Du relationnel au NoSql

SQL & SGBDR

Système de gestion de base de données

Relationnel

Modèle relationnel

SQL & SGBDR

➢ Algèbre Relationnelle➢ Théorie des ensembles➢ +40 ans d’optimisations

Ceux qui ne comprennent pas la théorie

des ensembles

????

Ceux qui comprennent la

théorie des ensembles

venn diagram sql

ACID

➢ Transaction ACIDAtomic: Everything in a transaction succeeds or the entire

transaction is rolled back.Consistent: A transaction cannot leave the database in an

inconsistent state.Isolated: Transactions cannot interfere with each other.Durable: Completed transactions persist, even when servers

restart etc.

Limites

➢ Difficilement distribuable➢ Schéma contraignant➢ Object Relational Mapper

■ Impedance mismatch➢ Performances➢ Coûts

Problème de lenteurs ?

Conception

➢ Lenteurs○ Souvent un problème de conception○ Tout modèle est optimisé pour

un cas d’utilisation

➢ SQL ou NoSQL ne change rien à ce problème

Index

➢ Un index est une structure de données redondante organisée de manière à accélérer certaines recherches.

Normalisation / Dénormalisation

Monté en charge verticale

n1-standard-11 CPU 3,75Go RAM51$/mois

n1-highmem-3232 CPU208Go RAM1900$/mois

Réplication pour la lecture

master

slave

slave

write

read

readwrite

Sharding pour augmenter les capacités d’écriture

A B Creadwrite

readwrite

readwrite

Concepts

➢ Monté en charge horizontale/verticale➢ Dénormalisation➢ Sharding/partitionning➢ Réplication

NoSQL

Qui ?

Google Big table map/reduce (2006)

Amazon Dynamo (2007)

Facebook Cassandra (2008)

=> Changement d’échelle

Comment ?

Distribution on commodity hardware

It will fail

Abandon des principes ACIDEn particulier la Consistance

➢ BASE○ Basic Availability○ Soft-state○ Eventual consistency

Comment ?

ACID � ➢ Strong consistency➢ Isolation � ➢ Focus on “commit”➢ Nested transactions

Availability?➢ Conservative (pessimistic)

➢ Difficult evolution (e.g. schema)

BASE � ➢ Weak consistency

○ stale data OK➢ Availability first➢ Best effort➢ Approximate answers OK➢ �Aggressive (optimistic)

➢ Simpler➢ Faster �➢ Easier evolution

NoSQL ACID vs BASE

continuum

CAP Theorem

Consistency

Availibity

SGBDR

1. The network is reliable.2. Latency is zero.3. Bandwidth is infinite.4. The network is secure.5. Topology doesn't change.6. There is one administrator.7. Transport cost is zero.8. The network is homogeneous.

Fallacies of distributed computing

CAP Theorem

Consistency

Availibity

SGBDR

CAP Theorem

Consistency

PartitionTolerance

BigtableMongoDbRedis

CAP Theorem

PartitionTolerance Availibity

DynamoCassandra

CAP Theorem

Consistency

PartitionTolerance Availibity

No

Pick two ! Eric A. Brewer 2000

Towards Robust Towards Robust Distributed Systems

Cas d’usages

➢ Performance & réduction des coûts○ Distribution○ élasticité (cloud)

➢ Agilité des développements○ NoORM○ NoSchema

➢ Modélisations alternatives○ Graph

NoSQL database type

One More Thing...

One More Thing...

?

La base de donnée NoSQL la plus utilisée est

One More Thing...

Lucene !

Solr / ElasticSearch

Ressources

➢ Presentations○ Dynamo and BigTable in light of the CAP theorem○ Towards Robust Towards Robust Distributed Systems

➢ Papers○ Dynamo: Amazon’s Highly Available Key-value Store○ Bigtable: A Distributed Storage System for Structured Data

➢ Videos○ Introduction to NoSQL by Martin Fowler

➢ Books○ NoSQL Distilled○ Seven Databases in Seven Weeks: A Guide to Modern Databases and

the NoSQL Movement