Setting up a shared NFS environment

NFS settings are determined by which of the following scenarios are most important:

• High availability (HA) in a failover scenario
• Better throughput configuration where HA and data durability are sacrificed

Setting up a shared NFS environment includes the following considerations:

• Redundancy of the NFS server. If you are using a Linux NFS server, you could use DRBD (Distributed Replicated Block Device), which has an active-passive kind of replication option.
  Attention: DRBD is open source software like Linux.
• When setting up the NFS server exports and the corresponding NFS buckets, you should avoid options like file delegation that bring a risk of file loss if the NFS server machine or client machine stops running.
• When running B2B Advanced Communications on multiple machines, make sure that any NFS exported directory is configured on all of the machines under the same path.

  For example, you cannot mount server1:/exported_dir on one machine as /data1 and on another machine as /data2. Every machine must use the same absolute path to access the shared directory.

After you set up NFS, if a single NFS server cannot keep up with B2B Advanced Communications traffic, and a separate NFS server is available, you could create a new bucket variant that is pointed to the second NFS server. B2B Advanced Communications can then spread its writing of blobs across the two NFS servers. You can have one server for each variant of a bucket, with a maximum of 64 variants for a single bucket.