---
title: '03 EC2 Storage: EBS & EFS'
disqus: hackmd
---

:::info
AWS Certified Developer Associate DVA-C01
:::

03 EC2 Storage: EBS & EFS
===

<style>
img{
/* border: 2px solid red; */
margin-left: auto;
margin-right: auto;
width: 80%;
display: block;
}
</style>

## Table of Contents

[TOC]

Elastic Block Store (EBS)
---
- ec2 machine loses root vol when manually terminated
- unexpected terminations might happen occasionally (aws will email you)
- need way to store instance data somewhr
- elastic block store (ebs) - network drive u can attach to instances when they run
- allows instances to persist data

### EBS Volume
- is network drive (not phy drive)
- uses network to comm with instances
- might have latency
- can be detached from ec2 instance & attached to another one quickly
- locked to AZ
- ebs vol in us-east-1a cannot be attached to us-east1b
- to move vol across, need to snapshot
- has provisioned capacity
- size in GBs & IOPS
- get billed for all provisioned capacity
- provisioned is diff from use
- Eg. provision 1tb but only use 500gb, still pay for 1tb
- can inc capacity of drive over time


### EBS Volume Types
- 4 types
- GP2 (SSD)
- general purpose ssd vol that balances price & perf for wide variety of workloads
- IO1 (SSD)
- highest perf ssd vol for mission-crit low-latency/high-throughput workloads
- ST1 (HDD)
- low cost hdd vol designed for frequently accessed, throughput-intensive workloads
- SC1 (HDD)
- lowest cost hdd vol designed for less frequently accessed workloads
- characterised in size/throughput/IOPS (I/O Ops per sec)
- when in doubt, consult aws documentation
- constantly updating
- only GP2 and IO1 can be used as boot vols

#### Use Cases - GP2
- recommended for most workloads
- system boot vols
- virtual desktops
- low latency interactive apps
- development & test envs
- 1 GB - 16 TB
- small gp2 vols can burst IOPS (input/output operations per second) to 3000
- max IOPS is 16k
- 3 IOPS per gb
- at 5334gb we at max IOPS

- IOPS dec/inc as size dec/inc

#### Use Cases - IO1
- crit business apps that require sustained IOPS perf or more than 16k IOPS per vol (gp2 limit)
- large db workloads like
- mongodb, cassandra, microsoft sql server, mysql, postgresql, oracle
- 4 GB - 16 TB
- IOPS provisioned (PIOPS) - min 100 to max 64k (nitro instances)
- else max 32k (other instances)
- max ratio of provisioned IOPS to requested vol in GiB is 50:1
- anything above that rejected
- Eg. 10gb = 500 max IOPS

#### Use Cases - ST1
- streaming workloads requiring constant, fast throughput at low price
- big data, data warehouses, log processing
- apache kafka
- cannot be boot vol
- 500 GB - 16 TB
- max IOPS is 500
- max throughput of 500mb/s - can burst

#### Use Cases - SC1
- throughput-oriented storage for large vols of data infrequently accessed
- scenarios whr lowest storage cost important
- cannot be boot vol
- 500 GB - 16 TB
- max IOPS is 250
- is fixed
- max throughput 250mb/s - can burst

#### Summary of Vol Types
- for gp2,
- 3000 iops per tb

### EBS VS Instance Store
- some instances dont come with root ebs vols
- instead come with __instance store__ (= ephemeral storage)
- ephemeral = lasting for a very short time
- instance store is phy attached to machine
- whereas ebs is network drive
- pros (cuz no network)
- better i/o perf
- ebs gp2 have max iops of 16k, io1 have 64k
- good for buffer/cache/scratch data/temp content
- data survives reboots
- cons
- on stop/termination, instance store lost
- cant resize instance store
- backups must be operated by user
- ask yourself, r u okay losing your data/is your data ephemeral?

#### Local EC2 Instance Store
- phy disk attached to phy server whr your ec2 is
- very high IOPS because phy attached
- network bandwidth limits iops
- disks up to 7.5tb (can change over time), stripped to reach 30tb (can change over time)
- number can change over time
- BUT once setup disk in local instance store, cannot change its size??
- block storage
- just like ebs
- can have file system
- cannot be increased in size
- risk of data loss if hardware fails


### Console
- when launching new instance, add new ebs volume
- free tier can get up to 30gb of ebs gp2 ssd

- view your volumes in the volumes tab
- 8gb is root, 2gb is ebs
- `lsblk` to list mounted volumes in terminal
- https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/ebs-using-volumes.html
- follow this guide to make vol avail for use
- must create and mount new file system since new vols are raw block devices


Elastic File System
---
- elastic file system (efs) - managed network file system (NFS) that can be mounted on many ec2
- works with ec2 instances in multi-az
- unlike ebs which is locked to az
- highly available, scalable, expensive (3x gp2) and pay per user
- might be cheaper than ebs if dont store too much data
- use cases
- content management
- web serving
- data sharing
- wordpress
- uses NFSv4.1 protocol
- uses security grp to control access to EFS
- compatible with linux based AMI
- not windows! windows cannot mount efs onto their file system
- encryption at rest using __key management service (KMS)__
- portable OS interface (POSIX) file system (linux) that has a standard file API
- EFS only used for POSIX file systems (basically in linux)
- file system scales automatically, pay per use and no capacity planning
- hence very easy to use


### Performance & Storage Classes
- efs scale
- 1000s of concurrent NFS clients, 10gbps throughput
- grow to petabyte-scale nfs automatically
- huge
- performance mode (set at efs creation time)
- general purpose (default)
- latency-sensitive use cases
- Eg. webserver, content management systems (CMS) etc.
- max i/o
- higher latency/throughput
- highly parallel
- Eg. big data, media processing etc.
- storage tiers - has lifecycle management feature (move file after N days)
- standard
- for frequently accessed files
- infrequent access (EFS-IA)
- cost to retrieve files
- lower price to store

### Console
- need search efs console from all aws services

- click customise to customise efs settings

- lifecycle management
- move files to EFS infrequent access storage class if not accessed for xx days
- save costs

- throughput mode
- provisioned allows u to choose throughput from 1-1024mbps
- bursting depends on size

- need define sec grp for ea az


- attach efs to instance by clicking attach btn at efs console
- can mount via dns or ip
- copy paste the commands into your ec2
- need install amazon efs utils package
- https://docs.aws.amazon.com/efs/latest/ug/installing-amazon-efs-utils.html

EBS VS EFS
---
### EBS
- ebs volumes
- can be attached to only 1 instance at a time
- locked to az lvl
- gp2 - i/o increases if disk size inc
- io1 - can inc io independently
- to migrate ebs vol across az
- take snapshot
- resotre snapshot to another az
- ebs backups use io so shouldnt run them while app is handling a lot of traffic
- else will get perf issues
- root ebs vols of instances get terminated by default if ec2 instance gets terminated
- can disable this


### EFS
- can mount to 100s of instances across az
- efs share website files
- Eg. wordpress
- only for linux (POSIX)
- has higher price pt than ebs
- can leverage EFS-IA for cost savings
- EFS only get billed for what you use
- for ebs, get billed on provisioned size not actual size
- NOTE
- rmb EFS VS EBS VS Instance Store
- efs for nfs to mount across multiple instances
- ebs is for network vol to mount only on 1 instance & locked to az
- instance store to get max amt of io on instance
- is ephemeral drive - easy lose data


###### tags: `AWS Developer Associate` `Notes`