Efficient Flash Memory Read Request Handling Based on Split Transactions

Flash memory is a storage class memory widely used in mobile computing systems due to its small size, low power consumption, fast access time, and high shock and vibration resistance. Flash memory based storage systems exploit chip-level parallelism and hide the latency of flash memory operations through request scheduling. However, conventional scheduling techniques are inadequate for handling series of small random requests, degrading the overall performance of the system. This paper presents a flash memory request handling technique based on split transactions to improve the read performance – in particular, the performance of multiple small read requests. This technique decouples the first and second phase of each flash memory read request and services them in a different manner. The first phase of the operation is issued at the earliest time possible, while the second phase is delayed until the operation latency expires or the next read request arrives to the same chip. This request handling technique based on split transactions has been incorporated into a prototype flash memory based storage system on an FPGA development board. Experimental results from this prototype show that for small random read requests, this technique improves read performance by up to 46%.