- On Demand (linux by the second, Windows by the hour)
- No long term commitments
- Spiky, short term work loads
- Development use
- Reserved (1 to 3 years in length)
- Production servers
- Standard RI up to 75% off on-demand costs
- Convertible RI up to 54% off on-demand costs
- Scheduled RI runs in a reserved time window (day, week, month)
- Spot (bid a price)
- Run a low use times
- Need high performance for a limited time
- Dedicated Hosts (Physical servers for you)
- Regulatory needs
- License needs
- Can be hourly priced
- Up to 75% off on-demand
- F1 Field Programmable Gate Array [Analytics]
- I3 High speed storage [DB]
- G3 Graphics [Video Encoding/Streaming]
- H1 High disk throughput [MapReduce]
- T2 Lowest Cost, GP [development/testing]
- D2 Dense Storage [Data Warehouse/Hadoop]
- R4 Memory optimized [Memory Intensive/DB]
- M5 GP [Application Servers]
- C5 Compute Optimized [CPU Intensive applications]
- P3 GPU [Machine Learning/Bit Coin Mining]
- X1 Memory Optimized (hugh amount of ram) [SAP/HANNA]
FightDrMcPx - remember instance types....
Root device volume - boot and image drive
- GP2 General Purpose SSD
- balance of price and performance
- 3 IOPS per GB up to 10,000 IOPS with burst up to 3000 IOPS for volumes above 3334 GB
- Pervisioned IOPS SSD
- Database use
- Need more than 10,000 IOPS
- ST1 Throughput optimized HDD
- cannot be boot volume
- log files
- big data SC1 - Cold HHD
- cannot be boot volume Magnetic (Std)
- legacy
- can boot from
- stored in ~/.aws/config ~/.aws/credentials
- aws configure --profile {profile_name}
- export AWS_PROFILE={profile_name} or setx AWS_PROFILE {profile_name}
- EBS volume must be in same availablity zone as ec2 instance.
- create encrypted root volume by copying root volume and creating an ec2 image from the copied and encrypted root volume.
- (OLTP) SQL Server, MySQL, PostgreSQL, Oracle, Aurora, MariaDB
- (OLAP) Redshift
- Elasticache [Memcached, Redis]
- automated backups, snapshot + transaction logs
- snapshot, done manually, stored until original RDS instance is deleted.
- restored instances have a different url
- encrypted if database is encrypted,
- create an encryted database by copying a snapshot and restoring
- for disaster recovery only - synchronous replication
- for performance - asynchronous replication
- NOT available for Oracle or SQL Server
- 5 read replica for each DB
- can be in different availablity zone or region
- can be promoted to main DB, but breaks all read replication
- store critical data in memory
- redis allows multi-az consistency
- redis is managed like a database, with persistence
- memcached auto scaling like EC2 auto scaling
- read heavy queries
- not for OLAP, use Redshift instead
- simple
- offline database data
- scale out
- advanced data types (lists, hashes, sets)
- pub/sub
- sorting, ranked datasets like leaderboards
- persistence
- multi-az
- max TTL 31536000 seconds (365 days)
- default TTL 86400 seconds (24hrs)
- protect content with signed URLs or signed cookies
- WAF projects you at the application layer, automatically projects you from
- SSL cert, can bring your own and store in ACM or IAM