My customer wishes to monitor between 800-1200 servers, the guide says a minimum of 10-12 disks.
Question, what is the size of the disks ? How much space should I reserve for this VM ? I know the default minimum is 205G and if it is reaching near capacity, we should always add an additional logical drive.
Question, how many objects can the DB keeps before it reaches 205G of space ? Can the customer keep a period of year data ? Would there be any performance impact if he keeps a year of data ?
The disk recommendations is about the number of spindle to meet the performance criteria, not capacity. Balance is an I/O intensive database application, so this recommendation prevents people from deploying Balance on a single local drive on an ESX host. (Which runs painfully slow). If you deploy on a disk aggregate that span a shelf of disks, you will be OK.
How much space should I reserve for this VM ?
The default VA size of 200 GB should be more than sufficient for most deployments. I recommend deploying the original size, as a VMware thin VMDK, and expand it later if needed. In general, it is better to start with the smaller size, and grow it. If you pick too large a size, you cannot shrink it back down, and you will be stuck with it. You can also Storage VMotion it to a larger datastore, if needed.
Question, how many objects can the DB keeps before it reaches 205G of space ?
For the size you are looking at, monitoring NFS datastores should fit within the 200 GB. If it is monitoring on a FC infrastructure, you will see more objects, and may hit the limit a few months sooner. Balance monitors its own disk space usage, so the admin receives an alert long before running out of disk space.
Can the customer keep a period of year data ?
By default Balance keeps one year of data. Anything older than one year is automatically purged.
Would there be any performance impact if he keeps a year of data ?
No, all of the sizing recommendations are made with a full year's worth of data in the database. A brand new Balance install will certainly run faster, but a fully loaded one will perform acceptably.
One other piece of advice. When deploying in larger environments, it is imperative that the CPU and memory reservations are set correctly, based on the table. Larger environments tend to have overloaded ESX hosts, and if Balance does not have enough resources, it will not finish all the data collections or analysis.