AWS Database Flashcards
Amazon RDS
- Managed RDMS Service
- Mysql, postgres, oracle, sqlserver, mariadb
- Encrypted during creation only
AWS Dynamo DB
- No SQL DB
- Sync replicated across 3 AZs in a Region
AWS Dynamo DB Read Models
- Eventual - Default
- Strongly
AWS Dynamo DB Global Tables
- read and write locally
- Allows Cross Region Replication
- Multi-Region and Multi-Master
- Specify the Regions - AWS does it all for you
Dynamo DB Accelerator (DAX)
- Fully managed & Highly available Cache
- In memory
- In proves performance on both reads and writes
Dynamo DB Transactions
- Used to achieve ACID compliance
-
AWS RDS Multi-AZ
- Used for DR option
- Highly Durable
- Only Primary is active
- Auto Failover - 1 to 2 mins
- Sync Replication
- Backup on Standby
- maintenance on standby
- Upgrade Engine on Primary/Standby together with downtime
Dynamo DB Streams
- Ordered Sequence of item level modifications
- Stored in a Log for 24 hours
Aurora Global Database
- For global applications
- Cross Region Cluster with Read scaling
- Can remove secondary and promote
- uses storage replication
- 1 primary regions / 5 secondary regions
- 16 Read Replicas per regions
- RTO < 1 min
Aurora Serverless
- On-Demand Serverless DB
- Does not support Read Replicas
- Infrequent, intermittent, or unpredictable workloads
- Can migrate between Standard and Serverless via console
Aurora Serverless Use Cases
- Infrequently used apps
- new applications
- variable workloads
- unpredictable workloads
- development and test
- multi-tenant apps
Aurora MySQL Replicas
- Cross Regions cluster with read scalling
- Failover target with mins of data loss
- Up to 5
- Performance impact is high
Aurora Multi-Master
- High Avail and ACID transactions in a single Region
- Scales our writes within a region
- All Nodes are reads and writes
- Open a connection to another writer on failure
RDS Read Replicas
- Used for Heavy Read workloads
- same AZ, cross AZ, or cross region - up to 5
- Async Replication
- Each has a unique endpoint
- Can promote to Standalone instance
- Not for MS SQL Server
RDS Multi-AZ vs Read-Replicas
- Sync vs Async
- Only Primary is active vs all replicas are active
- Backups from Standby vs no backups by default
- Always spans 2 AZs vs Same, Different, or Diff region
- Auto failover vs can be promoted to Standalone
Aurora Machine Learnign
- Add ML-Based predictions to your apps via sql
- Simple integration between DB and ML
Amazon Aurora
- Mysql or prostres compatible RDBMS for Cloud
- Distributed fault tolerant self-healing Storage sytems
- 6 copies in 3 AZs
- 4 Writes / 3 Reads
AWS Redshift
- SQL Based Data warehouse for OLAP UseCases
- complex queries against PBs of structured data
How can RDS Read Replicas increase availablity
- Multi-AZ
Dynamo DB read replicas
- There are no read replicas
RDS Database Failover
- Handled automatically by RDS
- CNAME is changed to point to Standby
- 1 to 2 mins
Aurora Replicas
- In region read scaling and failover targets
- Up to 15
- Peformance impact is low
- can act as auto failover with no data loss
- auto failover
Use as a web session Store
memcached
Use for Database Caching of RDS
Memcached
Used for Leaderboards for your mobile apps
Redis
MEMCACHED
- Simple Cache
- No Encryption
- No Compliance
- Multi-Threads
- Not highly available
Redis
- complex cache
- can be encrypted
- can support compliance
- can support failover
- Data partitioning - cluster mode only
Memcached High Availability
- no replication
- Using multiple nodes in each shard
- multiple nodes across AZs
Redis High Availablity
- Use multiple nodes in each shard across multiple AZs
- Enable Multi-AZ on the replication group to permit auto failover
- Schedule regular backups
What is the DB solution for Aurora MySQL DR strategy, cross region, with RTO of 10 minuets and RPO of 5 mins?
Aurora Global Database - Cross region
MySQL Replicas - will work cross region but the RPO/RTO is longer
Used for complex queries against an S3 Data Lake
Redshift spectrum
Allows a password to be used to access a Redis DB
Redis AUTH command
Steps to migrate a large DB using Snowball Edge
- Use the SCT to extract and load to snowball
- Ship the edge device
- Edge loads to S3
- DMS to migrate the data to a data store
Can efficiently query and retrieve structured and unstructured data from files in S3 without having to load the data into tables.
Redshift Spectrum
Type of replication used from primary to standby in RDS multi AZ?
Sync
Type of replication used from primary to read replica in RDS?
Async
Does RDS have multi-Master capability?
No. That’s an aurora feature
Dynamo DB Best practice
Store more freq and less freq accessed data in diff tablets
Store Objects larger than 400kb in s3 with pointers in dynamo
Time it takes for an RDS restore
5 mins
Does the RDS restore overwrite the DB?
No
What security group does the restored RDS DB use
Default SG
Are RDS Read Replicas only in a single region
No… they can be in multiple regions
Dynamo DB Auto Scaling
Amazon DynamoDB auto scaling uses the AWS Application Auto Scaling service to dynamically adjust provisioned throughput capacity on your behalf, in response to actual traffic patterns.
what configurations are needed when an on-prem app needs AWS RDS DB access.
- Security group to allow access from the on prem-ip
- Make RDS public so the on-prem app can see it
Can you scale DynamoDB using EC2 Auto Scaling groups
No… uses application scaling
A Read Replica of an Amazon RDS encrypted instance is also encrypted using the same key as the master instance when both are in the same Region. When in different Regions, a different key can be used.
true
When using the provisioned capacity pricing model DynamoDB does not automatically scale.
true
DynamoDB can automatically scale when using the new on-demand capacity mode
true
Can you enable storage auto-scaling for RDS MySQL
Yes
