There appears to be some confusion here; the older Isilon Hadoop docs recommended separate subnet pools for the NameNode (NN) and the DataNodes (DN), this changed recently and now the suggested recommendation is a single subnet pool for NameNode (NN) and DataNodes (DN), the pool should be dynamic to provide clean failover of HDFS services. The use of Isilon virtual rack requires more pools but more on that later.
In the sample configuration below; we have two pools, a static pool for stateful workflows [workflow1, this is non hdfs, let’s say it’s your smb workflow pool] and a second dynamic HDFS pool [HDFS] but you could get away with a single dynamic pool if needed.
see the latest Isilon Hadoop Best Practice doc for more:
https://www.emc.com/collateral/white-paper/h12877-wp-emc-isilon-hadoop-best-practices.pdf
# isi networks list pools –verbose
subnet0:pool0 – workflow1
In Subnet: subnet0
Allocation: Static
Ranges: 1
10.0.0.21-10.0.0.24
Pool Membership: 4
1:10gige-2 (up)
2:10gige-2 (up)
3:10gige-2 (up)
4:10gige-2 (up)
Aggregation Mode: Link Aggregation Control Protocol (LACP)
Access Zone: System (1)
SmartConnect:
Suspended Nodes : None
Auto Unsuspend … 0
Zone : isl-wkfl1.isilontest.foo.com
Time to Live : 0
Service Subnet : subnet0
Connection Policy: Round Robin
Failover Policy : Round Robin
Rebalance Policy : Automatic Failback
subnet0:pool1 – HDFS
In Subnet: subnet0
Allocation: Dynamic
Ranges: 1
10.0.0.25-10.0.0.50
Pool Membership: 4
1:10gige-2 (up)
2:10gige-2 (up)
3:10gige-2 (up)
4:10gige-2 (up)
Aggregation Mode: Link Aggregation Control Protocol (LACP)
Access Zone: hdfs (2)
SmartConnect:
Suspended Nodes : None
Auto Unsuspend … 0
Zone : isl-hdfs.isilontest.foo.com
Time to Live : 0
Service Subnet : subnet0
Connection Policy: Round Robin
Failover Policy : Round Robin
Rebalance Policy : Automatic Failback
This isl-hdfs.isilontest.foo.com is SmartConnect name used when connecting the compute cluster to the Isilon for HDFS access.
