Cloud, but there is nothing stopping me from packaging it up and having it run on premises.
OS vs Application is what I am asking here about. I could focus on OS/system monitoring or specific applications. If I go for generic application monitoring then I am competing with the big guys and that is what I am trying to avoid.
As for pricing - looks like it would be niche dependent. I was thinking of nothing less than $99/month and then up since I probably could not support many small customers on a lower price point.
So it's hard do charge anything or much for OS/Systems because there are so many free tools (it doesn't take long at all to deploy ELK + metricbeat for example). I would concentrate on applications, but then you are competing with DD and Librato anyways, unless you can cover applications that they don't monitor because they are too complex (I can't think of any though). I think the questions more than what niche to go after is what added value can you provide.
OS vs Application is what I am asking here about. I could focus on OS/system monitoring or specific applications. If I go for generic application monitoring then I am competing with the big guys and that is what I am trying to avoid.