The document discusses using JSON data with relational databases. It provides examples of defining JSON columns, querying and modifying JSON documents using functions like JSON_VALUE and JSON_INSERT. It also shows how to create constraints to enforce data integrity when storing JSON data and create virtual columns to index JSON attributes for improved querying performance. In summary, the document demonstrates how to flexibly store and query both structured and semi-structured data using JSON features in MariaDB.

1. Somewhere between schema
and schemaless
SHANE K JOHNSON
MARIADB CORPORATION

2. About me
I’m not a morning person,
but I like breakfast.

3. About me
I’m not a NoSQL person,
but I like JSON.

4. What’s the point?

5. What’s the point?

6. What’s the point?
If a movie can be both
science fiction and
western…

7. A data model can be comprised of
structured and semi-structured data.

8. JSON
Relational
Flexibility
Simplicity
Ubiquity
Data integrity
Transactions
Reliability

9. What’s the difference?
Structured data is externally described
Semi-structured data is self described

10. Structured and semi-structured data
with relational + JSON
Structured Semi-structured

11. Structured and semi-structured data
with relational + JSON
Structure described
by the schema
Structure
described by the
data

12. Creating and querying
JSON documents

13. Example #1: definition
CREATE TABLE IF NOT EXISTS products (
id INT NOT NULL PRIMARY KEY AUTO_INCREMENT,
type VARCHAR(1) NOT NULL,
name VARCHAR(40) NOT NULL,
format VARCHAR(20) NOT NULL,
price FLOAT(5, 2) NOT NULL,
attr JSON NOT NULL);

14. Example #2: create
INSERT INTO products (type, name, format, price, attr) VALUES
('M', 'Aliens', 'Blu-ray', 13.99,'{"video": {"resolution": "1080p", "aspectRatio": "1.85:1"}, "cuts":
[{"name": "Theatrical", "runtime": 138}, {"name": "Special Edition", "runtime": 155}], "audio":
["DTS HD", "Dolby Surround"]}');
INSERT INTO products (type, name, format, price, attr) VALUES
('B', 'Foundation', 'Paperback', 7.99, '{"author": "Isaac Asimov", "page_count": 296}');

15. Example #3: read field
SELECT name, format, price,
JSON_VALUE(attr, '$.video.aspectRatio') AS aspect_ratio
FROM products
WHERE type = 'M';
name format price aspect_ratio
Aliens Blu-ray 13.99 1.85:1

16. Example #4: null field
SELECT type, name, format, price,
JSON_VALUE(attr,$.video.aspectRatio') AS aspect_ratio
FROM products;
type name format price aspect_ratio
M Aliens Blu-ray 13.99 1.85:1
B Foundation Paperback 7.99 NULL

17. Example #5: contains field
SELECT name, format, price,
JSON_VALUE(attr, '$.video.aspectRatio') AS aspect_ratio
FROM products
WHERE type = 'M' AND
JSON_CONTAINS_PATH(attr, 'one', '$.video.resolution') = 1;
name format price aspect_ratio
Aliens Blu-ray 13.99 1.85:1

18. Example #6: contains value
SELECT id, name, format, price
FROM products
WHERE type = 'M' AND
JSON_CONTAINS(attr, '"DTS HD"', '$.audio') = 1;
name format price aspect_ratio
Aliens Blu-ray 13.99 1.85:1

19. Example #7: read array
SELECT name, format, price,
JSON_QUERY(attr, '$.audio') AS audio
FROM products
WHERE type = 'M';
name format price audio
Aliens Blu-ray 13.99 [
"DTS HD",
"Dolby Surround"
]

20. Example #8: read array element
SELECT name, format, price,
JSON_VALUE(attr, '$.audio[0]') AS default_audio
FROM products
WHERE type = 'M';
name format price audio
Aliens Blu-ray 13.99 DTS HD

21. Example #9: read object
SELECT name, format, price,
JSON_QUERY(attr, '$.video') AS video
FROM products
WHERE type = 'M';
name format price video
Aliens Blu-ray 13.99 {
"resolution": "1080p",
"aspectRatio": "1.85:1"
}

