Module 3 - Databases_on_AWS
Module 3 - Databases_on_AWS
Purpose-built databases,
the right tool for the right job
Vincent Oh
29 Apr 2022
Table of contents
Dev Ops
Relational Non-relational
databases databases
up to 128 TB
t0 t1 t2 t3 t4
Backtrack brings the database to a point in time without requiring restore from backups
• Backtracking from an unintentional DML or DDL operation
• Backtrack is not destructive. You can backtrack multiple times to find the right point in time
Net App
Old DB
Before ZDP
state state
Engine
Net App
state state New DB
User sessions terminate Engine
during patching
Storage Service
With ZDP
Old DB
Engine
Application
Networking
state
state
New DB
User sessions remain Engine
active through patching
© 2022, Amazon Web Services, Inc. or its Affiliates.
Fast database cloning
BENCHMARKS
Aurora
Primary Aurora Aurora
Replication Server
Replication Agent
Replica
Instance Replica Replica
(optional)
Async.
TYPE OF WRITE
© 2022, Amazon Web Services,
REDOInc. or its Affiliates.
LOG FRM FILES
Aurora Serverless
On-demand, auto-scaling database for applications with variable workloads
Application
Designed to support
Built for high durability 99.99%
of availability
WRITES READS
Hash(1) = 7B
CustomerOrdersTable
Global Table
Even faster—
DAX microsecond latency
Scales to millions of
requests per second
Amazon Amazon
DynamoDB RDS
Locality Global
Performance Milliseconds
to microseconds
Request Rate Millions
Amazon Neptune
• Recommendations
• Knowledge graphs
• Fraud detection
• Life sciences
Supports Apache TinkerPop Query billions of Six replicas of data across Build powerful queries
and W3C RDF graph models relationships with three Availability Zones easily with Gremlin and
millisecond latency with full backup and restore SPARQL
Amazon Timestream
Unnatural for Inefficient Rigid schema Difficult to scale Difficult to Limited data
time-series data time-series data inflexible for fast maintain high lifecycle
processing moving time-series availability management
data
Collect data at the rate of Adaptive query processing Built-in functions for Automated setup, configuration,
millions of inserts per engine maintains steady, interpolation, smoothing, server provisioning, software
second (10M/second) predictable performance and approximation patching
Amazon QLDB
Append-only, immutable All changes are Executes 2–3X as many Flexible document model,
journal tracks history of all cryptographically transactions as ledgers in query with familiar
changes that cannot be chained and verifiable common blockchain SQL-like interface
deleted or modified; get frameworks
full visibility into entire
data lineage
Amazon Keyspaces
• Restoring data
Single-digit-
millisecond
Apache Cassandra– No servers to performance at scale Highly available and
compatible manage secure
Use the same Cassandra No need to provision, Scale tables up and down 99.99% availability SLA
drivers and tools configure, and operate automatically within an AWS Region
large Cassandra clusters
Virtually unlimited throughput Data encrypted at rest;
and storage integrated with IAM
Keyspaces
Wide column
DMS | SCT