Solid-state drive | EPFL Graph Search

A solid-state drive (SSD) is a solid-state storage device that uses integrated circuit assemblies to store data persistently, typically using flash memory, and functioning as secondary storage in the hierarchy of computer storage. It is also sometimes called a semiconductor storage device, a solid-state device or a solid-state disk, even though SSDs lack the physical spinning disks and movable read–write heads used in hard disk drives (HDDs) and floppy disks. SSD also has rich internal parallelism for data processing. In comparison to hard disk drives and similar electromechanical media which use moving parts, SSDs are typically more resistant to physical shock, run silently, and have higher input/output rates and lower latency. SSDs store data in semiconductor cells. cells can contain between 1 and 4 bits of data. SSD storage devices vary in their properties according to the number of bits stored in each cell, with single-bit cells ("Single Level Cells" or "SLC") being generall

Querying Persistent Graphs using Solid State Storage

Manoussos Gavriil Athanassoulis

Recent advances in solid state technology have led to the introduction of Solid State Drives (SSDs). Todays SSDs store data persistently using NAND flash memory. While SSDs are more expensive than hard disks when measured in dollars per gigabyte, they are significantly cheaper when measured in dollars per random I/O per second. Additional storage technologies are under development, Phase Change Memory (PCM) being the next one to enter the marketplace. PCM is nonvolatile, it can be byte-addressable, and in future Multi Level Cell (MLC) devices, PCM is expected to be denser than DRAM. PCM has lower read and write latency compared to NAND flash memory, and it can endure orders of magnitude more write cycles before wearing out. Recent research has shown that solid statedevices can be particularly beneficial for latency-bound applications involving dependent reads. Latency-bound applications like path processing in the context of graph processing or Resource Description Framework (RDF) data processing are typical examples of these applications.We demonstrate via a custom graph benchmark that even an early prototype Phase Change Memory device can offer significant improvements over mature flash devices (1.5x - 2.5x speedup in response times). We take this observation further by building Pythia, a prototype RDF repository tailor-made for Solid State Storage to investigate the predicted benefits for these type of workloads that can be achieved in a properly designed RDF repository. We compare the performance of our repository against the state of the art RDF-3X repository in a limited set of tests and discuss the results. We finally compare the performance of our repository running on a PCM-based device against a state of the art flash device, showing that there is indeed significant gain to be achieved by using PCM jfor RDF processing.

2013

Path Processing using Solid State Storage

Manoussos Gavriil Athanassoulis

Recent advances in solid state technology have led to the introduction of Solid State Drives (SSDs). Todays SSDs store data persistently using NAND ﬂash memory. While SSDs are more expensive than hard disks when measured in dollars per gigabyte, they are signiﬁcantly cheaper when measured in dollars per random I/O per second. Additional storage technologies are under development, Phase Change Memory (PCM) being the next one to enter the marketplace. PCM is nonvolatile, it can be byte-addressable, and in future Multi Level Cell (MLC) devices, PCM is expected to be denser than DRAM. PCM has lower read and write latency compared to NAND ﬂash memory, and it can endure orders of magnitude more write cycles before wearing out. Recent research has shown that solid state devices can be particularly beneﬁcial for latency-bound applications involving dependent reads. Latency-bound applications like path processing in the context of graph processing or Resource Description Framework (RDF) data processing are typical examples of these applications. We demonstrate via a custom graph benchmark that even an early prototype Phase Change Memory device can offer signiﬁcant improvements over mature ﬂash devices (1.5x - 2.5x speedup in response times). We take this observation further by building Pythia, a prototype RDF repository tailor-made for Solid State Storage to investigate the predicted beneﬁts for these type of workloads that can be achieved in a properly designed RDF repository. We compare the performance of our repository against the state of the art RDF-3X repository in a limited set of tests and discuss the results. We ﬁnally compare the performance of our repository running on a PCM-based device against a state of the art ﬂash device, showing that there is indeed signiﬁcant gain to be achieved by using PCM for RDF processing.

2012

Flash-aware Database Transactions

Radu Ioan Stoica

Over the past 40 years, hard disks, the traditional building block of storage systems, have become increasingly slower when compared to the other system components such as the main memory and the CPU. Hard disks face mechanical constraints that cause their I/O bandwidth to lag behind capacity growth, while the access latency has remained virtually unchanged for the past 20 years. The growing gap between the main memory and the persistent storage brought us to a point where I/O accesses are the main bottleneck that limits the performance of database management systems. Several new solid-state storage technologies have been under development and are now commercially successful, with NAND flash memory being the most mature. Such technologies store data durably, have no mechanical constraints to limit their I/O performance, and can bridge the growing gap between the main memory and the persistent storage. Solid-state drives, however, have very different characteristics compared to hard disks, making it challenging to integrate such technology. For decades, database management systems have been implicitly developed and optimized around the model of a slow rotating hard disk. In this thesis, we focus on transactional database workloads composed of large numbers of small concurrent queries. Transactional workloads generate large numbers of scattered I/O requests at the level of the storage system that are traditionally challenging for hard disks to service. Solid-state drives, on the other hand, are a much better suited technology as they service natively I/O requests in parallel and their performance is not constrained by access locality. The approach we take in this thesis is to traverse the I/O stack of a database management system, i.e. the major software layers that define how data is stored and accessed, and show how these components can, with careful consideration of flash behavior and through novel algorithms, make efficient use of flash-based storage devices. The methodology we use is to observe the data access patterns and the data access frequencies at each level of the I/O stack and design new algorithms and data structures that match the characteristics of the workload with the constraints of flash memory. We use a bottom up approach to revisit the Storage Device, the Storage Manager, and the Caching layers of a data management system and show how each component can better exploit flash technology. At the level of the Storage Device, we show how a solid-state drive can minimize the overhead of storing data on the raw flash chips by exploiting the write skew present in real-life transactional workloads. We then introduce a log-structured data placement policy for the Storage Manager of a database system that can transparently adapt the data access pattern to eliminate all random I/O writes that are problematic for flash solid-state drives. Finally, we propose an architecture for the Caching layer that allows database systems to efficiently move data between main memory and a fast solid-state drive without having to incur the large overhead and limitations of the traditional buffer pool design.

EPFL2014

Flash-aware Database Transactions

Radu Ioan Stoica

EPFL2014