Once a single sample passes the test, the entire order will be placed, which is the starting point for rework on many projects. The role of small-batch trial assembly is to expose the three risks of batch consistency, real device identification and on-site environment in advance. The cost is usually a dozen devices and two to three weeks.
Sample size by batch and port type
- According to the supply batch: a certain proportion of each batch, generally no less than 5 to 10
- According to equipment type: switches and network cards of different brands or models occupy one group each.
- According to the link form: the shortest link, the longest link, and the link with redundant connectors each cover one group
- Take a separate group in the dense rack scenario and observe the status after the adjacent ports are fully plugged in.
By covering these categories, the sample size does not have to be large to expose the main differences. If the plan also containsGigabit single mode dual fiber one optical one electrical fiber optic transceiverAs with the module, the complete transceiver and module must be sampled separately.
The observation period and recording items must be agreed in advance
Continuous operation observation generally covers a complete working cycle, including nighttime and load peaks. Just keep three record items: port up/down times, bit errors or renegotiation events, and DDM reading drift over time. The on-site temperature, humidity and ventilation conditions in the cabinet are recorded together to facilitate future recurrence.
Release standards and rectification closed loop
The release criteria are clearly written before the trial installation begins: zero-tolerance items (such as repeated port restarts) and tolerable items (such as small drifts in readings) are processed in a graded manner. When the standard is not met, the phenomenon, sample number, and recurrence conditions will be sent back together. After the seller gives a rectification plan, only the failed items will be retested, without the need to retest the entire batch, and the cycle is controllable.
Three things should be left after the trial assembly: judgment table, batch correspondence, and follow-up sampling plan. EB-LINK supports providing trial installation prototypes and individual test records by project. For details of the plan, you can call 0755-83179002 to communicate.