The L2ARC is the 2nd Level Adaptive Replacement Cache, and is an SSD based cache that is accessed before reading from the much slower pool disks. The L2ARC is currently intended for random read workloads. This statistic shows L2ARC accesses if L2ARC cache devices are present, allowing its usage and performance to be observed.
As described in Overhead, breakdown such as by file name would be the most expensive to leave enabled.
To investigate L2ARC misses, check that the L2ARC has grown enough in size using the Advanced Analytic Cache: L2ARC size. The L2ARC typically takes hours, if not days, to warm up hundreds of Gbytes when feeding from small random reads. The rate can also be checked by examining writes from Cache: L2ARC I/O bytes. Also check the Advanced Analytic Cache: L2ARC errors to see if there are any errors preventing the L2ARC from warming up.
Cache: ARC accesses by L2ARC eligibility can also be checked to see if the data is eligible for L2ARC caching in the first place. Since the L2ARC is intended for random read workloads, it will ignore sequential or streaming read workloads, allowing them to be returned from the pool disks instead.