Elastic Compute Cloud (EC2)
- Choose your OS, Storage, Memory, Network Throughput. And launch and SSH into your server within minutes.
- EC2 is resizable compute capacity. (launch new instances)
EC2 - Instance Profile
- holds a reference to a role (정책을 프로필로 해서 인스턴스에 박음)
EC2 - Placement Groups
-
let you choose the logical placement of your instances to optimize for communication, performance or durability. (free)
-
Clusters : can't be multi-AZ, low latency network performance, high performance computing
-
Partition : spread instances across logical partitions, each partition don't share the underlying hardware with each other, large distributed and replicated workloads
-
Spread : can be multi-AZ, critical instances should be seperate from each other
EC2 - Pricing Model
- On-Demand : low cost & flexible (can't be interrupted), for short-term
- Spot : biggest savings, can handle interruptions, should be terminated by AWS to not get charged
- Reserved : steady-state, Best long-term, can resell, shared between multiple accounts
- Dedicated : Most Expensive, guarantee of isolate hardware(enterprise requirements)
EC2 - AMI (Amazon Machine Image)
-
Provides the information required to launch an instance
-
can create copies of your servers
-
Launch permissions that control which AWS accounts can use the AMI to launch instances.
-
AMIs are Region Specific
EC2 Auto Scaling Groups
- Set scaling rules which will automatically launch additional EC2 instance or shutdown instances to meet current demand
- Contains a collection of EC2 instances that are treated as a group for the purposes of automatic scaling and management.
- Capacity Settings : Min, max, and desired capacity(ideal)
- Health Check Type : if there is a sw or hw issue.
- ELB health check : by pinging an HTTP endpoint
- Target Scaling Policies : if avg cpu utilization exceeds 75%, then add another server
- ELB(Elastic Load Balancer) Integration :
Elastic Load Balancer (ELB)
-
들어오는 데이터를 이제 교통정리해서 뿌려줌 (EC2, lambda 등등)
-
must have at least two AZ
-
cannot go cross-region, only one per region.
-
can be physical hardware or virtual software that accepts incoming traffic to multiple targets.
-
Listeners : Incoming traffic 을 평가함
-
Rules : (no in classic load balancer) listener 들이 평가하는데 따르는 rule
-
Target Groups : (no in CLB)
-
CLB does not allow you to apply rules to listeners
-
Application Load Balancer : is designed to balance HTTP and HTTPS, WAF(Web Application Firewall) can be attached to ALB(Listener, Rules, Target Groups)
-
Network Load Balancer : is designed to balance TCP/UDP, can handle millions of requests per second, can perform Cross-Zone Load Balancing(Listener, Target Groups)
-
Classic Load Balancer : (legacy), can balance HTTP, HTTPS or TCP(not at the same time), will respond with a 504 error(timeout)(Listeners)
-
Sticky Session: specific EC2 instance로 연결, CLB, ALB,(no NLB), useful when specific info is only stored locally on a single instance
-
IPv4 address -> X-Forwared-FOR(XFF)
-
Health Checks : ELB doesn't terminate unhealthy instance, but will just redirect traffic to healthy instances.
-
Cross-zone load balancing : 다른 AZ에서도 나눠서 처리 가능
-
Request Routing : forward of redirect traffic(ALB)
Elastic File System (EFS)
- Scalable, elastic, cloud-native NFS(Network File System) file system.
- is a file storage service for EC2 instances.
- storage capacity grows and shrinks automatically
- your data is stored across multiple AZs within a region
- can mount multiple EC2 instance to a single EFS(only in the same VPC)
Elastic Block Store(EBS)
- virtual hard dist, snapshots are a point-in-time copy of that dist
- volumes exist on EBs, snapshots exist on S3
- snapshots only changes when the last snapshot are moved to S3
- can create AMI from Volumes or snapshot
- EBS very durable
- IOPS : Input/output per second, high I/O = lots of small fast reads and writes
- Throughput : transfer speed(water)
- Bandwidth : total possible speed of data movement along the network(Pipe)
- EBS : is a solution for attaching persistent block storage volumes to an EC2 instance, Volumes are automatically replicated within their AZ to protect from component failure.
- 5 types of EBS : General Purpose, Provisioned IOPS//SSD, Throughput Optimized HDD, Cold HDD, EBS Magnetic//HDD
HDD
-is a magnetic storage that uses rotating platters
- good at wrtiting a continuously amound of data
- not great for writing many small reads and writes
- better for throughput
SSD
- very good frequently reads and writes
- no physical moving parts
Moving volumes
-
From one AZ to another : take a snapshot of the volume -> create an AMI from the snapshot -> launch new EC2 instance in desired AZ
-
From one Region to another : take a snapshot of the volume -> create an AMI from the snapshot -> copy the AMI to another region -> launch a new EC2 instance from the copied AMI
Encrypted Root Volume
- you can encrypt the volume on creation
- Take a snapshot of the unencrypted volume -> create a copy of the snapshot and select the encryption option -> create a new AMI from the snapshot -> launch a new EC2
EBS vs Instance store volumes
- EBS Volumes : durable, can attach to a single EC2 instance, created from EBS snapshot, can start and stop instances, data will persist if your reboot your system, ideal for when you want data to persist.
- Instance Store Volumes : temporary storage, physical attach, is created from a template stored in S3, can't stop instances, only terminate, ideal for temporary backup, such as cache, logs, or other random data
