Application Development & Database Choices: Postgres Support for non Relational Data

Application
Development and
Database Choices:
Postgres Support for
non Relational Data
EDB CTO Team
December 2020

Slides and recording will be made available
Submit questions via Zoom – will be answering at end
Welcome – Housekeeping Items

This talk will cover the advanced features of
Postgres that make it the most-loved RDBMS
by developers and a great choice for non-
relational workloads.
Portions of this presentation were taken from https://momjian.us/presentations

© Copyright EnterpriseDB Corporation, 2020. All rights reserved.4
Agenda
The “Post” Relational Database
1.Postgres has won
2.Document-centric applications
3.Geographic Information Systems (GIS)
4.Business intelligence
5.Central data centers
6.Server-side languages

It is the most loved RDMS. Postgres is widely
adopted, is globally developed by hundreds of
people and is continually innovating.
1. Postgres Has Won

PostgreSQL Won
If you bet… you bet on PostgreSQL
Most Loved DatabaseMost Commonly Used Database Postgres Popularity
Source: Stack Overflow Developer Survey, 2019 Source: DB-Engines.com, 2020

Postgres as a service offering on all major clouds

Over 180 new features every year

Over 300 people contribute to each release

• 9.6 Parallel Query
• 10 Partitioning and Logical Replication
• 11 Transaction controlled stored procedures and Just In Time (JIT) Compilation
• 12 Pluggable Storage and Partitioning at Scale
• 13 Incremental Sort and Advanced Indexing and performance
About features
More recent release have taken on the more challenging features

• Strong support for the SQL Standard
• The optimizer has attracted top academics and practitioners
• Advanced features such as window functions
• The source for many commercial relational databases is Postgres and database add-ons
• True ACID compliance
A great relational database

Many Databases leverage Postgres
And … many more !!!

Postgres was designed from the start to be
extensible. This makes it a great choice for
non-relational (No SQL) applications.
2. Document-Centric Applications

• Full is naturally integrated with ANSI SQL in Postgres
• JSON and SQL queries use the same language, the same planner, and the same ACID compliant
transaction framework
• JSON and HSTORE are elegant and easy to use extensions of the underlying object-relational
model
JSON and ANSI SQL - A natural fit

JSON Examples
CREATE TABLE semi_structured_data (object JSONB);
….
{"name": "Apple Phone", "type": "phone", "brand": "ACME", "price": 200, "available": true,
"warranty_years": 1}
…
INSERT INTO semi_structured_data (object) VALUES
(’ { "name": "Apple Phone",
"type": "phone",
"brand": "ACME",
"price": 200,
"available": true,
"warranty_years": 1
} ')
Create a table JSONB Field
Simple JSON Data
Element
Insert this data element into the table
semi_structured_objects

{
"firstName": "John", // String Type
"lastName": "Smith",
"isAlive": true, // Boolean
"age": 25, // Number Type
"height_cm": 167.6,
"address": { // Object Type
"streetAddress": "21 2nd Street",
"city": "New York",
"state": "NY",
"postalCode": "10021-3100"
},
"phoneNumbers": [ // Object Array
{
"type": "home", "number": "212 555-1234"
},
{
"type": "office", "number": "646 555-4567"
}
],
"children": [],
"spouse": null
}
}
JSON Data Type Example

products=# insert into semi_structured_objects values('{ "firstName": "John", "lastName": "Smith", "isAlive": true, "age": 25,
"height_cm": 167.6,
"address": {
"streetAddress": "21 2nd Street", "city": "New York", "state": "NY", "postalCode": "10021-3100"
},
"phoneNumbers": [
{
"type": "home", "number": "212 555-1234"
},
{
"type": "office", "number": "646 555-4567"
}
],
"children": [],
"spouse": null
}
');
products=# select * from semi_structured_objects ;
{"name": "Apple Phone", "type": "phone", "brand": "ACME", "price": 200, "available": true, "warranty_years": 1}
{"age": 25, "spouse": null, "address": {"city": "New York", "state": "NY", "postalCode": "10021-3100", "streetAddress": ...
Can store different objects in same field
Different ROWs with different
attributes.