22. Example #10: read objects
SELECT name,
JSON_QUERY(attr, '$.audio') AS audio,
JSON_QUERY(attr, '$.video') AS video
FROM products
WHERE type = 'M';
name audio video
Aliens [
"DTS HD",
"Dolby Surround"
]
{
"resolution": "1080p",
"aspectRatio": "1.85:1"
}

23. Example #11: field equals
SELECT id, name, format, price
FROM products
WHERE type = 'M' AND
JSON_VALUE(attr, '$.video.resolution') = '1080p';
name format price aspect_ratio
Aliens Blu-ray 13.99 1.85:1

24. Example #12: indexing (1/2)
ALTER TABLE products ADD COLUMN
video_resolution VARCHAR(5) AS (JSON_VALUE(attr, '$.video.resolution')) VIRTUAL;
EXPLAIN SELECT name, format, price
FROM products
WHERE video_resolution = '1080p';
id select_type table type possible_keys
1 SIMPLE products ALL NULL

25. Example #13: indexing (2/2)
CREATE INDEX resolutions ON products(video_resolution);
EXPLAIN SELECT name, format, price
FROM products
WHERE video_resolution = '1080p';
id select_type table ref possible_keys
1 SIMPLE products ref resolutions

26. Modifying JSON documents

27. Example #14: insert field
UPDATE products
SET attr = JSON_INSERT(attr, '$.disks', 1)
WHERE id = 1;
name format price disks
Aliens Blu-ray 13.99 1
SELECT name, format, price,
JSON_VALUE(attr, '$.disks') AS disks
FROM products
WHERE type = 'M';

28. Example #15: insert array
UPDATE products
SET attr =JSON_INSERT(attr, '$.languages',
JSON_ARRAY('English', 'French'))
WHERE id = 1;
SELECT name, format, price,
JSON_QUERY(attr, '$.languages')
AS languages
FROM products
WHERE type = 'M';
name format price languages
Aliens Blu-ray 13.99 [
"English",
"French"
]

29. Example #16: add array element
UPDATE products
SET attr = JSON_ARRAY_APPEND(attr,
'$.languages', 'Spanish')
WHERE id = 1;
SELECT name, format, price,
JSON_QUERY(attr, '$.languages')
AS languages
FROM products
WHERE type = 'M';
name format price languages
Aliens Blu-ray 13.99 [
"English",
"French",
"Spanish"
]

30. Example #17: remove array element
UPDATE products
SET attr = JSON_REMOVE(attr,
'$.languages[0]')
WHERE id = 1;
SELECT name, format, price,
JSON_QUERY(attr, '$.languages')
AS languages
FROM products
WHERE type = 'M';
name format price languages
Aliens Blu-ray 13.99 [
"French",
"Spanish"
]

31. Creating JSON documents
from relational data

32. Example #18: relational to JSON (new)
SELECT JSON_OBJECT('name', name, 'format', format, 'price', price) AS data
FROM products
WHERE type = 'M';
data
{
"name": "Tron",
"format": "Blu-ray",
"price": 29.99
}

33. Example #19: relational + JSON (new)
SELECT JSON_OBJECT('name', name, 'format', format, 'price', price, 'resolution',
JSON_VALUE(attr, '$.video.resolution')) AS data
FROM products
WHERE type = 'M';
data
{
"name": "Aliens",
"format": "Blu-ray",
"price": 13.99,
"resolution": "1080p"
}

34. Example #20: relational + JSON (merge)
SELECT JSON_MERGE(
JSON_OBJECT(
'name', name,
'format', format),
attr) AS data
FROM products
WHERE type = 'M';
data
{
"name": "Tron",
"format": "Blu-ray",
"video": {
"resolution": "1080p",
"aspectRatio": "1.85:1"},
"cuts": [
{"name": "Theatrical", "runtime": 138},
{"name": "Special Edition", "runtime": 155}],
"audio": [
"DTS HD",
"Dolby Surround"]}

