Anyone use PowerRecon?
Can any one rate PowerRecon Scenario Modeling for gauging how many Hosts you would need for x monitored physical server.
I am not agreeing with allot of the results at first hand, I would appreciate any feedback/input from anyone.
Thanks.
Contact the PlateSpin support guys, they can likely explore whether or not it is something with the way you are running the scenario modeller or with the product.
We have used PowerRecon v2.5.1 (though version 3.0 is out now).
Its a great way of understanding max and average utilisation of your servers. I haven't used the scenario modelling though. How many servers are you intending to monitor?
How do these scenarios sound to you?
12 x HP C685c (4-Way CPU) to host 56 VMs
or
18 x HP C465c (2-Way CPU) to host 56 VMs
Anyone think that this is correct none of the servers are particulaly heavily used.
I'd be greatful for your input...
how much memory do you intend to install in your servers? You'll probably find your ESX hosts will eat memory before CPU.
Definitely a good product. Make sure you follow a couple of golden rules:
1. Analyze each target for at least 30 days; if anything is going to happen to a server it will most likely happen in any 30 day span.
2. Use experience and common sense, do not rely strictly on Recon
I have not used version 3.0 yet, but the previous releases were very nice. I laaake!
Yes using common sense says to me that it is floored where I would expect to get 3/4 VMs per core this is DEV by the way so a 2-way quadcore would easily provide 16 VMs, PowerRecon is saying 3/4 VMs per host.
Contact the PlateSpin support guys, they can likely explore whether or not it is something with the way you are running the scenario modeller or with the product.
We find you have to play with the disk and network i/o modelling options. By default they are crazy.
Also be sure to look at the HyperThreading multiplier options - otherwise you will be modelling HT as 2 x CPU (which is far from the truth).
In summary, find out what is tripping the scenario to add more ESX servers - it's probably network or disk (set them to 80% or higher).
Dave