The document discusses MongoDB transactions and concurrency. It provides code examples of how to perform transactions in MongoDB using logical sessions, including inserting a document into a collection and updating related documents in another collection atomically. It also discusses some of the features and timeline for implementing distributed transactions in sharded MongoDB clusters.

1. Eliot Horowitz, CTO & Co-Founder
@eliothorowitz

3. doc = { line_items : [ { item : 5,
quantity : 6 }, … ] }
db.orders.insertOne( doc );
for x in doc.line_items:
db.inventory.update( { _id : x.item },
{ $inc : { number : -1 * x.qty } })

4. doc = { line_items : [ { item : 5,
quantity : 6 }, … ] }
db.orders.insertOne( doc );
for x in doc.line_items:
db.inventory.update( { _id : x.item },
{ $inc : { number : -1 * x.qty } })

5. doc = { line_items : [ { item : 5,
quantity : 6 }, … ] }
db.orders.insertOne( doc );
for x in doc.line_items:
db.inventory.update( { _id : x.item },
{ $inc : { number : -1 * x.qty } })

6. session.start_transaction()
doc = { line_items : [ { item : 5, quantity: 6 }, … ] }
db.orders.insertOne( doc, session=session );
for x in doc.line_items:
db.inventory.update( { _id : x.item } ,
{ $inc : { number : -1 * x.qty } },
session=session )
session.commit_transaction()

7. Familiar
Conversational
Performant
Familiar
Conversational
Performant

8. Multi-document ACID transactions
4.0 4.2

9. Marathon to transactions: the homestretch
MongoDB 3.0 MongoDB 3.2 MongoDB 3.4 MongoDB 3.6 MongoDB 4.0
Single Replica Set
Transactions
MongoDB 4.2
Sharded
Transactions
New Storage engine
(WiredTiger)
Enhanced replication
protocol: stricter
consistency & durability
Shard membership
awareness
Consistent secondary
reads in sharded clusters
Storage support for
prepared transactions
Transaction – compatible
chunk migration
WiredTiger default storage
engine
Logical sessions
Make catalog
timestamp-aware
More extensive
Wired Tiger repair
Config server
manageability
improvements
Retryable writes
Replica set
point-in-time reads
Transaction manager
Read concern "majority" Causal Consistency
Recoverable rollback via
WT checkpoints
Global point-in-time reads
Cluster-wide logical clock Recover to a timestamp
Oplog applier prepare
support for transactions
Storage API to changes to
use timestamps
Sharded catalog
improvements
Collection catalog
versioning
Make collection
drops two phrase
UUIDs in sharding
Fast in-place updates to
large documents in WT
In Progress
Done
Sharded Transaction Feature

12. {
order_id,
customer_id
}
{
line_item_id,
order_id
}
session.start_transaction()
db.line_items.update(
{ order_id : my_order_id } ,
{ $mul : { price : discout } },
sesison = my_session)
db.orders.update(
{ order_id : my_order_id } ,
{ $mul : { order_total : discount
} },
session = my_session )
session.end_transaction()

13. {
order_id,
customer_id
}
{
line_item_id,
order_id
}
session.start_transaction()
db.line_items.update(
{ order_id : my_order_id } ,
{ $mul : { price : discout } },
sesison = my_session)
db.orders.update(
{ order_id : my_order_id } ,
{ $mul : { order_total : discount
} },
session = my_session )
session.end_transaction()

14. {
order_id,
customer_id,
line_iems: [
{
name : ,
price : ,
qty : ,
}
]
}

15. db.orders.update(
{ order_id : my_order_id },
{ $mul :
{ order_total :
discount,
"line_items.$[].price" :
discount }
}
)
{
order_id,
customer_id,
line_iems: [
{
name : ,
price : ,
qty : ,
}
]
}

16. Distinct business objects
Rollups / Views
Coming soon:
change shard keys

18. 4.0

21. Relationships
{
order_id,
line_items: [ … ],
}
{
product_id,
inventory
}
REFERENCE

22. Relationships
EMBEDDED
{
order_id,
line_items: [
{ product_id,
qty,
subtotal
}
]
}
{
order_id,
line_items: [ … ],
}
{
product_id,
inventory
}
REFERENCE