35. Enforcing data integrity with
JSON documents

36. Example #21: constraints (1/2)
ALTER TABLE products ADD CONSTRAINT check_attr
CHECK (
type != 'M' or (type = 'M' and
JSON_TYPE(JSON_QUERY(attr, '$.video')) = 'OBJECT' and
JSON_TYPE(JSON_QUERY(attr, '$.cuts')) = 'ARRAY' and
JSON_TYPE(JSON_QUERY(attr, '$.audio')) = 'ARRAY' and
JSON_TYPE(JSON_VALUE(attr, '$.disks')) = 'INTEGER' and
JSON_EXISTS(attr, '$.video.resolution') = 1 and
JSON_EXISTS(attr, '$.video.aspectRatio') = 1 and
JSON_LENGTH(JSON_QUERY(attr, '$.cuts')) > 0 and
JSON_LENGTH(JSON_QUERY(attr, '$.audio')) > 0));

37. Example #21: constraints (1/2)
INSERT INTO products (type, name, format, price, attr) VALUES
('M', 'Tron', 'Blu-ray', 29.99, '{"video": {"aspectRatio": "2.21:1"}, "cuts": [{"name": "Theatrical",
"runtime": 96}], "audio": ["DTS HD", "Dolby Digital"], "disks": 1}');
ERROR 4025 (23000): CONSTRAINT `check_attr` failed for `test`.`products`
no resolution field!

38. Example #22: constraints (2/2)
INSERT INTO products (type, name, format, price, attr) VALUES
('M', 'Tron', 'Blu-ray', 29.99, '{"video": {"resolution": "1080p", "aspectRatio": "2.21:1"}, "cuts":
[{"name": "Theatrical", "runtime": 96}], "audio": ["DTS HD", "Dolby Digital"], "disks": "one"}');
ERROR 4038 (HY000): Syntax error in JSON text in argument 1 to function 'json_type' at
position 1
"one" is not a number!

39. One more thing…

40. Example: identifying movie attributes
id name price attributes
1 Aliens 17.99 {"video": {
"resolution": "1080p",
"aspectRatio": "2.35:1",}}
2 Event Horizon 9.75 {"video": {
"resolution": "1080p",
"aspectRatio": "2.34:1"}}
3 Pandorum 11.49 {"video": {
"resolution": "1080p",
"aspectRatio": "2.35:1"}}
4 Sunshine 9.99 {"video": {
"resolution": "1080p",
"aspectRatio": "2.38:1"}}

41. When the attributes have been defined,
you can create “virtual” columns
id name price resolution ratio attributes
1 Aliens 17.99 1080p 2.35:1 {"video": {
"resolution": "1080p",
"aspectRatio": "2.35:1",}}
2 Event Horizon 9.75 1080p 2.34:1 {"video": {
"resolution": "1080p",
"aspectRatio": "2.34:1"}}
3 Pandorum 11.49 1080p 2.35:1 {"video": {
"resolution": "1080p",
"aspectRatio": "2.35:1"}}
4 Sunshine 9.99 1080p 2.38:1 {"video": {
"resolution": "1080p",
"aspectRatio": "2.38:1"}}

42. Until there is a new attribute…
id name price resolution ratio attributes
1 Aliens 17.99 1080p 2.35:1 {"video": {
"resolution": "1080p",
"aspectRatio": "2.35:1",}}
2 Event Horizon 9.75 1080p 2.34:1 {"video": {
"resolution": "1080p",
"aspectRatio": "2.34:1"}}
3 Pandorum 11.49 1080p 2.35:1 {"video": {
"resolution": "1080p",
"aspectRatio": "2.35:1"}}
4 Sunshine 9.99 1080p 2.38:1 {"video": {
"resolution": "1080p",
"aspectRatio": "2.38:1"}}
5 The Darkest Hour 34.99 1080p 2.40:1 {"video": {
"resolution": "1080p",
"aspectRatio": "2.40:1",
"3d": true}}

43. But, you can create “virtual” columns
using functions
id name price resolution ratio 3d attributes
4 Sunshine 9.99 1080p 2.38:1 false {"video": {
"resolution": "1080p",
"aspectRatio": "2.38:1"}}
5 The Darkest Hour 34.99 1080p 2.40:1 true {"video": {
"resolution": "1080p",
"aspectRatio": "2.40:1",
"3d": true}}
IFNULL(JSON_VALUE(attr, '$.video.3d'), false)

44. Questions?

45. Check out the JSON microsite:
mariadb.com/database/topics/json
Check out the JSON white paper:
goo.gl/sw34cx
Try it out!
What’s next?

46. Thank you!