tlantic @ elasticsearch poa meetup

26
Elasticsearch for retail You know, for search!

Upload: vinny-linck

Post on 18-Jul-2015

569 views

Category:

Technology


1 download

TRANSCRIPT

Elasticsearch for retailYou know, for search!

@vinnylinck # Chief Mobile Architect

More than a search system....

... the agenda is about:

how

uses

to build

TLANTIC

elasticsearch

TLANTIC

We do software for retail

10 years of success

7

ERP

WMS

POS

DW

IN-STORE

LOGISTICS

MANAGEMENTCUSTOMER

SALES

MOBILE RETAIL SUITE™ A PROVEN SOLUTION

http:// aboutTheRetailers .com:9200/

{ok: true,

status: 200,

name: "RetailersWorld",

facts: {

data_amount: "millions of records",

availability: "high",

search_scenario: “not sure how to find things",

common_approach: “fulltext",

does_fulltext_works: false

},

tagline: “Retail is about people…"

}

... but people are messy...

SELECT *FROM myawesomesoftware

WHERE ‘text’ LIKE‘%stuff%’

SELECT *FROM mysearchtableWHERE MATCH(text)

AGAINST (‘stuff’)

SELECT *FROM mysearchtableWHERE MATCH(text)

AGAINST (‘+stuff –”stuffler”’IN BOOLEAN MODE)

SELECT * FROM jobsWHERE role = ‘DEVELOPER’

AND MATCH(job_description) AGAINST (‘javascript’)

SELECT * FROM jobs JJOIN jobs_benefits jb ON j.id =

jb.job_idWHERE j.role = ‘DEVELOPER’

AND (MATCH(job_description) AGAINST (‘javascript -asp’) IN

BOOLEAN MODE)AND jb.free_coffee = TRUE

THIS IS NOT A SEARCH SYSTEM!You will need one...

There is a “software” for that:

There is a software SOLUTION for that:

elasticsearch is a real search engine

• Efficient indexing of data• All fields / combination

• Analyzing data:• Text Search: tokenizing, stemming, filtering

• Understanding locations

• Date parsing

• Relevance scoring

• Not just split(‘ ‘, $text)

• Understand patterns: URLs, e-mail, currencies, hashtags, twitter @mentions.

• Analyze conjugations and plurals:• Fishing, Fished, Fish, Fisher > Fish• Better > Good

• Filters stop words: not every character is meaningful when indexing a HTML code or web contents.

• Distance searches: geo polygons, bounding box searches, searching nearby...

• Relevance: think of Google PageRank

• Mappings generated on the fly

Retailers want more!And we are using ElasticSearch for that.

With elasticsearch...

• We found performance: 1M records on 40s (bulk insert)

• The app users will be able to find a product regarding the nearest store.

• We can scale easily, handling BIG DATA scenarios with safety.

• We found a stable nosql search system and a powerful JSON API.

• Customers can find the information even when they don’t know how to write that properly... In miliseconds!

• We found the cloud: (bonsai.io)

• It fits like a glove on mobile solutions.

• We found integration with out existing systems...

elasticsearch will be part of our next product generation...

http:// tlantic .com/thank_you

{

ok: true,

status: 200,

name: “Vinicius Linck",

e_mail:“[email protected]”,

e_mail_2: “[email protected]”,

twitter: “@vinnylinck",

}