25. MongoDB is the best
database for relationships

26. Natural
Dynamic schema
Query language
Native drivers
Document model

27. High Availability
Scale-out
Workload isolation
Zoned sharding
Distributed systems

28. Consistency
Portability
Security
Runs anywhere

29. The best database for modern
applications
Distributed Systems Runs EverywhereDocument Model
Any use case, any platform

31. db.products.find().limit(1).pretty()
{
"_id" : ObjectId("5b116f1b3e1c824cb7950480"),
"id" : "298157062",
"name" : "Uptown Laptop TSA-Friendly Backpack",
"sku" : "MON25232000",
"image" : {
"large" : "/images/products/bags/large_backpack.jpg",
"thumb" : "/images/products/bags/thumb_backpack.jpg"
},
"color" : "black",
"inventory" : 100,
"category" : [
"products",
"bags",
"backpack",
"travel"
],
"price" : NumberDecimal("1233423"),
"reviews" : [ ],
"likes" : 101,
"tweets" : 13,
"type" : "backpack"

35. DEMO
Atlas Data Explorer
Jonathan Balsano

38. grep -o "[0-9]+ms" <filename> |
sort -n |
tail -n 20 |
uniq |
xargs -n 1 -I x grep x <filename>

39. Composable
Debuggable
Powerful

40. db.logs.aggregate( [
{ $sort : { time : -1 } },
{ $limit : 20 },
{ $group : { _id : "$queryShape",
count : { $sum : 1 } }
}
] )

41. Date
Manipulation
String
Manipulation
Type
Casts

47. DEMO
Aggregation Pipeline Builder
Asya Kamsky

50. db.products.find( {} )
https://api.swagstore.com/products

51. Categories:
S shirts
 hats
 socks
https://api.swagstore.com/products?cat=shirts
db.products.find( { category : "shirts" } )

52. Categories:
S shirts
 hats
 socks
Brands:
 IBM
S MongoDB
https://api.swagstore.com/products?cat=shirts&brand=mongodb
db.products.find( { category : "shirts",
brand : "mongodb"
} )

53. Categories:
S shirts
 hats
 socks
Brands:
 IBM
S MongoDB
Price:
…api.swagstore.com/products?cat=shirts&brand=mongodb&priceMin=10&priceMax=120
$Min $Max
G
O
db.products.find( { category : "shirts",
brand : "mongodb",
price : { $lt : 120, $gt: 10 }
} )

54. Categories:
S shirts
 hats
 socks
Brands:
 IBM
S MongoDB
Price:
Price: Low to High
…ducts?cat=shirts&brand=mongodb&priceMin=10&priceMax=120&sort=price&order=asc
$Min $Max
G
O
db.products.find( { category : "shirts",
brand : "mongodb",
price : { $lt : 120, $gt: 10 }
} ).sort( { price : 1 } )

55. Categories:
S shirts
 hats
 socks
Brands:
 IBM
S MongoDB
Price:
Rating:

…mongodb&priceMin=10&priceMax=120&sort=price&order=asc&sort2=rating&order2=desc
$Min $Max
G
O
db.products.find( { category : "shirts",
brand : "mongodb",
price : { $lt : 120, $gt: 10 }
} ).sort( { price : 1, rating : -1 } )
?!

