Microsoft launched a new project today that’s designed to help make flash storage more useful to large enterprises and cloud providers. Called Project Denali, it’s supposed to help deal with several issues with implementing solid state drives in large scale datacenters.
Denali works by splitting apart the monolithic components of an SSD into two different modules. The drive deals with bad blocks, media, and power failures, presenting a set of clean memory blocks. Companies can then create an optimized software interface to handle tasks like garbage collection and wear leveling, based on the applications that they run.
Right now, SSDs behave like hard drives, even though they perform very differently. Hardware manufacturers produce devices that manage all of the functions that Denali aims to split up. That’s great for driving adoption, but large-scale cloud providers like Microsoft miss out on opportunities to optimize their workloads for the flash storage they have available.
Denali’s split approach is supposed to reduce cost of SSD deployment, as well as allow accelerated development of both hardware and software for datacenter applications. It’s part of Microsoft’s work with the Open Compute Project, which is supposed to accelerate datacenter innovation through the creation of open source hardware.
Splitting apart those functions also allow the development of dedicated silicon that can handle the management of data over Denali-compliant SSDs across entire datacenters at high speeds. That makes particular sense for Microsoft, which has deployed a fleet of field-programmable gate arrays in its datacenters that may be useful for providing this sort of data management.
Microsoft has built a prototype Denali SSD with its partner Cnex Labs, and it plans to finalize the specification for the hardware in the coming months and make it broadly available later this year.