Japanese researchers claim to have developed a method that allows to increase write performance up to 300%. Their method also lowers power consumption and decreases the amount of write and erase cycles. The researchers of the Chuo University announced their method at the 2014 IEEE International Memory Workshop (IMW), an international academic conference on semiconductor memory technology, which took place last week in Taipei, Taiwan.
The method from the Japanese researchers is based on the fact that NAND memory does not allow overwriting of the same memory area. An previous used area first has to be erased, a time consuming operation, and therefore SSDs try to write to space that's empty first. This results in data that's fragmented. SSDs try to solve the fragmentation using a technology called "garbage collection". This rearranges fragmented data so it can be read sequentially. It also cleans up (erases) areas that are no longer used so the SSD can write to it again. All these commands take 100ms or longer which decreases the performance of the SSD, the Japanese researchers argue.
The researchers claim to have developed a method that prevents the data from fragmenting. They have developed a method called "LBA (logical block address) scrambler" that sits between the filesystem of the operating system and the SSD controller. The LBA scrambler converts the logical addresses of data being written to reduce the effect of fragmentation.
Specifically, instead of writing data on a new blank page (see image above), data is written on a fragmented page located in the block to be erased next. As a result, the ratio of invalid pages in the block to be erased increases, reducing the number of valid pages that need to be copied to another area at the time of garbage collection.
When tested, the research team found that the writing speed of SSD was increased up to 300% and power consumption reduced up to 60%. The method also reduced he number of write/erase cycles by up to 55%, which has the benefit of increasing the life of the SSD. The researchers state that the method can be applied to existing production methods.
