If you're in I.T., you'll be familiar with daft ideas that senior management come up with. I'm still puzzling this one over:
1) Company A decides to reduce exposure to SPOF (Single Point of Failure) by doubling up on servers using clusters etc.
2) Company A decides it's a good idea to reduce costs by placing as many applications on one machine using VMWare, containers and shared database servers.
I'm quite certain these two aims cancel each other out very neatly. It's a good company A doesn't exist. Surely no one would be that daft right?! 
SeasideMan
Pro 
I simplified for comic effect. 
Those ideas are not at all mutually exclusive. The clustering prevents a SPOF, how many applications are on that cluster doesn't affect the SPOF avoidance. What this strategy does do is magnify the effect of a disaster in which multiple servers are lost (e.g. by partial network failure).
One big mistake companies make when clustering is to forget that if there is a failure of one node, the remaining nodes must be powerful enough to run all the applications. So with a 2-node cluster, each node shouldn't be more than about 40% busy on average.
Tom.