Graphs all the Way Down - The BERNEXPO Configurator, Neos CMS and Neo4j

Transcript

1. THE BERNEXPO CONFIGURATOR, NEOS AND NEO4J GRAPHS ALL THE WAY DOWN
2. 2 HI THERE!
3. 3 MICHEL Senior Full-Stack Dev @JvMTECH Zurich c4ll-m3-j4ck Neos: Michel Löw
4. THE BERNEXPO PROJECT NEOS V9 NEO4J AND THE CONTENTGRAPH SPACETIME WHAT’S GONNA HAPPEN? 1 2 3 4
5. BERNEXPO T H E PR OJ E CT
6. THE PROJECT
7. T H E C US TO ME R ONE OF THE LARGEST EXHIBITION VENUES AND ORGANIZERS OF EXHIBITIONS IN SWITZERLAND
8. A USER-FRIENDLY BOOKING PLATFORM
9. A USER-FRIENDLY BOOKING PLATFORM AN ELEVATED EDITING EXPERIENCE
10. A USER-FRIENDLY BOOKING PLATFORM AN ELEVATED EDITING EXPERIENCE AN EXTENSIBLE DATA HUB
12. © 12
13. THE STACK not-so-microservices TM
14. NUXT JS GRAPH QL NEOS/FLOW REDIS MARIA DB BITBUCKET CLOUD UPSUN.COM GOOGLE CLOUD
15. DATENBANK CONTENT REPOSITORY NEOS FLOW NODE TYPES NEOS UI ADMIN INTERFACE GRAPHQL API KONFIGURATOR FRONTEND DASHBOARD BACKEND
16. NEOS V9 T H E G O O D , TH E B A D - A ND TH E UG LY
17. NEOS v9 PROVIDED A PERFECT PLATFORM FOR EDITORIAL WORK
18. STRUCTURAL LOGIC IN NODETYPES
19. STRUCTURAL LOGIC IN NODETYPES PACKAGE AND SERVICE-GROUP BUNDLING IN THE BACKEND
20. STRUCTURAL LOGIC IN NODETYPES PACKAGE AND SERVICE-GROUP BUNDLING IN THE BACKEND BUSINESS LOGIC WITH PROVIDERS AND DTOS IN FLOW CONTROLLERS
21. THE ARCHITECTURE WAS ALWAYS MEANT AS A STEPPING STONE
22. INSTEAD OF PHASING OUT, THE PLATFORM GREW RAPIDLY
23. THE PITFALLS
24. WE WERE MAYBE A BIT OPTIMISTIC ABOUT THE STATE OF THE BETA
25. INITIAL ISSUES WITH PUBLISHING, REBASING AND DISCARDING
26. INITIAL ISSUES WITH PUBLISHING, REBASING AND DISCARDING HIERARCHY-RELATIONS TABLES WITH ABOUT 1 MIO ROWS
27. INITIAL ISSUES WITH PUBLISHING, REBASING AND DISCARDING HIERARCHY-RELATIONS TABLES WITH ABOUT 1 MIO ROWS PERFORMANCE DEGRADED WITH LARGER DATABASE
28. THE SOLUTION
29. A COMPLETELY SEPARATE CACHING LAYER IN REDIS
30. FULL MANUAL CACHING OF ALL COMPOSITE DTOS
31. FULL CACHE REBUILD AFTER PUBLISH TO LIVE WORKSPACE FULL MANUAL CACHING OF ALL COMPOSITE DTOS
32. EDITOR WORKFLOW NEEDED TO BE ADAPTED AND DOCUMENTED FULL CACHE REBUILD AFTER PUBLISH TO LIVE WORKSPACE FULL MANUAL CACHING OF ALL COMPOSITE DTOS
