See this from the admin guide re Placeholder datastores:
For every virtual machine in a protection group, Site Recovery Manager creates a placeholder virtual machine at the recovery site. You must identify a datastore on the recovery site in which Site Recovery Manager can store the placeholder virtual machines.
After you select the datastore to contain the placeholder virtual machines, Site Recovery Manager reserves a place for protected virtual machines in the inventory on the recovery site. Site Recovery Manager creates a set of virtual machine files on the specified datastore at the recovery site and uses that subset to register the placeholder virtual machine with vCenter Server on the recovery site.
You are correct that they don't need to be very big as the placeholder VM is very small. 5-10GB should be more than sufficient for the placeholder datastores. More than likely, 1 placeholder datastore per site would be sufficient. Make sure you meet the criteria outlined above.
Not sure what you mean by: Do you recommend to use datastores for placeholder which virtual machines stored?
If you are asking if it's ok to store regular VMs on the placeholder datastore, it is ok, though I wouldn't recommend it as it makes things clearer to just have placeholders on the placeholder datastore. If this isn't what you were asking please clarify.
Best practice is, designate a smaller size datastore as please datastore, let's say a 10 GB. More than sufficient with couple of hundred VMs.
when your VMs are replicating from Protected Site to recovery site either using vSphere Replication or Arraybased replication, SRM on top of this is helping you to create Protection Groups and then create Recovery Plans using same set of Protection Groups. as part of this process, at your recovery(destination) site, your placeholder datastore is going to be used to place VM Placeholders [VM's vmx file, vmxf file, vmsd file etc] per VM basis,
since those limited files are participating in creating your Placeholder VM, it doesn't consume too much space, so better to have smaller size datastore as Placeholder Datastore.
Actual VM is anyway being replicated from source and destination datastore. So in case of failure, when you recover, those placeholder VMs will become actual VMs.
Additionally I would also like to mention that if you are doing replication from Site A(Source) to Site B(Destination), ideally you will configure Placeholder Datastore at destination site only but please also configure a Placeholder datastore at your Source site too, so that doing 'Re-protect' and 'Failback' will be working fine.
I have 200 Virtual machine replicated .How many placeholder datastore i should use ?
One datastore per site as Placeholder datastore with minimal possible capacity would be sufficient enough to store those 200 VM placeholders.
Do you recommend to use datastores for placeholder which virtual machines stored?
No, please use a separate datastore as placeholder
Thank you very much for your helps. thats enough for me.
hive a nice days