The test results are based on a model in which the constant-bandwidth safety and security video traffic remained unaffected during disk rebuild and non disruptive upgrades. Some minimal video loss on random cameras occurred during activities caused by archive reconstruct processes when an Isilon node was failed.
Each recorder was tested on the same share with 2 or 4 statically defined paths to the file share across unique nodes. In a configuration with 2 connections, any single recorder must not be connected to the same pair of nodes as another recorder.
Cognyte actively load balances across all active paths to the file share so that the bandwidth from any one recorder to the file share via any node is the aggregate write bandwidth divided by the number of paths.
To control and predict the maximum bandwidth to any one node, be careful to avoid a node failure where 2 paths from different recorders increase the bandwidth to the same surviving node.
When a failure occurs, the additional bandwidth on a surviving node increases by the bandwidth of only one additional recorder.
We used 10 GbE interfaces with no more than two SMB connections per interface. A 10 GbE interface can accommodate up to four Recorder connections at the maximum Symphia VMS-supported values.
We performed all tests with a per-camera bandwidth of 4 Mb/s, so a single Recorder that handles 40 MB/s can support 40 to 80 such cameras.
We performed all tests with node or drive failures in place in the cluster (for example, with Isilon FlexProtect running) to ensure a worst-case scenario for all sizing parameters.
The following table provides bandwidth-sizing guidelines based on our test results.
Array | Symphia VMS version | OneFS version | Recorders per node | Bandwidth (MB/s) with disk SmartFail | Bandwidth (MB/s) with node SmartFail 1 | Drive size | ||
Per Recorder | Per node | Per Recorder | Per node | |||||
A2000 | 7.7.2 | 9.1 | 2 | 40 | 80 | 25 | 50 | 10 TB |
The CPU utilization percentage is between 30 and 50 during disk failure and recovery scenarios, such as Smartfail . This percentage falls within the desired performance range for Cognyte Symphia VMS.
The network throughput rate displays the total amount of data being read from and written to the disks in the cluster during a disk SmartFail. Network throughput is performing as expected.
The CPU utilization percentage is between 50 and 70 during node failure and recovery scenarios, such as Smartfail and adding the failed node back in. This percentage falls within the desired performance range for Cognyte Symphia VMS.
The network throughput rate displays the total amount of data being read from and written to the disks in the cluster during a node SmartFail. Network throughput is performing as expected.