Recently words cloud and virtualization are more and more common. And when you think about it, it actually make sense. To accomplish best scaling possibilities, to extract the full benefit of your hardware investment, virtualization make a lot of sense! NetApp with ONTAP 7-mode approach did a little bit of that bringing the vFilers. However, it still had a lot of limitations, for instance all resources were limited to a single Filer that was hosting the vFiler. With introduction of Clustered ONTAP NetApp brings much more benefits from virtualization to storage systems, to give you a few:
- Introduction of Storage Virtual Machine (SVM) which I have already briefly discussed in NetApp cDOT – what is SVM or vserver ? post. In a nutshell SVM is acting as a dedicated virtual storage controller, with its own data volumes, CIFS shares, NFS exports, and LUNs. It is not “linked” to single node – it can utilize physical storage resource pool, containing all nodes within the cluster
- Different type of physical storage controller models can be connected to build a single cluster
- NetApp cDOT brings Quality of Service (QoS) to help managing resource utilization between storage workloads
- Almost all maintenance ops can be executed without downtime, including software and firmware updates.
What is a cluster?
Cluster is a group of physical storage controllers. Each Storage controller is a cluster’s node. All nodes in the cluster can pool their resources together – data workload can be distributed across the member nodes. All communication and data trasfer between member nodes take place over a 10Gb cluster-interconnect network to which all nodes are connected (I mention that a little bit already in NetApp – Clustered vs 7-mode Scaling post).
A cluster is typically build upon one or more HA pair. HA (High Availability) pair is a setup of two storage controllers in which both nodes are actively host and serve data, but they are also capable of taking over their partner’s job. All disk shelves in HA pair are cabled to both controllers.
Such configuration is not required. As I mentioned in my previous post (Clustered vs 7-mode Scaling) there can be even single-node cluster configurations. However to ensure the best quality and stability production clusters are almost always build upon HA pairs. Single node cluster is a special confiugration that supports small storage deployments.
Maximum number of nodes
ONTAP 9 that only serve NFS and/or CIFS data (often called NAS clusters) can scale-up to maximum 24 nodes (in other words cluster can be build up 12 HA pairs). cluster that hosts iSCSI and FC (often called SAN clusters) can scale up to a maximum of 8 nodes. The limit can be lowered – depending on the hardware controller model.
Verify Your configuration via Command Line
If you want to see the basics information about your cluster, You can execute the command cluster show:
nyc_nas_cl001::> cluster show Node Health Eligibility --------------------- ------- ------------ nyc_node16 true true nyc_node17 true true nyc_node25 true true nyc_node26 true true nyc_node27 true true nyc_node28 true true 6 entries were displayed.
As you can see, my cluster is called nyc_nas_cl001, and it is build upon 6 nodes. If you wish to get more information about the nodes, you can execute the command node show:
nyc_nas_cl001::> node show Node Health Eligibility Uptime Model Owner Location --------- ------ ----------- ------------- ----------- -------- --------------- nyc_node16 true true 479 days 16:38 FAS3250 NYC, NY nyc_node17 true true 479 days 17:10 FAS3250 NYC, NY nyc_node25 true true 223 days 13:31 FAS8060 NYC, NY nyc_node26 true true 223 days 13:31 FAS8060 NYC, NY nyc_node27 true true 118 days 13:02 FAS8060 NYC, NY nyc_node28 true true 118 days 13:12 FAS8060 NYC, NY 6 entries were displayed.
As I mentioned previously, in most enterprise setups Clustered ONTAP is build upon HA pairs. If you wish to verify that you can execute command:
nyc_nas_cl001::> storage failover show Takeover Node Partner Possible State Description -------------- -------------- -------- ------------------------------------- nyc_node16 nyc_node17 true Connected to nyc_node17 nyc_node17 nyc_node16 true Connected to nyc_node16 nyc_node25 nyc_node26 true Connected to nyc_node26 nyc_node26 nyc_node25 true Connected to nyc_node25 nyc_node27 nyc_node28 true Connected to nyc_node28 nyc_node28 nyc_node27 true Connected to nyc_node27 6 entries were displayed.
In my example there are 3 HA pairs: nyc_node16-nyc_node17, nyc_node25-nyc_node26, nyc_node27-nyc_node28.