With new technology comes great … experimentation.
Marketing is a great tool to confuse people about the abilities and ease of use o certain products. I set out to build my own, home grown cloud environment. One that I can build on and grow as new hardware becomes old, and can be plugged into the Collective.
I started with a base OpenStack setup as provided through OpenSuSE and aborted that mission after I read about the terrible and complex setup procedure fro OpenStack in general.
So I thought that instead of spending a lot of time to setup OpenStack on OpenSuSE I would go with something which is specifically tailored towards OpenStack. Welcome to StackOps
StackOps is based on Ubuntu and is easy enough to install. Download and burn the DVD. Plug it into the DVD drive of your computer and it will install the OS and all required components.
The idea behind plug-n-play is not new and it is amazing when it works. Only this time it did not. So after spending some time on checking out what was going on I decided that it may be better to move on with my life without StackOps.
Also StackOps seems to be tailored towards up-selling its OS and services. Something that I want to avoid at any cost … Vendor Lock In.
So on I went to the next and much easier to handle offering. This time from none other that the Apache foundation. Welcome to CloudStack.
I installed CentOS 6.4 on my two servers and went straight for the CloudStack repos. The installation went just fine and the FE came up in my browser. That was about the point where the fun ended.
I could not get the network configuration right to create a single VM and even after research and playing around for some time I thought that this may not be the best possible option for me.
The first big issue I have with it is the Java based FE, which looks usable but is missing the boat IMHO by looking very much Java like. Now this is very subjective but after some meditation I thought that what I really wanted was
- Learn about Cloud Computing
- Setup my own small cluster
- extend, replace, and assimilate my existing HW.
- Write ZeroMQ based SW for a highly scalabe cloud environment.
- start the collective commonly known as The Borg …
So I went full circle and I came crawling back to OpenStack. Not only do I like to be affiliated with NASA and all the other cool companies involved with OpenStack, but also RackSpace, the main company behind OpenStack seems to have a pretty cool work place … The Castle
I uninstalled CloudStack, burned the repos and installed RDO on my two computers. After about 15 minutes I had the first running single node installation up and running. I could configure and start my very first home grown VM through the user interface and more importantly I could start playing around with the technology behind cloud computing instead of reading through tons of HowTo guides to setup the system in the first place.
Learning is so much more fun if you can experiment with a running system and tweak here and there a bit to see what will happen. If you spend all your time getting the system to run in the first place, you just don’t know where to start troubleshooting if certain things are not working.
Even though you understand the basic, layered approach it becomes hard to know if you have to fix Keystone, Nova, Cinder or any of the other parts of OpenStack if nothing works in the first place.
I am now working to get the second computer, a storage node with approx 2TB of storage into the system. since OpenStack has block storage, and object storage I know now that I should have started out with a different partition layout when installing the OS. A lesson well learned.