need not be a large and powerful box, since it provides very few services. The installation of Linux on this machine will be a ‘stripped’ installation–only software that are actually required will be installed.

The gateway server has no user accounts, only administrative ones. This further increases the server security by removing the possibility of local-user attacks. Depending on how it’s connected to the Internet, this machine may run a pass-through mail service. This is required if the machine is always connected to the Internet with a fixed IP address, in which case mail from outside is delivered to your LAN through this gateway machine.

The gateway machine is connected via a dedicated network link to the second (intranet) server. All services deployed on the network will be hosted here, including DNS, DHCP, user-mail, file and print, and Web-proxy. This machine will have all the LAN’s user accounts on it. All services will authenticate against these accounts, giving users a single user ID and password for all services. All services will require authentication. This is required so that logging of activity can be done in a sensible way. For example, access to the World Wide Web will require the user to configure his/her workstation to connect to the Web proxy running on the Intranet server. The first time (during each session) that the user wishes to access the World Wide Web, he/she will be asked for a user id and password.

Failure to provide it will result in failure to access the Web. Success will allow the user to access the Web, and at the same time, a detailed log will be kept of each user’s activities while surfing. This will be typical of the kind of access control that will be experienced for all services provide by the Intranet server. It is to be noted that at no time will any of the LAN users actually access the gateway server–all services will be availed of via the Intranet server. In the envisioned setup, we assume that users will mostly be using Windows workstations. However, you will also be able to access services using Unix/Linux, Sun Solaris, BeOS, Macintosh and other OSs and architectures.

The setup requires no Linux knowledge by the LAN users. The setup places security before functionality. In case of conflict of interest, we will always be inclined towards the security side.

Finally, we assume that the setup is maintained by a person who either has some knowledge of Linux, or is willing to learn. This person must understand the setup, document changes and additions, and must be a responsible system administrator.

Atul Chitnis is CTO of Exocore Consulting, and Consulting Editor of PCQuest