Architecting VMware vSAN 6.2 : vSAN Technology and Features Overview : 3.2 vSAN Key Features : 3.2.16 Comparing Hybrid and All-Flash Configurations
   
3.2.16 Comparing Hybrid and All-Flash Configurations
As described previously, vSAN 6.0 and later provides support for two different configuration options—a hybrid configuration that leverages both flash-based devices and mechanical disks, and an all-flash configuration. The hybrid configuration is the traditional approach used in the previous release of vSAN. Hybrid configurations use a flash-based device to provide the caching tier, and mechanical disks to provide capacity and persistent data storage. An all-flash configuration employs entirely flash-based devices for both the caching and capacity tiers. The cost of using all-flash configurations can be significant, because 10-GbE networking is an absolute requirement, in addition to the additional costs of flash storage.
The all-flash vSAN configuration brings improved, highly predictable and uniform performance, regardless of workload, and additional features when compared with hybrid configurations. However, it do not automatically assume that an all-flash configuration delivers improved performance over a hybrid design. While it is true to say that performance is more consistent when employing an all-flash environment, the actual performance, in terms of read and write latency, is very much dependent on the workload’s dataset, the hardware being employed, and overall design.
Both hybrid and all-flash clusters carry a 10 percent of consumed capacity recommendation for the flash cache layer, even though cache is used differently in each configuration:
Hybrid clusters – The caching algorithm attempts to maximize both read and write performance. By default, 70 percent of the cache is allocated for storing frequently read disk blocks, which minimizes accesses to slower mechanical disks. The remaining 30 percent is used for write caching.
All-flash clusters – Employ two types of flash: very fast and durable write cache, and larger and cost-effective capacity flash. In an all-flash configuration, 100 percent of cache is allocated for writes, because read performance from capacity flash is more than sufficient to handle demanding enterprise workloads. In addition, far more writes are held by the cache, and written to the capacity layer only when needed, extending the life of the capacity flash tier.
Consider the following general guidelines about drives (flash or mechanical):
Compatibility – The model of the PCIe or SSD devices must be listed in the vSAN section of the VMware Compatibility Guide.
Performance – PCIe devices generally have faster performance than SSD devices.
Capacity – The maximum capacity available for PCIe devices is generally greater than the maximum capacity that is currently listed for SSD devices for vSAN in the VMware Compatibility Guide.
Write endurance – The write endurance of the PCIe or SSD devices must meet the requirements for capacity or for cache in all-flash configurations, and for cache in hybrid configurations.
Cost – PCIe devices generally have higher cost than SSD devices.
In addition to these general guidelines, the following are considerations for using all-flash configurations:
vSAN 6.0 at a minimum must be used.
10-GbE network is required.
Maximum number of all-flash nodes is 64.
Flash Read Cache reservation SPBM capability is not used.
All drives must be marked as flash.
Drive endurance becomes an important design consideration.