Scribd
Upload
55 views7 pages

Newsql Databases: Corso Di Sistemi E Architetture Per Big Data

The document discusses NewSQL databases, which aim to provide the scalability of NoSQL systems while maintaining the ACID guarantees of traditional databases. It describes Google's Spanner and VoltDB as examples of NewSQL databases, and notes key aspects of how they implement replication and concurrency control to achieve these goals. The document also provides background on the motivations for building NewSQL databases and an overview of features like Spanner's use of TrueTime for synchronization.

Uploaded by

Nhat Nguyen
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
55 views7 pages

Newsql Databases: Corso Di Sistemi E Architetture Per Big Data

The document discusses NewSQL databases, which aim to provide the scalability of NoSQL systems while maintaining the ACID guarantees of traditional databases. It describes Google's Spanner and VoltDB as examples of NewSQL databases, and notes key aspects of how they implement replication and concurrency control to achieve these goals. The document also provides background on the motivations for building NewSQL databases and an overview of features like Spanner's use of TrueTime for synchronization.

Uploaded by

Nhat Nguyen
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 7

Università degli Studi di Roma Tor Vergata

Dipartimento di Ingegneria Civile e Ingegneria Informatica

NewSQL Databases

Corso di Sistemi e Architetture per Big Data


A.A. 2016/17

Valeria Cardellini

The reference Big Data stack

High-level Interfaces
Support / Integration

Data Processing

Data Storage

Resource Management

Valeria Cardellini - SABD 2016/17 1


Relational database services

•  RDBMS pros:
–  ACID transactions
–  Relational schemas (and schema changes without
downtime)
–  SQL queries
–  Strong consistency

•  RDBMS cons:
–  Lack of horizontal scalability, to hundreds or
thousands of servers

Valeria Cardellini - SABD 2016/17 2

NewSQL databases

•  How to build a relational database service that is both


strongly consistent and horizontally scalable?

•  NewSQL: a class of modern RDBMSs that seek to


provide the same scalable performance of NoSQL
systems for OLTP read-write workloads while
maintaining ACID guarantees of traditional DB
systems
–  Support SQL

Valeria Cardellini - SABD 2016/17 3


NewSQL examples

•  Google’s Spanner
–  Also available as cloud service in Google Cloud Platform:
Cloud Spanner https://cloud.google.com/spanner/
•  Google’s F1
–  Built on top of Spanner
•  VoltDB
-  And H-Store, its research prototype predecessor
http://hstore.cs.brown.edu
-  Developed by M. Stonebraker (2015 ACM Turing award)
•  Clustrix
–  Closed source
•  NuoDB
–  Closed source, no support for stored procedures

Valeria Cardellini - SABD 2016/17 4

Replication in NewSQL
•  Multi-master or masterless schemes
–  Any node can receive update statements

•  VoltDB and Clustrix


–  A transaction/session manager receives the updates, which
are forwarded to all replicas and executed in parallel
•  Google Spanner
–  Uses Paxos state machine replication to guarantee that a
sequence of commands will be executed in the same order
in all the replica nodes

Valeria Cardellini - SABD 2016/17 5


Spanner

•  Motivations:
–  “We provide a database instead of a key-value
store to make it easier for programmers to write
their applications”
–  “We consistently received complaints from users
that Bitable can be difficult to use for some kinds
of applications”

Valeria Cardellini - SABD 2016/17 6

What is Spanner

•  Wide-area distributed multiversion database


-  General-purpose transactions (ACID)
-  SQL query language
-  Schematized tables
-  Semi-relational data model

•  Running in production
-  Storage for Google’s ad data
-  Replaced a sharded MySQL database

Valeria Cardellini - SABD 2016/17 7


Spanner overview

•  Feature: lock-free distributed read transactions


•  Property: external consistency of distributed
transactions
–  First system at global scale
•  Implementation: integration of concurrency control,
replication, and 2PC
–  Correctness and performance
•  Enabling technology: TrueTime
–  Interval-based global time
–  Based on hardware-assisted time synchronization using
GPS clocks and atomic clocks
–  Accuracy: ~ 1 ms!

Valeria Cardellini - SABD 2016/17 8

Concurrency control in Spanner

•  Hybrid approach
–  Read-write transactions are implemented through read-write
locks, but read-only transactions are lock-free

•  Why is it possible?
–  Spanner stores multiple versions of data, and a read
transaction is basically a read at a “safe” timestamp

Valeria Cardellini - SABD 2016/17 9


VoltDB
•  In-memory database
•  Starting point
–  Open source RDBMS ran on memory-based file system
•  Over 80% of time spent on page buffer management, index
management, and concurrency management
•  Only 12% of time spent doing the real work
–  Lead to H-Store
•  Features
–  Horizontal scale-out on commodity hardware with linear
scalability
–  Full and strong ACID compliance
–  High concurrency
–  Reliable disk persistence
–  High availability

Valeria Cardellini - SABD 2016/17 10

VoltDB

•  Tables are partitioned over multiple servers,


and clients can call any server
–  Transparent distribution but the user can choose
the sharding attribute
•  Selected tables can be replicated over
servers, e.g. for fast access to read-mostly
data
•  Shards are replicated, so that data can be
recovered in the event of a node crash
•  Database snapshots are also supported,
continuous or scheduled
Valeria Cardellini - SABD 2016/17 11
VoltDB and concurrency control

•  Alternative design based on two assumptions


–  Assumption 1: total available memory is large
enough to store the entire data store
–  Assumption 2: all user transactions are short-lived
and can be very efficiently executed over in-
memory data
•  Then, all transactions are executed
sequentially in a single-threaded, lock-free
environment

Valeria Cardellini - SABD 2016/17 12

References

•  Golinger at al., “Data management in cloud


environments: NoSQL and NewSQL data stores”, J.
Cloud Comp., 2013. http://bit.ly/2oRKA5R
•  Corbett et al., “Spanner: Google’s Globally
Distributed Database”, OSDI 2012.
http://bit.ly/2nyJBrb
•  Stonebraker and Weisberg, “The VoltDB Main
Memory DBMS”, 2013. http://bit.ly/2okw837

Valeria Cardellini - SABD 2016/17 13

You might also like