One of the good parts of my job consists in helping customers benefit from implementing virtualization management solutions. In most cases they start looking for solutions only after they experience problems. Slow VMs, slow applications, you know….
Then what? Fire in the house. The first person to call is the “guy who manages” the whole system – my client. Complains of slow applications or network performance. It’s of course already too late to fix the problem. Especially if there is no virtualization management suite which would allow to not only monitor, but also predict with “what if scenarios”.
Wouldn't it be better to fix the problems before they occurs? Because the slow VMs performance do have source somewhere and it would be nice if you could be alerted that, for example, “In one month time you’ll run out of space in your datastore”, or “In 2 months the capacity of your 2 brand new ESXi hosts will not be sufficient” because you’re creating 5 new VMs per day…
Usually the admins have to firefight two things (but usually much more):
One of the first functions of every virtualization management suite is the capability to get the « full picture », to see, for example, that half of the client’s VMs are performing poorly because they’re over (or under) provisioned with CPU or memory, that each second VM is running with snapshots.
Other problems can be arising from bad performance of existing hardware, which is often out of warranty and too old to do the job that is required.
Now what? Buy first and optimize later? No, that’s not what I would usually advise. The optimization phase would consist in helping the client solve their problem first and then give them advice to implement a solid virtualization management solution. The client thanks you not only for your help, which saved his bacon, but also for the advice you gave him to save him from future problems.