Tweet
Tweet

Slide 1

Slide 1 text

It's getting faster: moving from relational databases to MongoDB Michel Krämer

Slide 2

Slide 2 text

What data?

Slide 3

Slide 3 text

3D city models

Slide 4

Slide 4 text

Owner Street Number Metadata 3D geometry

Slide 5

Slide 5 text

Heterogeneous data sources

Slide 6

Slide 6 text

Municipality Urban planners Utility Environmental Emergency Citizens

Slide 7

Slide 7 text

Why?

Slide 8

Slide 8 text

Layer Feature Geometry Metadata 1 n n 1 n m n n 1 1 n 1 ...

Slide 9

Slide 9 text

Mesh Face Geometry n 1 Vertex Color Texture coord. 1 n 1 1 1 n n n ...

Slide 10

Slide 10 text

Mesh Face Geometry n 1 Vertex Color Texture coord. 1 n 1 1 1 n n n ... Blob

Slide 11

Slide 11 text

MySQL PostgreSQL Throughput (relatively) Oracle

Slide 12

Slide 12 text

Downtime during backups

Slide 13

Slide 13 text

Downtime when scaling

Slide 14

Slide 14 text

Hibernate Pro Contra

Slide 15

Slide 15 text

Documents

Slide 16

Slide 16 text

Geometry Metadata ... Building Geometry Metadata ... Building ...

Slide 17

Slide 17 text

NoSQL NoSQL NoSQL NoSQL NoSQL NoSQL NoSQL NoSQL NoSQL NoSQL NoSQL NoSQL NoSQL NoSQL NoSQL NoSQL NoSQL NoSQL NoSQL NoSQL MongoDB CouchDB NoSQL NoSQL NoSQL NoSQL NoSQL NoSQL NoSQL NoSQL NoSQL NoSQL

Slide 18

Slide 18 text

MongoDB CouchDB

Slide 19

Slide 19 text

MongoDB CouchDB fast stable community scalable ad-hoc queries fast stable community scalable ad-hoc queries ... ...

Slide 20

Slide 20 text

MySQL MongoDB 800 200 400 600 Objects/s (read)

Slide 21

Slide 21 text

MySQL MongoDB 500 125 250 375 Objects/s (write)

Slide 22

Slide 22 text

MongoDB

Slide 23

Slide 23 text

MongoDB update A update B

Slide 24

Slide 24 text

MongoDB update A update B update A update C

Slide 25

Slide 25 text

MongoDB update A update B update A update C ?

Slide 26

Slide 26 text

1 + 1 = 2

Slide 27

Slide 27 text

MongoDB does not use [...] transactions with rollback, as it is designed to be lightweight and fast [...]. By keeping transaction support extremely simple, performance is enhanced [...]. MongoDB Developer FAQ

Slide 28

Slide 28 text

MVCC

Slide 29

Slide 29 text

snapshot client 1 ... snapshot client 2 ...

Slide 30

Slide 30 text

CouchDB's approach doc A, rev 1 client 1 doc A, rev 2 client 2

Slide 31

Slide 31 text

Git's approach File A C1

Slide 32

Slide 32 text

Git's approach File B File A C1 C2

Slide 33

Slide 33 text

Our approach snapshot (index) client 1 ... snapshot (index) client 2 ... C1 C2 C1 C2

Slide 34

Slide 34 text

doc A C1 index client 1

Slide 35

Slide 35 text

doc B doc A C1 index client 1

Slide 36

Slide 36 text

doc B doc A C1 C2 index client 1

Slide 37

Slide 37 text

C1 C2 client 1 client 2

Slide 38

Slide 38 text

C1 C2 client 1 client 2 C3

Slide 39

Slide 39 text

C1 C2 client 1 client 2 C3 conflict!

Slide 40

Slide 40 text

C1 C2 client 1 client 2 C3

Slide 41

Slide 41 text

C1 C2 client 1 client 2 C3 C4

Slide 42

Slide 42 text

C1 C2 C4 C3 C5 C6 master project a project b Named branches

Slide 43

Slide 43 text

C1 C2 C3 Accessing the history client 1 client 2

Slide 44

Slide 44 text

Downside Performance Memory API

Slide 45

Slide 45 text

MongoDB MongoMVCC MVCC + = https://github.com/igd-geo/mongomvcc

Slide 46

Slide 46 text

Michel Krämer Fraunhofer IGD [email protected] @michelkraemer +Michel Krämer