3. About LINE
•Messaging service
•More than 200 million active users1 in countries with top market
share like Japan, Taiwan, Thailand and Indonesia
•Many family services
•News
•Music
•LIVE (Video streaming)
1 As of June 2017. Sum of 4 countries: Japan, Taiwan, Thailand and Indonesia.
4. Scale Metrics
Messages delivered 25 billion per day
Peak time Messages sent
(new year greetings)
420K per second
Total API requests per day 80 billion per day
Number of event records published > 150 billion per day
Accumulated analytics data 40PB
6. LEGY
• LINE Event Delivery Gateway
• API Gateway/Reverse Proxy
• Written in Erlang
• Deployed to many data centers all over the world
• Features focused on needs of implementing a messaging service
• Zero latency code hot swapping w/o closing client connections
• Durability thanks to Erlang process and message passing
• Single instance holds 100K ~ connection per instance =>
huge impact by single instance failure
7. talk-server
• Java based web application server
• Implements most of messaging functionality + some other
features
• Java8 + Spring + Thrift RPC + Tomcat8
8. Datastore with Redis and
HBase
• LINE’s hybrid datastore =
Redis(in-memory DB, home-
brew clustering) +
HBase(persistent distributed
key-value store)
• Cascading failure handling
• Async write in app
• Async write from background
task processor
• Data correction batch
Primary/
Backup
talk-server
Cache/
Primary
Dual write
9. When user sends a
message
LEGY
LEGY
talk-server
Storage
1. Find nearest LEGY
2. sendMessage(“Bob”, “Hello!”)
3. Proxy request
4. Write to storage
talk-server
X. fetchOps()
6. Proxy request
7. Read message
8. Return fetchOps() with message
5. Notify message arrival
Alice
Bob
10. Communication between
internal systems
• Communication for queuing, transactional
updates:
• Query authentication/permission
• Synchronous updates
• Communication for data synchronization, update
notification:
• Notify user’s relationship update
• Synchronize data update with another service
talk-server
Auth
Analytics
Another
Service
HTTP/REST/RPC
12. • Asynchronous HTTP/2 RPC/REST client/server library
• Built on top of Java8, Netty and RPC frameworks
• Open sourced: https://github.com/line/armeria
• Find more details: https://speakerdeck.com/trustin/
armeria-lines-next-generation-rpc-layer
Armeria
13. DocService
• Quick API test call + Documentation
sb.serviceUnder("/docs", new DocServiceBuilder()
.exampleHttpHeaders(HttpHeaders.of(AUTHORIZATION, "bearer b03c4fed1a"))
.exampleRequest(new HelloService.hello_args("Armeria"))
.build());
14. Other features
• Client side load balancer
• Retrieve server list from pluggable centralized
configuration directory and use for load balancing
• Structured logging
• Supports logging request/response including payload
of RPC
• Supports external system as an output such as Apache
Kafka
15. Armeria
• Most of new services are built on top of Armeria
• and some old projects are/will be rewritten based on it
• Sophisticated yet minimum LINE’s “best practices”
• PRs of line/armeria would be worth visiting for you
• https://github.com/line/armeria/pulls
• You will see how LINE developers are working for
development
16. Apache Kafka
• A distributed streaming platform
• (narrow sense) A distributed persistent message queue
which supports Pub-Sub model
• Built-in load distribution
• Built-in fail-over on both server(broker) and client
17. How it works
Producer
Brokers
Consumer
Topic
Topic
Consumer
Consumer
Producer
AuthEvent event = AuthEvent.newBuilder()
.setUserId(123)
.setEventType(AuthEventType.REGISTER)
.build();
producer.send(new
ProducerRecord(“events", userId, event));
consumer = new KafkaConsumer("group.id" ->
"group-A");
consumer.subscribe("events");
consumer.poll(100)…
// => Record(key=123, value=...)
21. However…
• OSS usually isn’t a perfect solution
• We add/fix what we need
• Typically we need more than others, because of our
scale and challenging usages
• https://issues.apache.org/jira/browse/KAFKA-4614
• https://issues.apache.org/jira/browse/KAFKA-4024
23. Sometimes we share
• Kafka Summit SF 2017
• One Day, One Data Hub, 100
Billion Messages: Kafka at
LINE
• https://youtu.be/
X1zwbmLYPZg
24. but we have more things to
do
• Performance improvement to enable true stability
• Kafka based reliable async task processing system
• Multi-DC deployment
25. Summary
• LINE’s server architecture is keep evolving
• Almost everything are “Distributed”
• Software we use:
• Java/Earlang/Redis/HBase/Kafka/Armeria
• We love OSS
• we Use/Understand/Develop/Contribute
• and it’s encouraged
• 6 years since we started our service, still have many challenges
• More stable and durable service architecture
• Dealing with a lot of new feature developments
• There are a lot of chances you make a great contribution to our systems and services,
please come and talk to me!