MetaData persistence using storage class memory

Abstract

Storage Class Memory (SCM) blends the best properties of main memory and hard disk drives. It offers non-volatility and byte addressability, and promises short access times with low cost per bit. Earlier research in this field explored designs exploiting SCM features and used either simulations or theoretical models for evaluations. In this work, we explore the design challenges for achieving non-volatility using real SCM hardware that is available now: Flash-Backed DRAM. We present performance analysis of flash-backed DRAM and describe the system issues involved in achieving true non-volatility using the system memory hierarchy which was designed assuming that data is volatile. We present software abstractions which allow applications to be redesigned easily using SCM features, without having to worry about system issues. Furthermore, we present case studies using two applications with different characteristics: an SSDbased caching layer used in enterprise storage (Flash Cache) and an in-memory database (SolidDB), and redesign them using software abstractions. Our performance evaluations reveal that SCM aware Flash Cache design could enable persistence with less than 2% degradation in performance. Similarly, redesigning SolidDB persistence layer using SCM improved the performance by a factor of two. To the best of our knowledge, this is the first work that evaluates SCM performance and demonstrates application redesign using real SCM hardware.

Meeting Real-Time Constraint of Spectrum Management in TV Black-Space Access

“…In [14], the average response time of a query in disk-based database for dataset size of 1,200, 4,800, and 19,000 is reported as 6, 26, and 220 milliseconds, respectively. For telecommunications, fast response is achieved by in-memory database [15], [16]. For example, IBM solidDB [17], can reduce the response time from 375 milliseconds in disk-based database to 50 milliseconds, and achieve an average of 1.2 million transactions per second for a dataset of 1 million records.…”

“…For example, IBM solidDB [17], can reduce the response time from 375 milliseconds in disk-based database to 50 milliseconds, and achieve an average of 1.2 million transactions per second for a dataset of 1 million records. In [15], SolidDB is reported to achieve 140,000 transactions per seconds on Flash-backed DRAM storage device.…”

Isotope

“…In the past, smart block devices such as HP AutoRAID [Wilkes et al 1996] were restricted to enterprise settings due to their reliance on battery-backed RAM; today, SSDs routinely implement indirection in FTLs, using supercapacitors to flush metadata and data on a power failure. Software block stores in turn can store metadata on these SSDs, on raw flash, or on derivatives such as flash-backed RAM [Jose et al 2013] and Auto-Commit Memory [SanDisk 2015b].…”