Putting the SQL Back in NoSQL - October 2022 - All Things Open

Confidential and Proprietary. Do not distribute without Couchbase consent. © Couchbase 2021. All rights reserved.
Power of Relational With the Flexibility of JSON
August 2022
Matthew D. Groves | Product Marketing Manager
Putting the SQL Back in NoSQL

Confidential and Proprietary. Do not distribute without Couchbase consent. © Couchbase 2021. All rights reserved. 2
Querying with SQL

Querying JSON with JavaScript/JSON

01/
02/
03/
04/
Relational Data to JSON
Basics of SQL++
How to Get Started
Questions
Agenda

1 Relational Data to JSON

Modeling Data in a Relational World
Billing
Connections
Purchases
Contact
Customer

Relational Challenges
• "Vertical" scaling –
expensive and has a
ceiling
• "Horizontal" scaling –
complex and difficult to
maintain with a
relational database
• Rigid table-based data
modeling
• Schema changes can
be time consuming,
costly
Flexibility
• Limited by disk access
• Often needs an external
caching solution
• Complex queries and
large number of JOINs
• Single database server
means single point of
failure
• Replication/distribution,
if available, requires
more overhead
Lorem ipsum Lorem ipsum
In other words, why consider NoSQL?
Scaling Performance Availability

Modeling Data in a JSON World
Customer
{ JSON }
{ JSON }
{ JSON }
{
"Name" : "Jane Smith",
"DOB" : "1990-01-30",
"Billing" : [ … ],
"Connections" : [ … ],
"Purchases" : [ …. ]
}

{
"Name" : "Jane Smith",
"DOB" : "1990-01-30",
"Billing" : [
{
"type" : "visa",
"cardnum" : "5827-2842-2847-3909",
"expiry" : "2019-03"
},
{
"type" : "mcard",
"cardnum" : "6274-2842-2847-3909",
"expiry" : "2019-03"
}
],
"Connections" : [
{
"CustId" : "XYZ987",
"Name" : "Joe Smith"
},
{
"CustId" : "SKR007",
"Name" : "Sam Smith"
}
],
"Purchases" : [
{ "id":12, item: "mac", "amt": 282.52 }
{ "id":19, item: "ipad2", "amt": 623.52 }
], . . .
}
DocumentKey: CBL2016
CustomerID Name DOB
CBL2016 Bob Jones 1980-01-29
CustomerID Type Cardnum Expiry
CBL2016 visa 5927… 2020-03
CBL2016 mcard 6273… 2019-11
CustomerID ConnId Name
CBL2016 XYZ987 Joe Smith
CBL2016 SKR007 Sam
Smith
CustomerID item amt
CBL2016 mac 282.52
CBL2016 ipad2 623.52
CustomerID Address Email
CBL2016 123
Main
a@b.gov
CBL2016 456
Elm
c@d.gov
Contacts
Customer
Billing
Connections
Purchases

But at what cost?
• Easy to read, designed to be read
by humans
• Lots of training material, taught in
schools and bootcamps
• Concise, declarative syntax
• Some data has to be modelled as
many-to-many and many-to-one
• Even with JSON consolidation, the
need for joining data together is
critical for some use cases
JOINs
• Within one JSON document,
writes are ACID
• What about when transferring
money between accounts?
• Even with JSON consolidation, the
need for interaction between
multiple documents is critical for
some use cases
SQL standard is the most popular
data query language
What if I can't put everything into one
document?
Not every use case can fit into one
document either
There are GOOD things about relational databases!
SQL ACID transactions

Enter SQL++
SQL for JSON?
https://resources.couchbase.com/analytics/sql-book

Relational-to-NoSQL Translation
SQL Server Couchbase Note
Server Cluster ➕ Scalability / High Availability / Built-in caching
Catalog (Database) Bucket
Schema Scope Often just "dbo" in SQL Server or "_default" in Couchbase
Table Collection ➖ Pre-defined columns/constraints
Row Document ➕ Flexible JSON
tSQL SQL++ (Previously known as N1QL in Couchbase)
Primary Key Document Key ➖ Compound keys
➖ "No" keys
Index Index