products=# select * from product;
weight | sku | name | manufacturer | instock | price
--------+-----+---------------+--------------+---------+---------
0.5 | 1 | Apple iPhone | Apple | t | $950.00
0.3 | 2 | Apple Watch | Apple | t | $220.00
0.2 | 3 | Apple earpods | Apple | t | $220.00
3 | 4 | Macbook Pro | Apple | t | $220.00
products=# select to_json(r) from (select sku as id, name as prod_name from product) r;
--------------------------------------
{"id":1,"prod_name":"Apple iPhone"}
{"id":2,"prod_name":"Apple Watch"}
{"id":3,"prod_name":"Apple earpods"}
{"id":4,"prod_name":"Macbook Pro"}
(4 rows)
Transform tables to JSON Format

products=# select * from semi_structured_objects ;
{"name": "Apple Phone", "type": "phone", "brand": "ACME", "price": 200, "available": true, "warranty_years": 1}
{"age": 25, "spouse": null, "address": {"city": "New York", "state": "NY", "postalCode": "10021-3100", "streetAddress": ...
products=# select object->>'name' as "Product Name" from semi_structured_objects where object->>'brand'='ACME';
Product Name
--------------
Apple Phone
(1 row)
products=#
SQL constructs to query the JSON DATA

JSON and ANSI SQL Example
SELECT DISTINCT
product_type,
data->>'brand' as Brand,
data->>'available' as Availability
FROM json_data
JOIN product
ON (product.product_type=semi_structured_objects.object->>'name')
WHERE semi_structured_objects.object->>'available'=true;
product_type | brand | availability
---------------------------+-----------+--------------
AC3 Phone | ACME | true
ANSI SQL
JSON
No need for programmatic logic to combine SQL and NoSQL in the
application – Postgres does it all

Performance Comparison: PostgreSQL 11/MongoDB 4.0
● Benchmarks published in June 2019 (follow up to similar analysis in 2014)
● Executed by ongres.com
● Using m5.4xlarge (16 vcores) on AWS EC2
● Details at (including the code and the engine to verify the findings)
http://info.enterprisedb.com/Performance-Benchmarks-PostgreSQL-vs-MongoDB.html

Performance Comparison: PostgreSQL 11/MongoDB 4.0
OLTP (sysbench) - Many Small Transactions - Large Data Set
PGBouncer (connection pooler) is key to
manage highly concurrent access

PostGIS is one of the most popular geospatial
database offerings in the market. Turning
Postgres into one of the most popular and
powerful geospatial database is free and
simple.
3. Geographic Information Systems (GIS)

Power through extensibility — Geo spatial
Postgres=# create EXTENSION postgis;
CREATE EXTENSION
Now have one of the world’s most popular
Geospatial Databases built on well established
industry standards.

PostGIS enables a new type of analysis
● What is the largest city with 100 miles of the Grand Canyon?
● How many households are within 1 mile of this fault line?
● If we relocate, the office how does the average commute distance change?
● How many cities within 150KM of Tampa have median income over $50,000.00?
● What truck drove the greatest distance yesterday?

PostGIS enables a new type of analysis
CREATE TABLE public.interesting_cities
(
id integer,
name character varying(100),
location geometry(Point,4326),
medium_hval money,
int population,
PRIMARY KEY (id)
)
A type introduced by
PostGIS

Analysis can now involve spatial references
-- How many Meters are between Boston (TOM) and Philadelphia (BRUCE)?
select ST_Distance (ST_Transform ((select location from interesting_cities
where name ='Boston'), 3587),
ST_Transform((select location from interesting_cities
where name = 'Philadelphia'), 3587));
st_distance
--------------------
431332.35327121057
(1 row)
ST_Distance and ST_Transform
are functions introduced by PostGIS
extension.

