Managing System Dependencies [closed]

Question

I'm trying to find a tool that will help me map system dependencies. For example I have a web application that runs on two web servers, they both plug into different switches, connect to a SQL cluster which runs on two virtual servers, which are on a 3 node ESX cluster, which depends on the SAN storage, etc, etc

I would like to find a tool that helps me map all these dependencies, let me track who owns each of those pieces, etc. We have around 500 distinct applications and many inter-dependencies.

Answer 1

This is a case for documentation. This sounds like a small installation, so don't over-think this. If you over-document, there's a chance that this documentation won't be updated as the environment changes.

Perhaps a Visio diagram and a wiki entry per host, detailing the dependencies and pertinent information about each system. Some of the open-source asset management tools could also provide a way to assign ownership as well.

Edit:
Managing 500 applications and tracking interdependencies will require something way more automated. Look into cloud management applications or even cluster management and job scheduling software. There are applications that could perform that level of discovery. Without giving too much of a product recommendation, see here and here.

Answer 2

As trite as it might sound: pen, paper, communication and investigation. Those are the tools.

Once you've "mapped out" the dependencies you can put it all into a diagram, document, spreadsheet or some combination of those.

Answer 3

Usually I do this in system documentation on install. So say for example I'm installing a web server, I will create a document for this webserver. The document will explain it's physical hardware and/or virtual hardware, the network interfaces it uses, protocol information (ip info, etc), software reqs (i.e. needs sql or some api installed), etc.

Now this method won't tell you for example, if I unplug X swtich what system will be affected which might be what you're looking for. However, in our managed switches all of the ports are documented what system connects to it with the exception of workstation ports. Now if you wanted to know what systems would be affect by say the sql server going down, I would reference that servers documentation.