InvisibleWriteRule: the sufficient condition for safely omittable write operations in concurrency control

04/17/2019
by   Sho Nakazono, et al.
0

Concurrency Control (CC) is the heart of the correctness and is the current bottleneck of OLTP databases. Many schedulers for CC are proposed to reduce coordination between transaction processing. In fact, they only improve concurrency for read operations. Generally, databases based on conventional schedulers execute transactions with lock and wait. Thus, in write intensive and contended workload, all conventional schedulers cannot scale enough in many-core environments. To reduce coordination between write operations, we propose an InvisibleWrite (IW) operation which is an omittable write operation under the specific condition. Omitting IW operations result in reducing coordination between write operations. However, a schedule including IW operations sometimes violates the correctness conditions of OLTP databases such as serializability. To guarantee serializability and other correctness conditions, we propose InvisibileWriteRule (IWR) which is the sufficient condition for schedulers. All IWR schedulers, which satisfy IWR and generate IW operations, have the guarantee of serializability, recoverability, and linearizability. We can extend all conventional schedulers to IWR schedulers with tiny overhead and implementation cost. To indicate the efficiency and extensibility of IWR schedulers, we extend three conventional schedulers such as Silo, TicToc, and MVTO. Our experimental results show that every IWR scheduler achieves significant performance, especially in write intensive and contended workload. In the best case, Silo-based IWR scheduler achieves about 6.5x faster throughput than the original scheduler.

READ FULL TEXT

page 1

page 2

page 3

page 4

research
04/07/2019

Obtaining Progress Guarantee and GreaterConcurrency in Multi-Version Object Semantics

Software Transactional Memory Systems (STMs) provides ease of multithrea...
research
12/25/2018

A 256kb 9T Near-Threshold SRAM With 1k Cells per Bit-Line and Enhanced Write and Read Operations

In this paper, we present a new 9T SRAM cell that has good write-ability...
research
03/29/2018

Unleashing and Speeding Up Readers in Atomic Object Implementations

Providing efficient emulations of atomic read/write objects in asynchron...
research
08/31/2021

Latency-Redundancy Tradeoff in Distributed Read-Write Systems

Data is replicated and stored redundantly over multiple servers for avai...
research
11/04/2016

Flat ORAM: A Simplified Write-Only Oblivious RAM Construction for Secure Processors

Oblivious RAM (ORAM) is a cryptographic primitive which obfuscates the a...
research
06/09/2023

CAWL: A Cache-aware Write Performance Model of Linux Systems

The performance of data intensive applications is often dominated by the...
research
11/04/2020

Predict and Write: Using K-Means Clustering to Extend the Lifetime of NVM Storage

Non-volatile memory (NVM) technologies suffer from limited write enduran...

Please sign up or login with your details

Forgot password? Click here to reset