Performance
Fault tolerance
Load Balancing
Client Scalability
Robustness and Stability
StorNext File System Tuning Distributed LAN Client Vs. Legacy Network Attached Storage
DLC outperforms NFS and CIFS for
DLC handles faults transparently, where possible. If an I/O is in progress and a NIC fails, the I/O is retried on another NIC (if one is available). If a Distributed LAN Server fails while an I/O is in flight, the I/O is retried on another server (if one is running). When faults occur, applications performing I/O will experience a delay but not an error, and no administrative intervention is required to continue operation. These fault tolerance features are automatic and require no configuration.
DLC automatically makes use of all available Distributed LAN Servers in an active/active fashion, and evenly spreads I/O across them. If a server goes down or one is added, the load balancing system automatically adjusts to support the new configuration.
As the following table shows, DLC supports a significantly larger number of clients than legacy NAS protocols:
Largest Tested Configuration
| NFS | CIFS | DLC | |
Number of Clients Tested (via |
|
|
| |
4 | 4 | 1000 | ||
simulation) | ||||
|
|
| ||
|
|
|
|
The code path for DLC is simpler, involves fewer file system stacks, and is not integrated with kernel components that constantly change with every operating system release (for example, the Linux NFS code).
StorNext File System Tuning Guide | 22 |