The test objective is to maintain constant safety and security video traffic during select storage failure scenarios, such as disk rebuild and controller failure. During these scenarios, frame loss was observed on the ME4 platform as tier 1 storage.
We performed tests using 2, 5, and 12 Mb/s cameras. Video data was written to the capacity tier or disk groups consisting of NL-SAS drives.
The following tests were performed for all sizing parameters:
Test | Description | Result |
Controller restart | Reboot controller A storage. | This test was performed with all three Disk Groups (12, 24, and 48). The controller restarted within five minutes of the time to complete the restart and resume the recording. |
Controller Shutdown and start | Shutdown controller A and then start. | During this test, it has taken more than one hour to start recording at the correct bandwidth. |
Nic Failure | Fail the iSCSI target on the storage side to identify the impact. | During this failure test, there is a drop in write bandwidth for less than a minute, however there is not any media loss. |
Disk Failure | Disk fail test simulated to identify rebuild rate and video loss during disk reconstruction. | This is most impactful in carried out tests. Write bandwidth drops 30-50% of the bandwidth achieved during the normal load benchmark. |
Copy Back test | Removed disk will be inserted back to check the copy back time and impact. | This test is less impact full than the copy back test. We have seen 20-40% of media loss achieved during normal load. |
Single Power Failure | Check the impact on storage when single power is failed on the storage. | There is no impact because of single power source failure on the storage. |
Results during normal loads
The following table provides bandwidth-sizing guidelines based on our test results during normal loads:
Array | Version | RAID | Camera bit rate (Mb/s) | Bandwidth MB/s (Mb/s) | Camera count | Controller CPU load | IOPS | Disk IOPS |
ME4012 | 9 | Adapt Linear | 2 | 92 (736) | 350 | 15 | 1220 | 150 |
5 | 133 (1064) | 200 | 10 | 1381 | 145 | |||
12 | 236 (1888) | 150 | 12 | 2224 | 135 |
Array | Version | RAID | Camera bit rate (Mb/s) | Bandwidth MB/s (Mb/s) | Camera count | Controller CPU load | IOPS | Disk IOPS |
ME4012 | 9 | Adapt Linear | 2 | 134 (1072) | 520 | 19 | 1846 | 145 |
5 | 237 (1896) | 350 | 18 | 2416 | 145 | |||
12 | 288 (2304) | 185 | 11 | 2693 | 135 |
Array | Version | RAID | Camera bit rate (Mb/s) | Bandwidth MB/s (Mb/s) | Camera count | Controller CPU load | IOPS | Disk IOPS |
ME4012 | 9 | Adapt Linear | 2 | 177 (1416) | 680 | 24 | 2350 | 100 |
5 | 257 (2056) | 385 | 18 | 2619 | 90 | |||
12 | 297 (2376) | 185 | 18 | 2760 | 80 |
Results during failure tests
The following table provides bandwidth-sizing guidelines based on our test results during failure tests:
Array | Version | RAID | Camera bit rate (Mb/s) | Bandwidth MB/s (Mb/s) | Camera count | Controller CPU load | IOPS | Disk IOPS |
ME4012 | 9 | Adapt Linear | 2 | 50 (400) | 200 | 13 | 890 | 145 |
5 | 124 (992) | 180 | 10 | 1015 | 140 | |||
12 | 130 (1040) | 85 | 10 | 1620 | 130 |
Array | Version | RAID | Camera bit rate (Mb/s) | Bandwidth MB/s (Mb/s) | Camera count | Controller CPU load | IOPS | Disk IOPS |
ME4012 | 9 | Adapt Linear | 2 | 75 (600) | 290 | 18 | 1424 | 140 |
5 | 132 (1056) | 205 | 16 | 1860 | 140 | |||
12 | 160 (1280) | 105 | 11 | 2078 | 130 |
Array | Version | RAID | Camera bit rate (Mb/s) | Bandwidth MB/s (Mb/s) | Camera count | Controller CPU load | IOPS | Disk IOPS |
ME4012 | 9 | Adapt Linear | 2 | 95 (760) | 370 | 15 | 1723 | 115 |
5 | 143 (1144) | 225 | 13 | 1916 | 100 | |||
12 | 165 (1320) | 100 | 13 | 2010 | 95 |