Thursday, May 16, 2013

Basics of NetApp Data ONTAP Cluster-mode

With ever increasing business needs the amount of data need to be stored, managed and backed up is getting complicated. To handle this kind of requirements industry has two models: scale up and scale out.

Scale up: 
1. limits performance and capacity
2. additional controllers are managed independently and do not provide any shared resources
3. not the best model for larger environments

Scale out or clustering:
1. no limit on performance and capacity
2. additional controllers are added seamlessly to take advantage of the shared resources
3. all the operations including hardware replacements, upgrades and almost everything can be done non-disruptively
4. NetApp Data ONTAP cluster mode supports both SAN and NAS environments along with features like thin provisioning, deduplication, compression and data replication
5. Unified architecture with ability to support multiple data access protocols like FC, FCoE, iSCSI,  NFS, pNFS, CIFS

Non Disruptive Operations:
1. Businesses today have rigid requirements and downtime cannot be tolerated.
2. Data ONTAP Cluster-Mode was built in with NDO as its fundamental feature to keep storage infrastructure remain up and to serve data during hardware and software maintenance and refresh operations
3. Data volumes can be move non-disruptively with data motion across the cluster at any time

NetApp 7G to Cluster-mode transistion:
It has always been a big gap for netapp to merge both the products, but with ONTAP 8 it allowed 7G customers to run ONTAP 8 in 7-mode while giving first step in the eventual move to a clustered environment. ONTAP 8.1 OS is a major step in merging 7G and cluster-mode.

Difference between 7G and Cluster-mode:

In ONTAP cluster-mode systems, each node has a vol0 root volume similar to the 7G systems. In addition, each Vserver has a root volume that is the root of the namespace that can be composed of many volumes spread throughout the cluster

7-mode features not in Cluster-mode:
1. MetroCluster
2. SnapVault
3. Snaplock
4. Synchronous snapmirror, Qtree Snapmirror
5. Volume snapmirror between 32-bit and 64-bit aggr
6. Virtual storage console, SMVI
7. DataMotion for vFiler
8. SyncMirror
9. NetApp storage encryption disks
10. IPv6

Cluster-mode features not in 7-mode:
1. LIF migration and failover
2. Cluster management as a single system
3. SMB 2.1
4. NFSv4.1, pNFS
5. Asynchronous volume-level replication
6. Load-sharing SnapMirror functionality
7. Onboard antivirus scanning (McAfee and Sophos)

1. 2 to 24 nodes, by looking like a single file system
2. Throughput scales linearly to multiple gigabytes per second
3. Can be scaled to petabytes
4. Online load balancing and scaling
5. High-performance computing (HPC)

1. Mix of FC, SATA and SAS drives
2. NDU data movement between tiers

Note: data doesn't need to be moved between tiers for clients to be able to access it. FAS 6200 series can access the SATA storage just as easily as they access the FC and SAS storage. The reason to use tiered storage within a cluster is to improve performance, not accessibility.

Cluster Resources:
1. Three types of Vservers: nodes, cluster, administration
2. Nodes
3. Aggregates
4. Volumes
5. Namespaces

Node Vserver:
-Physical node
-Associated with cluster LIFs and node management LIFs

Administration Vserver:
-Physical cluster
-Associated with cluster management LIF

Cluster Vservers:
-Not associated with any single node
-Contains resources within scope: Namespace, volumes, data LIFs, protocol servers (NFS,CIFS,FC,FCoE and iSCSI)

Note: When a cluster is created, the administration Vserver is automatically created, where as a node Vserver is automatically created when a node is joined to cluster. Cluster Vservers are created by administrators to build global namespaces.
Post a Comment
Read more: