Restoring Backups : 다시 복원시킬때는 새로운 instance 가 만들어짐 (endpoint 도)
Multi AZ
Automatic Failover protection : 다른 AZ로 바로 복구
Synchronous replication (highly durable)
Only db engine on primary instance is active
always span two AZ within a single Region
Read Replicas
multiple copies 가능한데, 이것들은 오직 reads 만 허용함. (primary db 의 performance 향상시키기위해)
primary RDS 와 replica 는 asynchronous (병렬적, 동시에 x) 로 작동한다.(highly scalable)
문제가 생기면 manually 백업 instance
can be within an AZ, cross-az, or cross region
Aurora (Severless)
serverless is inexpensive option
Fully Managed and really fast
high-end db, cost-effectiveness
can span multiple regions via Aurora Global DB
infrequent db usage is more ideal
Scaling
Storage is autoscaling
알아서 늘어남 필요한만큼
Avalability
한 az는 2개의 copies 만들수있음
Fault Tolerance and Durability
Aurora Backup and Failover is handled automatically
다른 aws 계정에 스냅샷 공유 가능
Storage is self-healing
Serverless
automatically start up, shut down, and scale capacity up or down
너가 쓴 만큼 자동으로 청구
Redshift
petabyte scale solution for Data Warehousing
Business Intelligence
Online Analytics Processing System
Columnar Storage : important factor in optimizing analytic query performance because it drastically reduces the overall disk I/O requirements and reduces the amount of data you need to load from disk(SQL-like Queries)
can be loaded from S3, EMR, DynamoDB...
병렬처리 가능(Asynchronously)
Data Warehouse
Online Analytical Processing : multiple records at the same time, you save memory because you fetch just the columns of data you need instead of whole rows.
built to store large quantities and enable fast, complex queries
long transaction
Node Types
Dense Compute(dc) : best for high performance, but less storage
Dense Storage(ds) : clusters in which you have a lot of data
Processing
Massively parallel processing(MPP)
automatically distributes
add new nodes easily
Backups
default -> 1day, Retention period can be up to 35 days
different region snapshot replicate available
Security
Encrypted using SSL, KMS
Availability
Single-AZ
To run Multi-AZ, you would have to run multiple RedShift Clusters in different AZs with same inputs.
Snapshots can be restored to a different AZ in the event an outage occurs