Hi to all!

Premise: I should start working at the final course project for my undergraduate degree course in Computer Science
(I know... this is usually one of the most hated kinds of topic, I'm sorry and I beg your pardon, but, please keep reading... please)

My field of interested is almost everything concerning Linux/Unix and system administration in general.
Firewall rules, system/server administration/configuration, hardening a system and so on.
In the future I'm looking forward to a job as a system administrator (with time, efforts, and job experience obvioiusly!)

Anyway, I'm not very interested in coding and programming stuff (and I'm not so good at it, either).

So a project in the Linux/Unix/System Administration field will be useful for my future (I hope).
A project involving coding, programming will not.
Scripting and bash scripting is ok, though (the same for everything about CLI, editing configuration files.....).


My ideas: hardening a Linux system; Snort; something about VPN; Linux Cluster; configuring some specific server/service
IMO they are useful topics (you can learn a lot working on these) but I feel they are too general and maybe a bit outdated


What I'm NOT looking for:
- I'm NOT looking for anyone else to do my "homeworks" (or job) for me

What I'm looking for:
- some advice to "tune up", to narrow, and to update maybe, my ideas
- other ideas, suggestions (maybe related to more "up to date" topics)
- some help to avoid a silly/too easy or, on the other hand, impossible topic


Any serious help will be really really much appreciated. Thanks in advance!

Good to hear that you want to build your career in linux.

Back to the topic. Is this project will be on paper or you have been asked to create a setup to show them. I am asking this because on piece of paper it might not look that attractive but in real environment I hope it will.

Important topics/services that I can think of:

1. Email service. very important for today's environment.
2. Clustering for fault tolerance.
3. LDAP. Nobody wants to keep syncing /etc/passwd on all servers :-)
4. DNS/DHCP services
5. FTP, HTTP services, IPTABLES, selinux.
6. Backups. Can be demostrated using rsync.
7. Scripting. A simple script to perform run rsync everynight for backups will be good I guess.
8. Logs generated by rsync to be sent to adminstrator to analyze if there were any issues.

As of now I can only thing of the above points and in the environment I would prefer to go with the following configs:

1. Two LDAP servers (primary and secondary)
2. Clustered HTTP services. To maintaine uptime for my main website.
3. DNS/DHCP to be run on primary LDAP server. I would prefer to go with DDNS instead DNS.
4. One backup server. All the server will initiate rsync connection either over ssh or by local mounting destination server mount points for backup.
5. One seperate server for email. Postfix will be a good option for it.
6. IPTABLES and selinux tightly configured on public facing system as well as on all other systems.

I hope this helps for the start.

What I found about these more OS level things is that it's hard to find a suitable "bigger picture" to align things to. You can't talk initially about service tuning and config files... you need to have a good reason to want to do it. These things need to be the means, not the end. Maybe some sort of clever self configuring parachuted in security device that does *something* interesting? Maybe something about cluster nodes? automatically being able to drop in nodes to expand a computing cluster from bare metal etc.

What I thought about coding for my equivalent work was a concurrent logic design tool, where you could use drag and drop to build up logic models using atomic blocks like "double" or "add 1" to create sequences, but you don't want to code. I didn't really either so I build a oscilloscope, 50% hardware, 50% software. Worked ace.

First of all, thanks for your replies.

I have to "develop" something, usually to configure something or to code something (the latest I would like to avoid, for the reasons I explained early)

The time frame is about 4 months.

So, if the topic consists in setting up (configuring, installing, etc.) something, actually I have to set up a real working environment (and this is good, IMO)

Obviously, talking about a final course project for an undergraduate degree, the topic should NOT be too easy or too general, either
(a too easy topic is "unsuited", a too general topic is impossible to fully develop in a good way and a reasonable time by only one person)

Something specific, a useful topic is the best choice. Something about a solution for a possible real problem, or a possible real situation.

I think a great project would be to go beyond the work done by the NSA to secure Linux. Everyone who knows Linux in any sort of depth knows about the work they've done with SELinux. But they've gone --much-- further. I'd take the tips that they've created, ref http://www.nsa.gov/ia/guidance/secur..._systems.shtml , and I'd find enhancements to what they've done, to create an even more secure release, to document improved "best practices."

The current NSA documentation is for RHEL 5. Perhaps you could set up new documentation for RHEL 6, perhaps you could show how the NSA guidelines would work for the next Ubuntu Server Edition.

I appreciate this idea.... maybe hardening the next Ubuntu Server edition according with (and beyond, maybe) the NSA specifics.
Ubuntu is free and perhaps the most used Linux "flavour" nowadays.... What do you think about this topic?

In your opinion (mike rhce and everyone else), would this topic be suitable for an Undergraduated Degree in Computer Science?
(I mean.... could it be perceived as too easy by the professors? or, on the other hand, be impossible to fulfill for one person only?)

Depending on your experience with various services noted by T3RM1NVT0R, I'd ballpark the job at between 200 and 400 hours. Since SELinux is not normally enabled for Ubuntu, I'm guessing that would be the biggest challenge (though I haven't checked up on Ubuntu since 10.04). Last I checked, the minimal install for Ubuntu Server edition was already pretty good in many ways.

If it takes "too long," you could always limit the number of services to configure.

I'd visualize the "deliverable" as a CD that you would hand in -- from which a sysadm could install new workstations automatically. In general, I think best security practices involve VMs, with one service (plus SSH) configured per system.

Of course, all this depends on your advisor / sponsoring faculty.

(p.s. this is a potentially interesting topic for a book. I'd appreciate it if you keep me posted. FYI, I've enabled emails in my profile.)

sorry if I reply only now, I'm looking for a professor to talk about these ideas (because I need their approval, anyway)
I will update you ASAP (maybe using your email too, mike rhce)

thanks again to all (for every other ideas or whatsoever... don't be shy, please post them)