NFS will not give you resilience against machine failure either: the NFS server is still a single-point-of-failure.

If you want to run more than one instance of SLAPD with the same data, you should set up replication as described in the Admin Guide. This avoids using NFS to store the databases and also allows you to build a system that is resilient to machine failures.

Always store SLAPD databases on local disks (and choose the fastest ones available if performance is important to you).