발표 영상 - https://www.youtube.com/watch?v=LhXFxjNzHws
AWS re:Invent는 매년 신규 서비스를 통해 클라우드 기술의 미래를 조망하고, 새로운 기술을 익히고 배우는 글로벌 기술 컨퍼런스입니다. 분야별 클라우드 기술 전문가이신 AWS Hero와 함께 개발자, 아키텍트 및 데브옵스 엔지니어, 데이터과학자 및 분석가 등 직무별로 새해 주목할 만한 AWS 신규 서비스를 고객의 입장에서 생생하게 전달해 드리고자 합니다.
16. Container Image Support for AWS Lambda
PA C K A G E A N D D E P L OY F U N C T I O N S A S C O N TA I N E R I M A G E S
Benefit from sub-second autoscaling, high
availability, 140 native service integrations, pay
for use
Use a consistent set of tools for containers and
Lambda-based applications
Deploy large applications with AWS provided or
third party images of up to 10GB (ML, data
analytics, large dependencies)
19. Simple Developer Experience
CREATE FUNCTION AS CONTAINER IMAGE
docker push
Amazon Elastic
Container Registrycontainer image
Upload image to registry
Invoke
Status: ACTIVE
Ready for invoke
Lambda
Function
Container Image
1. Pull image from ECR
2. Optimize image
3. Deploy image to Lambda
CreateFunction
container image
Status: PENDING
AWS Lambda
20. 1. EFS 안써도 된다.
• EFS 원래 목적은 고성능 File system
• Lambda에서는 주로 모자른 Local storage (250mb…)를 보충하기 위해 사용
• Lambda 보다도 더 비싸지는 경우가 잦음
21.
22. 2. 더 쉬운 Serverless로의 Migration
• 기존엔 AWS Lambda로 기존 Application을 옮기려면
상당한 코드를 다시 작성해야함
• Docker를 지원하면서, 기존에 ECS / ECR / Kubernates 등으로 구현한
Application을 Migration 하는게 훨씬 쉬워짐
24. 1 millisecond billing granularity
AW S L A M B D A
In effect starting with December 2020 billing
cycle
Reduced the billing granularity for function
duration from 100ms to 1ms
Functions with short duration will see largest cost
benefits – interactive web calls, data streaming,
and more
25. 1. Microtask 에도 Lambda를 적용
• 실제로 서비스 운영할때는 cache가 완벽하게 걸린 GET요청이 많음
• 이런 경우, 실제 Lambda 사용 시간이 길어야 5~10ms…
• 기존엔 최소 billing이 100ms
28. Amazon Aurora
MySQL and PostgreSQL compatible relational database built for the cloud
Performance and availability of commercial-grade databases at 1/10th the cost
Performance
and scalability
5x throughput of MySQL
3x throughput of PostgreSQL
15 read replicas
Scale out reads and writes across
multiple data centers
Fully managed
Managed by RDS:
no hardware provisioning,
software patching, setup,
configuration, or backups
Availability
and durability
Fault-tolerant self-healing storage
Six copies of data across three AZs
Continuous backup to S3
Single Global database with cross-
region replication
Network isolation
Encryption at rest/transit
Highly secure
31. Introducing Amazon Aurora Serverless v2
Highly Scalable Highly available
Power your business critical
workloads with the full breadth
of Aurora features, including
Multi-AZ, read replicas, and
Global Database.
Scale capacity in fine-grained
increments and save up to 90% of
costs vs provisioning capacity for
peak load.
Save up to 90%
PREVIEW
NEW
Scale instantly, from hundreds to
hundreds-of-thousands of
transactions, in a fraction of a
second.
Support the most demanding applications without managing database capacity.
Aurora MySQL – Preview today, Aurora PostgreSQL – Preview early 2021
32. What does this mean?
Serverless v2 is built for high performance applications;
Compared to v1, v2 has full Aurora functionality (multi-AZ, Read/
Replica, and global Database), more scale, faster performance
Enables applications that are optimal for serverless
• Apps that need to scale quickly
• Apps that handling inputs quickly
Competitive differentiator: There is no truly serverless database
from either GCP or Microsoft