• SLES (Novell Suse Enterprise)
• Redhat (Enterprise Linux)

I have been using both OSes to hosting Oracle DB and so far it’s been working great. My personal choice would be Redhat, but I know you can get better pricing and features (from the licensing point of view) with Novell… if you know how to deal with these guys.

suse_register -a email=”your@emailaddresshere.com” -a regcode-sles=”REGCODE”

REGCODE is your Suse registration code (e.g. Order code), usually consists of 14 characters.

You must be root to execute this command and make sure you network configuration is set-up and your server on-line. After registration you can start server updating process via Yast or rug command line. Good luck!

Xen Virtual Private Server is perfect for sites that require custom software installation, or even custom daemon configuration that canāt be run on a shared server. Plus, VPS offers better security because you have your own file system that is not shared with other web sites.

However, VPS is a like a small dedicated server – you will probably need skills to manage it. I suggest you read some Linux books and on-line manuals to get familiar with it, otherwise someone may compromise your VPS.

I don’t like much Fedora either because of quick end-of-life (EOF) for updates and upgrading usually costs money, requires downtime and somewhat complex for novice users. So we have rounded down to Centos Linux or FreeBSD. Very good, let’s continue.

FreeBSD is powerful BSD oriented operating system, very reliable, secure (of course if you update it as all other OS) and scalable. Centos is a Redhat Enterprise whitebox clone that is reliable and easy to keep updated due to YUM interface. Much longer EOF comparing to Fedora makes it a much better choice.

While I am more oriented towards Linux – Centos would be my choice of the best OS from the list above, however if you are BSD savvy user – more Unix oriented, go for FreeBSD – it’s powerful, IO-fast, high performance box you could count on.

Type ‘cmd’ and hit enter in Start->Run

To run Traceroute under the Windows you must use traceroute’s command line utility

tracert

traceroute help:

C:\Documents and Settings\Administrator>tracert

Usage: tracert [-d] [-h maximum_hops] [-j host-list] [-w timeout] target_name

Options:
-d Do not resolve addresses to hostnames.
-h maximum_hops Maximum number of hops to search for target.
-j host-list Loose source route along host-list.
-w timeout Wait timeout milliseconds for each reply.

for example to trace the network path to host host.net you should enter

tracert host.net

and hit enter. Voila

find the following setting:

Rebuild Bayes Every = 0

and replace it with

Rebuild Bayes Every = 12000

This will rebuild Bayesian spamassassin database/files every 200 minutes. You can of course aim for a higher value if you have enough disk space.

It’s always good to detect SYN floods and then turn on SYN cookies to better use the CPU power – so the SYN flood must be detected and sysctl SYN cookie value set to 1 in the real time. If you have Syn cookies turned on you server CPU and Network stack will be slightly affected, however, no RAM usage will increase in case of SYN flood. The kernel does not allocate the TCP buffers unless there is a positive reply to ACK/SYN packet.
So if you have a dual core box with 1GB RAM and good enough connectivity, you can leave Syn Cookies enabled without much of the performance loss, even if you get quite decent legitimate traffic.

Sysctl Syn cookie protection for Linux can be activated in real time, by:
sysctl -w net.ipv4.tcp_syncookies=1

You can also increase Syn backlog, to let’s say 1536:
sysctl -w net.ipv4.tcp_max_syn_backlog=1536

For FreeBSD you can tune TCP queue length by issuing:
sysctl -w kern.ipc.somaxconn=1024

We have seen companies offering one front-end and one back-end MySQL server. Usually, they are run by people who never ran the blogs themselves, thus, no experience and skills required for such set-ups. Load balancing is not that easy, assuming you even know what data synchronization and database replication, or master-slave configuration means. Some “experts” will say round-robin is the key for the load balancing, but we know it is not. You need more complex and more advanced load balancing that works and delivers. You need complex solutions that can easily handle “slashdot” and digg effect – withstand heavy and high load hits when your site is getting hammered by real visitors.

At this time, there are not that many companies offering cluster and true load balancing blog hosting. Multiple front-end boxes and back-end database server data must be kept up to date at any time and quickly responding, otherwise this doesn’t simply make a sense.

But not many know how to properly configure and run server platforms, thus, managed dedicated servers are better choice. However, it will cost you much more – there are some management companies that offer service from $30 and up to $500 per month. Starting from standard OS patching and upgrades, and ending with Apache recompilation and adding modules and tunning the platform for better reliability.