Performance Efficiency - Storage performance options

Question 1

Storage considerations

Answer 1

• Sharing needs
• Latency needs
• Throughput
• Persistence or volatile
• File size
• Access method: sequentially or randomly
• Availability
• Durability
• Frequency of access and update

Question 2

ACID

Answer 2

• Atomicity: either all of the transaction succeeds or none of it does
• Consistency:
• All data will be valid according to all defined rules. Also refers to a query returning the same result every time the same request is made
• Strong consistency means the latest data is returned, but, it may result with higher latency
• With eventual consistency results are less consistent early on, but they are provided much faster with low latency
• Isolation: all transactions will occur in isolation. No transaction will be affected by any other transaction
• Durability: once a transaction is committed, it will remain in the system – even if there’s a system crash immediately following the transaction

Question 3

S3 - Characteristics 1

Answer 3

• Offers shared access, low latency, high throughput, durability and data availability. Doesn’t support NFS
• Provides encryption, access management, lifecycle management, and query-in-place (SQL like query capability)
• S3 Reduced Redundancy is an storage option for non critical / reproducible storage with less durability and less cost than S3
• S3 Transfer Acceleration enables fast and secure transfers of files over long distances between a client and S3. It takes advantage of CloudFront’s ELs

Question 4

S3 - Characteristics 2

Answer 4

• Objects are private by default. Can store an object with up to 5TB size
• S3 Object Lock allows to store objects using write-once-read-many (WORM) model. It prevents objects from being deleted or overwritten for a fixed amount of time or indefinitely
• S3 is a public service so it’s not under a VPC. To make a private connection can use a VPC gateway endpoint
• Lifecycle rules can be configured to abort all multipart uploads that are failing to complete in a specific time period

Question 5

S3 - Storage classes / tiers 1

Answer 5

• Standard: for general-purpose storage of frequently accessed data
• Intelligent tiering:
• For unknown or changing access frequency. Doesn’t require minimum storage duration
• Has some sub-tiers classified by access frequency, and amount of data retrieved
• Can store smaller objects than 128KB

Question 6

S3 - Storage classes / tiers 2

Answer 6

• Standard Infrequent Access (Standard IA):
• For data that is accessed less frequently, but requires rapid access when needed. As durable as Standard
• The minimum billable object size is 128 KB. Objects are charged for a minimum of 30 days
• Has less retrieval cost than Standard class
• One Zone Infrequent Access (One Zone IA):
• For the same purpose as Standard IA but with less availability
• Same charges as Standard IA for durability and retrieval

Question 7

S3 - Storage classes / tiers 3

Answer 7

• Glacier
• Glacier Deep Archive
• Outposts:
• Delivers object storage to on-premises AWS Outposts environment
• Ideal for workloads with local data residency requirements, and for performance needs that keeps data close to on-premises applications

Question 8

S3 - Glacier Characteristics

Answer 8

• Provides Archival storage. Minimum storage time is 90 days
• Offers encryption, access control, and audit logging
• Can query stored data without retrieval
• Latency options (can retrieve data from 1 minute to 12 hours)
• Expedited: premium price, data available in minutes
• Standard: data available in hours
• Bulk: cheaper and data available longer than the rest

Question 9

S3 - Glacier Deep Archive characteristics

Answer 9

• Very economical
• Minimum storage time is 180 days
• Can be accessed once or twice in a year and retrieved within 12 hours

Question 10

S3 - Organizing objects using prefixes

Answer 10

• A prefix is a logical grouping of the objects in a bucket. It’s similar to a directory name
• A delimiter allows to collect all the keys with a common prefix into a single summary list
• The prefix and delimiter parameters help to organize and then browse the keys hierarchically
• For example, if you store information about cities, you might organize them by country, then by province or state. You might use ‘/’ as the delimiter. So ‘Europe/France/Nouvelle-Aquitaine/Bordeaux’ and ‘North America/Canada/Quebec/Montreal’ would be some of the instances

Question 11

S3 - Presigned URLs

Answer 11

• A presigned URL gives access to the object identified in the URL, as long as the creator of the presigned URL has permissions to access that object
• Useful if you need a user / customer to be able to access / upload a specific object, but you don’t require them to have AWS security credentials or permissions
• On a presigned URL creation need the following: security credentials, bucket name, object key, HTTP method (GET to download, PUT to upload), and an expiration date and time

Question 12

S3 - CORS

Answer 12

• First S3 receives a preflight request from a browser, then evaluates the CORS configuration for the bucket and uses the first CORS Rule that matches the incoming browser request to enable a cross-origin request
• A CORS configuration is a document with rules that identify the origins to be allowed, the operations (HTTP methods) supported, and other specific operation information. JSON or XML can be used to describe rules

Question 13

EBS - Characteristics 1

Answer 13

• It’s always attached to an EC2 instance. Can be multi-attached to as many as 16 instances but with some conditions, only in the same AZ (not across many AZs)
• Provides replication on multiple AZs, access control, and encryption
• Snapshots:
• They’re automatically saved in S3 and constrained to the region they were created
• Can create new volumes from it. Can be generated and accessed by other EC2 instances
• Can copy them across regions
• The EBS fast snapshot restore feature charges for each snapshot and each enabled AZ

Question 14

EBS - Characteristics 2

Answer 14

• Elastic volumes is a feature that will automatically grow and shrink a volume based on the size of storing unit. You pay for what you use, not by the provisioned capacity
• RAID arrays allow to use multiple EBS volumes to improve performance or redundancy
• Use RAID 0 when performance is more important than fault tolerance
• Use RAID 1 when fault tolerance is more important than performance

Question 15

EBS - Storage options

Answer 15

• HDD (Throughput is the dominant attribute, measured in MB/s):
• Throughput optimized HDD: designed for frequently accessed, throughput-intensive workloads. Throughput: maximum 500 MB/s per volume
• Cold HDD: for less frequently accessed data at the lowest cost. Throughput: maximum 250 MB/s per volume
• SSD (IOPS is the dominant attribute):
• General purpose SSD: a balance of price and performance. Recommended for most workloads. IOPS: maximum 16,000 IOPS per volume
• Provisioned IOPS SSD: highest performance mission-critical, low-latency, or high-throughput workloads. IOPS: maximum 256,000 IOPS per volume
• Previous generation: HDD that can be used for workloads with small datasets where data is accessed infrequently

Question 16

EFS - Characteristics 1

Answer 16

• Provides highly-available, durable, and scalable file storage. Can be used as common data source for resources across many AZs and Regions
• Can be shared through many EC2. Can create mount targets in each AZ to lower latency
• Can shrink and grow based on data stored in it

Question 17

EFS - Characteristics 2

Answer 17

• Provides lifecycle policies to move to infrequent access storage classes
• Can use security groups and encryption
• Cannot use Windows instances with EFS
• It’s the only one that offers a standard file system interface provided through the operating system

Question 18

EFS - Modes

Answer 18

• Performance modes:
• General purpose (default option): has low latency on sequential and random access methods
• Max I/O: high throughput and parallel access from many EC2 instances. Not available when using One Zone storage classes
  NOTE: Cannot be changed after the FS is created
• Throughput modes:
• Bursting: scales as the size of the file system grows
• Provisioned: throughput is provisioned independent of storage’s size. For applications with more consistent throughput
  NOTE: Can be switched between them

Question 19

EFS - Storage classes

Answer 19

• Standard and Standard IA: high level of availability and durability by storing in multiple AZs. Useful for frequent and infrequent access
• Standard One Zone and One Zone IA: Amazon automatically backs up file systems using One Zone storage classes with AWS Backup. Useful for frequent and infrequent access