56. db.products.find( { category : "shirts",
brand : "mongodb",
price : { $lt : 120, $gt: 10 }
} ).sort( { price : 1, rating : -1 } )
Categories:
S shirts
 hats
 socks
Brands:
 IBM
S MongoDB
Price:
Rating:

…mongodb&priceMin=10&priceMax=120&sort=price&order=asc&sort2=rating&order2=desc
$Min $Max
G
O
MQL
REST

57. db.products.find( { category : "shirts",
brand : "mongodb",
price : { $lt : 120, $gt: 10 }
} ).sort( { price : 1, rating : -1 } )
Categories:
S shirts
 hats
 socks
Brands:
 IBM
S MongoDB
Price:
Rating:

…mongodb&priceMin=10&priceMax=120&sort=price&order=asc&sort2=rating&order2=desc
$Min $Max
G
O
REST (Shitty Query
Language)

58. Categories:
S shirts
 hats
 socks
Brands:
 IBM
S MongoDB
Price:
Rating:

$Min $Max
G
O
db.products.find( { category : "shirts",
brand : "mongodb",
price : { $lt : 120, $gt: 10 }
} ).sort( { price : 1, rating : -1 } )

59. The serverless platform
for building modern
applications

60. QueryAnywhere Functions

62. Atlas Stitch Native SDK
Auth
Rules

63. Fully integrated
authentication
Declarative rules
Full MongoDB query
language

65. DEMO
Stitch QueryAnywhere
Eric Daniels

72. QueryAnywhere Functions

75. BI 2.5 ChartsODBC Driver

77. DEMO
ODBC & Charts
Grigori Melnik

78. Download a public beta from
mongodb.com/charts
Play, learn, provide your feedback! Happy charting!

79. Atlas CRUD
Aggregation Pipeline
Builder
Stitch QueryAnywhere
Analytics

81. High
Availability
Scale
Out
Workload
Isolation
Zoned
Sharding

82. Latency
Regulation

87. DEMO
Atlas Global Clusters
Andrew Davidson

91. # Subscribe to change notifications in a collection.
cursor = client.my_db.my_collection.watch([
{'$match': {'operationType':
{'$in': [
'insert',
'replace' ]
}
}},
{'$match': {'fullDocument.n':
{'$gte': 1}}}
])
# Loops forever.
for change in cursor:
print(change['fullDocument'])

93. QueryAnywhere Functions Stitch Triggers

99. DEMO
Stitch DB Triggers
Jane Fine

101. QueryAnywhere Functions Stitch Triggers

102. Global Clusters
Stitch Triggers

104. M0

106. No lock-in
Regions
Services

109. DEMO
Ops Manager / Kubernetes
James Broadhead

111. final Barcode barcode =
data.getParcelableExtra(BarcodeActivity.BARCODE_FIELD);
final Document query = new Document("barcodeId",
barcode.displayValue);
final Document update = new Document("$inc",
new Document("itemsScanned", 1));
final Document after = itemsColl.findOneAndUpdate(
query,
update,
new FindOneAndUpdateOptions().
upsert(true).
returnDocument(ReturnDocument.AFTER));

112. final Barcode barcode =
data.getParcelableExtra(BarcodeActivity.BARCODE_FIELD);
final Document query = new Document("barcodeId",
barcode.displayValue);
final Document update = new Document("$inc",
new Document("itemsScanned", 1));
final Document after = itemsColl.findOneAndUpdate(
query,
update,
new FindOneAndUpdateOptions().
upsert(true).
returnDocument(ReturnDocument.AFTER));

114. DEMO
MongoDB Mobile
Andrew Morrow

115. Collection.sync(
documentID,
conflictResolver,
eventListener);
Mobile sync syntax specifics

116. Mobile sync architecture
Stitch Mobile: Sync
Change
Streams
Atlas
Stitch

117. Mobile SyncQueryAnywhere Functions Stitch Triggers

118. Atlas free tier on
Google Cloud Platform
Cross-cloud
Kubernetes
Mobile Sync

119. GA
QueryAnywhere
Functions
Triggers
Mobile Sync

120. CRUD
Global Clusters
Free Tier on GCP
Enterprise security
features
Sharding import

121. Query enhancements
Free monitoring
Multi-document ACID
transactions

122. Best database for modern applications

123. M040 course call-out

125. Innovation Award Winners
Global Go-to-Market
Partner of the Year The Enterprise Launch Fast Scale
Data-Driven Business Customer Experience Healthcare Internet of Things
Delivery Partner of the Year
The William Zola Award for
Community Excellence Atlas Savvy Startup