EC2

Elastic Compute Cloud (EC2)

Provides virtual servers in the AWS cloud.

You can launch one or thousands of instances simultaneously and only pay for what you use.

EC2 Autoscaling

• Allow you to dynamically scale your EC2 capacity up or down automatically according to defined conditions.

• It can also perform health checks on those instances and replace them when they become unhealthy.

Types of Autoscaling

Keep instances in the same level

• You configure the Auto scaling group to maintain the specified number of instances executing.

• EC2 Autoscaling runs periodic integrity verifications on the running intances, and when it finds defective ones it terminates these and start new ones.

Manual scaling

• Most basic scaling way, where you only specify the change in max/min/target capacity in the auto dimention group.

• EC2 Autocaling will just manage the start/finish of intances to keep the capacity up to date.

Scheduled scaling

• It auto scales based on specified date and time.

• Useful when you know when to rise and lower the instance number, based on predictable needs.

Scaling based on demand

• A more advanced way to scale resources.

• Using scaling policies that allows to define parameters that control the scaling process.

• Useful when you don't know when to rise or lower the intance number, since the demand is unpredictable.

  • For instance EC2 Autoscaling can look at CPU or memory load, and it scales up/down based on a threshold.

Predictive scaling

• Uses EC2 Autoscaling with AWS Autoscaling to scale resource in several services.

• Maintains availability and performance at ideal levels.

• Uses predictive and dynamic scaling (proactive and reactive approaches).

Pricing

Instance Types

• General Purpose (T2, M3, M4): Small and mid-size databases, data processing tasks that require additional memory, caching fleets, etc.

• Compute Optimized (C3, C4): High performance front-end fleets, web-servers, batch processing, distributed analytics, high performance science and engineering applications, etc.

• Memory Optimized (X1, R3, R4): High performance databases, distributed memory caches, in-memory analytics, genome assembly and analysis, large deployment of SAP and etc.

• GPU/Accelerated Computing (G3, G2): 3D application streaming, machine learning, video encoding, and other server-side graphics or GPU compite workload.

• Storage Optimized:

  • (I3, I2): NoSQL databases like Cassandra and MongoDB, scale out transactional databases, data warehousing, Hadoop, and cluster file systems.

  • (D2): Massively Parallel Processing (MPP) data warehousing, MapReduce and Hadoop distributed computing, log or data processing applications.

Instance States

• start

• stop (EBS Backed only): Instance is shut down with no instance charges, still charged for EBS volumes.

• Stop-Hibernate (EBS Backed only): Suspend-to-disk, saves RAM to EBS.

• Reboot: Operating system reboot.

• Terminate

Storage Options

Elastic Block Store (EBS)

• Most common.

• Replicated within AZ.

• EBS volumes attached at instance launch are deleted when instance terminated.

• EBS volumes attached to a running instance are not deleted when instance is terminated but are detached with data intact.

Instance Store

• Temporary storage.

• Physically attached to the host server.

• Data NOT LOST when OS is rebooted.

• Data LOST when:

  • Underlying drive fails.

  • Instance is terminated.

• Do not rely on for valuable, long-term data.

• You cannot detach and attach to another instance.

General Purpose SSD (gp2)

• Default choice.

Provisioned IOPS SSD (io1)

• Consistent and low-latency performance.

• I/O intensive applications such as large relational or NoSQL databases.

Cold HDD (sc1)

• Lowest cost per gigabyte.

Throughput Optimized HDD

• Low cost per gigabyte.

• Frequently accessed workloads.

EBS Snapshots

• Point in time backup of EBS volume to Amazon S3.

• Incremental backup.

• Can be copied to other regions or accounts.

EBS Encryption

• AWS Key Management Service. (KMS)

• Data stored at rest encrypted as well as data in transit between EBS and EC2.