Posts Tagged ‘Ubuntu’

Be Unlimited and Name Servers

Monday, December 1st, 2008

I’ve been a Be-Unlimited ADSL user for a while now, and of late, I’ve noticed that their DNS servers have been very unreliable. They did send an email to their users stating that they had some issues (I’ve deleted that mail! doh!) a few weeks back, and it was mentioned that the problem was resolved. However, I’ve been keeping a close eye on it, and it seems it hasn’t been resolved entirely – roughly 1 out of every 5 queries to their name servers (that they provide via DHCP) times out.So, being really frustrated with hitting refresh so many times when browsing the web, I decided to just take the bullet and install my own BIND server locally. I’m not a big fan of running my own name server – its a pain as it eats up resource, and I really shouldn’t have to worry about resolving names from my home laptop – that’s the responsibility of the darn ISP! To get it working on Ubuntu is as simple as:

So, being really frustrated with hitting refresh so many times when browsing the web, I decided to just take the bullet and install my own BIND server locally. I’m not a big fan of running my own name server – its a pain as it eats up resource, and I really shouldn’t have to worry about resolving names from my home laptop – that’s the responsibility of the darn ISP! To get it working on Ubuntu is as simple as:

sudo apt-get install bind9

If you’re using DHCP, edit /etc/dhcp3/dhclient.conf. And ensure that there’s an entry for prepend domain-name-servers:

prepend domain-name-servers 127.0.0.1;

And thats it!

If you don’t want to run your own bind server, then you’re not out of luck just yet! You could OpenDNS and plugin their servers locally or on your router. I’ve set them up for my router as other PCs and devices use the router to resolve names.

Ubuntu As A Server

Saturday, October 11th, 2008

Ubuntu Wikipedia

Wikipedia, who use a mix of Red Hat and Fedora distributions for their infrastructure are moving over to Ubuntu, as part of an effort to standardise their infrastructure.

Ubuntu is very extremely popular as a desktop Linux distribution, but seem to have taken a while to penetrate the server market. If you have a look at Distrowatch.com, Ubuntu has had the top position in their Page Hit Rank since 2005, which just generally means that it is pretty darn popular!

My general take on Ubuntu is that it has always been promoted as a desktop distribution. I remember having a discussion with a former colleague about using Ubuntu as a server. He laughed out loud in disbelieve and discounted the fact at face value. It will take time to adjust to the fact that Ubuntu is actually a very viable server distribution as there’s a stigma attached to it – imagine a system administrator coming up to you with a solution to a business proposal and suggesting “Oh yes sir, we’ll be using the Windows XP”. To the average Joe, that’ll be a textbook failure in making first impressions. Ubuntu’s image is that deep in the desktop zone.

However, not many people know this fact – the current server that hosts this blog runs on Ubuntu, and as far as I can remember, Ubuntu has always had a “minimal”/”server” installation option along with their standard “desktop” option. In the not so far past, I ran a bunch of servers from home behind an ADSL line (hardly the enterprise environment!) and as soon as Ubuntu came out, I wiped out the mix of Debian/Gentoo/FreeBSD that I was running and installed Ubuntu everywhere – desktops and servers!

Why would I do such a thing ? Why would I use a desktop distribution on a server ? Why would Wikipedia use Ubuntu on their enterprise environment which process 50,000 HTTP requests during peak times ?

Well – the answer is actually very simple. Canonical took Debian, gave it a major cleanup, ease of use and (importantly) standardised the whole distribution. They basically wanted something that was easy to use and maintain. There would be a few releases a year and they will be properly supported, but only for a certain period of time. The best part is that the distribution feels like it was not standardised from a desktop point of view, i.e. a top-down approach. It really feels like they took a bottom-up approach in modifying and cleaning up Debian – which results in a great distribution that has everything consistent! It takes more work to implement the bottom-up approach, as you’re not only cleaning up Gnome or KDE, but you’re cleaning up the thousands, upon thousands of standard software packages that come with Ubuntu.

So, having Ubuntu basically means that I get ultra awesome support for security patches and bug fixes, and at the same time, I get a Debian distribution that is extremely clean, standardised and consistent. All the default config files are at the right locations. It makes it faster, easier and more intuitive to setup a package under Ubuntu as the default options for most packages are good enough.

I can see why it makes sense for Wikipedia to switch to Ubuntu – it keeps the servers very standardised, which in turn means administration costs are kept low and architecture is less prone to failure. My only hope is to actually see more adoption of Ubuntu in the data centres worldwide. So, if you’re in a team that uses Linux, or even Windows, do convince them to switch over to Ubuntu.