A financial analytics VM on a hybrid (SSD plus HDD) AHV cluster runs a high-read workload whose hot data is small enough to fit on the SSD tier, and Prism shows read latency as the primary bottleneck when data is served from HDD. You need a feature that favors read performance rather than capacity efficiency. Which option should you enable?

On hybrid clusters, AOS tiers data automatically: hot extents live on SSD and cold extents are moved down to HDD by Information Lifecycle Management. A read-heavy workload can still hit HDD latency when part of its data is classified as cold and moved down. Flash Mode, enabled per VM or per vDisk in Prism, pins that data to the SSD tier so it is not tiered down, which keeps read latency consistently low for a working set that fits within the SSD capacity allotted to Flash Mode. Compression reduces capacity consumption and usually costs little, but it doesn't keep data on SSD, so it doesn't fix HDD read latency. Deduplication targets identical blocks to save space and does not change which tier the VM's data lives on. Erasure coding converts cold data to data and parity strips to save capacity; reads of encoded data can involve more nodes, so it favors efficiency over latency. Exam caveat: Flash Mode applies only to hybrid clusters and uses a limited share of SSD capacity, so reserve it for a few critical VMs; Prism Central storage policies cover encryption, compression, QoS throttling and replication factor, not read caching. Operational check: enable Flash Mode on the VM, then compare its read latency and the share of reads served from SSD in Prism over the next business cycle.