Can use Spatial AND Traditional analysis tools
-- Interesting cities within 150 KM of Boston with the 10 lowest medium home prices
select name , medium_hval, location from interesting_cities where
(select ST_Distance (ST_Transform (location, 3587),
ST_Transform( ( select location from interesting_cities
where name = 'Boston'), 3587) ) ) < 150000
ORDER BY medium_hval limit 10;
Spatial Query
Traditional RDBMS expression

PgAdmin 4 is part of PostGIS eco-system

Postgres has advanced functionality for
business intelligence.
4. Business Intelligence

Variable from first
expression passed to next
expression. (and again to
third expression)
Business Intelligence — Advanced CTE Features
WITH source (order_id) AS (
DELETE FROM orders WHERE name = ’my order’ RETURNING order_id
), source2 AS (
DELETE FROM items USING source WHERE source.order_id =
items.order_id )
INSERT INTO old_orders SELECT order_id FROM source;
Delete a given order,all the items associated with order and place order in a historical table.
Less code to maintain than on any other database
Fewer round trips with the server than on any other database

Business Intelligence — Windows Functions
SELECT depname, empno, salary, rank() OVER (PARTITION BY depname ORDER BY salary DESC)
FROM empsalary;
depname | empno | salary | avg
-----------+-------+--------+-----------------------
develop | 11 | 5200 | 5020.0000000000000000
develop | 7 | 4200 | 5020.0000000000000000
develop | 8 | 6000 | 5020.0000000000000000
develop | 10 | 5200 | 5020.0000000000000000
personnel | 5 | 3500 | 3700.0000000000000000
personnel | 2 | 3900 | 3700.0000000000000000
sales | 3 | 4800 | 4866.6666666666666667
sales | 1 | 5000 | 4866.6666666666666667
sales | 4 | 4800 | 4866.666666666666667
(9 rows)
compare each employee's salary with the average salary in his or her department

Business Intelligence — Specialized Indexes
Specialized Indexes for all data types and access patterns
Index Type Optimized For
B-Tree Range queries with low selectivity and largely unique
values. The traditional database index.
HASH Equality lookups on large datasets (key / value store)
use cases.
GiST Unstructured Data i.e. Geo Spatial Types
GIN JSON Data, Full Text Search, JSONB Data
SP-GiST SP-GIST is ideal for indexes whose keys have many
duplicate prefixes

Business Intelligence — Specialized Indexes
Specialized Indexes for non-relational data
Index Type Optimized For
BRIN Time series data, multi-terabyte tables
PARTIAL When only a specific set of values will be looked up
COVERING For access patterns to unindex values navigated to
by an index.
EXPRESSION Allow for variances in keys

Postgres can function as a central integration
point for your data center using Foreign Data
Wrappers.
5. Central Data Center

Power through extensibility — Foreign Data Wrappers
postgres=# create extension postgres_fdw;
CREATE EXTENSION

Foreign Data Wrappers
CREATE SERVER postgres_server FOREIGN DATA WRAPPER postgres_fdw OPTIONS (host ’localhost’, dbname
’fdw_test’);
create SERVER oracle_server foreign data wrapper oracle_fdw options (dbserver
'//<oracle_servefr_IP>/<sid>' );
CREATE SERVER mongo_server FOREIGN DATA WRAPPER mongo_fdw OPTIONS (address '127.0.0.1', port '27017');
CREATE SERVER hadoop_server FOREIGN DATA WRAPPER hdfs_fdw OPTIONS (host '127.0.0.1');
CREATE SERVER mysql_server FOREIGN DATA WRAPPER mysql_fdw OPTIONS (host '127.0.0.1', port '3306');

Foreign Data Wrapper Access
Application Stack

Postgres has server-side language support for
almost all developers.
6. Server-Side Languages

Server-Side Programming Languages
• PL/Java
• PL/Python
• PL/R
• PL/pgSQL (like PL/SQL)
• PL/Ruby
• PL/Scheme
• PL/sh
• PL/Tcl
• PL/v8 (JavaScript)
• SPI (C)
CREATE LANGUAGE plpython3u;
CREATE OR REPLACE FUNCTION pymax (a integer,
b integer) RETURNS integer AS
$$
if a > b:
return a
return b
$$ LANGUAGE plpython3u;
SELECT pymax(12, 3);
pymax
-------
12
(1 row)

Learn More
Other resources
Thank You
Next Webinar: December 16
Postgres Enterprise Manager 8.0:
Something For Everyone
Postgres Pulse EDB Youtube
Channel

Application Development & Database Choices: Postgres Support for non Relational Data

Recommended

More Related Content

What's hot (20)

Similar to Application Development & Database Choices: Postgres Support for non Relational Data (20)

More from EDB (20)

Recently uploaded (20)

Application Development & Database Choices: Postgres Support for non Relational Data