2 Basics of SQL++

Route document example
route_5966 ← key
{
"airlineid": "airline_24", ← "foreign key"
"sourceairport": "MCO",
"destinationairport": "SEA",
"equipment": "737",
"schedule": [
{"day": 1, "utc": "13:25:00", "flight": "AA788"},
{"day": 4, "utc": "13:25:00", "flight": "AA419"},
{"day": 5, "utc": "13:25:00", "flight": "AA519"}
]
}

Airline document example
airline_24 ← key
{
"active": "Y",
"callsign": "AMERICAN",
"country": "United States",
"iata": "AA",
"icao": "AAL",
"name": "American Airlines",
}

SELECT

How do I get that airline name?
route_5966
{
"airlineid": "airline_24",
"airlineName": "Delta", ← denormalize / duplicate
"sourceairport": "MCO",
"destinationairport": "SEA",
"equipment": "737",
"schedule": [
{"day": 1, "utc": "13:25:00", "flight": "AA788"},
{"day": 4, "utc": "13:25:00", "flight": "AA419"},
{"day": 5, "utc": "13:25:00", "flight": "AA519"}
]
}

SELECT . . . JOIN

SELECT . . . JOIN . . . HAVING . . . ORDER

BEGIN/COMMIT/ROLLBACK
ACID Transactions
ID: 1924
{
"name": "Emma",
"balance": 0
}
ID: 4872
{
"name": "Matthew",
"balance": 1400
}

BEGIN/COMMIT/ROLLBACK
ACID Transactions
BEGIN TRANSACTION;
UPDATE customer
SET balance = balance + 100
WHERE META().id = "1924";
UPDATE customer
SET balance = balance - 100
WHERE META().id = "4872";
SELECT name, balance
FROM customer
WHERE META().id IN ["4872","1924"];
COMMIT TRANSACTION;

Indexing
bucket scope collection
bucket scope collection field(s) / expression(s)

Index Advisor

Cost-based Optimization
Which index to use?
CREATE PRIMARY index ON customer;
CREATE INDEX ix1 ON customer(name, zip, status);
CREATE INDEX ix2 ON customer(zip, name, status);
CREATE INDEX ix3 ON customer(zip, status, name);
CREATE INDEX ix4 ON customer(name, status, zip);
CREATE INDEX ix5 ON customer(status, zip, name);
CREATE INDEX ix6 ON customer(status, name, zip);
CREATE INDEX ix7 ON customer(status);
CREATE INDEX ix8 ON customer(name);
CREATE INDEX ix9 ON customer(zip);
CREATE INDEX ix10 ON customer(status, name);
CREATE INDEX ix11 ON customer(status, zip);
CREATE INDEX ix12 ON customer(name, zip);
SELECT *
FROM customer
WHERE name = "Matt"
AND zip = "43123"
AND status = "Standard";
UPDATE STATISTICS for `customer`
(status, name, zip);

User-Defined Functions (UDFs)
CREATE FUNCTION getAirlineRank(in_country) { (
SELECT a.id,
a.name AS airline,
COUNT(r.id) AS routecount,
PERCENT_RANK() OVER (ORDER BY COUNT(r.id)) AS `rank`
FROM `travel-sample`.ìnventory`.àirline` a
JOIN `travel-sample`.ìnventory`.`route` r
ON META(a).id = r.airlineid
WHERE a.country = in_country
GROUP BY a.id, a.name
)};

Eventing
function OnUpdate(doc, meta) {
var minDistance = 1000;
var results =
SELECT *
FROM `travel-sample`.inventory.route
WHERE distance > $minDistance;
for (var route of results) {
log(route);
break;
}
results.close();
}

Analytics

Tools for Moving Data

3 How to Get Started

Couchbase Capella
https://www.couchbase.com/products/capella

Couchbase Playground
https://couchbase.live

Demo

THANK YOU

Putting the SQL Back in NoSQL - October 2022 - All Things Open

More Related Content

Putting the SQL Back in NoSQL - October 2022 - All Things Open