33. THIS WORKED, BUT MADE US REALIZE WE NEEDED A DIFFERENT APPROACH FOR THE UNDERLYING DATA STRUCTURE
34. NEO4J T H E NE W KI D I N TO W N ? !
35. WHAT IS
36. A GRAPH NATIVE DATABASE BUILT WITH CONNECTED DATA IN MIND
37. NATIVE RELATIONSHIPS WITHOUT INDEX LOOKUPS
38. NATIVE RELATIONSHIPS WITHOUT INDEX LOOKUPS OPTIMIZED FOR GRAPH TRAVERSAL
39. NATIVE RELATIONSHIPS WITHOUT INDEX LOOKUPS OPTIMIZED FOR GRAPH TRAVERSAL EXCELLENT PERFORMANCE FOR CONNECTED DATA SCENARIOS
40. EASY TO UNDERSTAND QUERY LANGUAGE (CYPHER)
41. EASY TO UNDERSTAND QUERY LANGUAGE (CYPHER) BUILT-IN DATA-SCIENCE PROCEDURES
42. EASY TO UNDERSTAND QUERY LANGUAGE (CYPHER) BUILT-IN DATA-SCIENCE PROCEDURES FULLY OPEN-SOURCE
43. WHY NEO4J FOR NEOS?
44. THE CR ALREADY LEVERAGES GRAPH RELATIONSHIPS TO MANAGE CONTENT STRUCTURE
45. OUR PERFORMANCE BOTTLE-NECKS WERE MAINLY CAUSED BY RELATIONSHIP LOOKUPS
47. THE ADAPTER
48. DROP-IN REPLACEMENT FOR THE CURRENT DBAL-ADAPTER
49. SAME EVENT-LOG – DIFFERENT PROJECTION
50. SAME EVENT-LOG – DIFFERENT PROJECTION HANDLES NODES AS NODES, HIERARCHICAL RELATIONS AS EDGES
51. HANDLES NODES AS NODES, HIERARCHICAL RELATIONS AS EDGES IMPLEMENTS ALL INTERFACES REQUIRED BY THE CR SAME EVENT-LOG – DIFFERENT PROJECTION
53. LEVERAGE PATH ALGORITHMS FOR COMPLEX LOOKUPS
54. C o nt e n t S u b g ra p h : : f i nd S ub t r e e()
55. N e o 4 j C o nt e n t S u b g ra p h : :f i nd S ub t r e e( )
56. DEMO TIME
57. SPACETIME 57 W H E R E D O W E G O F R O M H E R E?
58. THE IDEA
59. CURRENTLY PROJECTIONS ONLY SHOW THE SUM OF ALL CHANGES – A SINGLE POINT IN TIME
60. THERE IS NO EASY ROLLBACK MECHANISM AND NO HISTORICAL STATE
61. NODES HAVE A NOTION OF THEIR TEMPORAL POSITION
62. NODES HAVE A NOTION OF THEIR TEMPORAL POSITION A SPACE-TIME-GRAPH PROVIDES ACCESS TO ALL VERSIONS
63. NODES HAVE A NOTION OF THEIR TEMPORAL POSITION A SPACE-TIME-GRAPH PROVIDES ACCESS TO ALL VERSIONS PROJECTIONS CAN BE MAPPED TO A POINT IN TIME
64. WHY?
65. PERFORMANCE OF LARGE CONTENT-GRAPHS IS NOT AN ISSUE WITH A GRAPH DATABASE
66. IT ADDS BUSINESS VALUE
67. PROVIDES PROOF OF DIGITAL CONTENT
68. PROVIDES PROOF OF DIGITAL CONTENT REMOVES IMMEDIATE NEED FOR INCREMENTAL CONTENT-BACKUPS
69. PROVIDES PROOF OF DIGITAL CONTENT REMOVES IMMEDIATE NEED FOR INCREMENTAL CONTENT-BACKUPS ALLOWS CONTENT-AUDITING AND DEBUGGING
70. WHERE DO WE GO FROM HERE?
71. JUST TO REITERATE: ADOPT NEOS 9 IN YOUR PROJECTS
72. THE ADAPTER IS AVAILABLE ON GITHUB AS AN EXPERIMENTAL RELEASE: gh://jvm-tech/JvMTECH.ContentGraph.Neo4jAdapter
73. QUESTIONS QUESTIONS ! QUESTIONS?
74. IMAGE SOURCES https://www.pexels.com/de-de/foto/foto-von-galaxy-207529/ https://www.pexels.com/de-de/foto/selektive-fokusfotografie-des-pinsels-in-der-nahe-der-farbpalette-1047540/ https://www.pexels.com/de-de/foto/zeitrafferfotografie-von-blauen-lichtern-373543/