RDS + AURORA + ELASTICACHE Flashcards
1
Q
What is RDS?
A
Relational Database Service. It works with SQL as query language
2
Q
What kind of SQL databases does RDS support?
A
Posgres
My SQL
Aurora (AWS)
Microsoft SQL Server
MariaDB
Oracle
IBM DB2
3
Q
What is Aurora
A
AWS’s SQL Relational Database
4
Q
What are some benefits of running SQL on RDS rather than EC2 instances?
A
The database provisioning is automated
Automatic OS patching
continuous backups being made, which you can restore to a specific timestamp
Monitoring dashboards, to view the database’s performance
Read replicas to improve read performance
Easily setup multi-az for DR: disaster recovery
Automatic maintenance windows for minor db pgrades
You can easily scale horizontally and vertically.
Storage is backed by EBS
5
Q
What is a disadvantage of running your SQL in RDS?
A
you can’t ssh to your RDS instances
6
Q
What is storage auto scaling?
A
It’s for RDS. It scales automatically when it detects your database is about to run out of storage.
7
Q
How does storage auto scaling work?
A
You set a maximum storage threshold.
It auto increases storage if:
free space less than 10%
this low storage lasts more than 5 minutes
8
Q
Where can you scale your read replicas to?
A
Within same AZ, cross AZ, and cross Region.
9
Q
How many read replicas can we create?
A
Up to 15
10
Q
What is eventually consistent ASYNC replication? and how does it apply to RDS read replicas?
A
Its how the replication process works for read replicas. It means the replicas have a delay in getting the data replicated from the original Database instance. So when you read from a replica you may get old information.
11
Q
What is a read replica promotion?
A
Read replicas can be promoted to their own database. They are out of the replication system after that. They become a separate database.
12
Q
What is a connection string in SQL?
A
Its what tells the application where the database is, for example:
Data Source=190.190.200.100,1433;Network Library=DBMSSOCN;Initial Catalog=myDataBase;User ID=myUsername;Password=myPassword;
13
Q
How do you connect the read replicas to the app that uses the database?
A
By adding connection strings in the app
14
Q
What are use cases for read replicas?
A
When you have a production app that uses a database, and you need to add for example a reporting tool that needs to read the database.
This can slow down the production app. So what you can do instead is use a read replica for the reporting tool to read, and you dont affect the production app and the main database instance.
15
Q
What are SELECT, INSERT, UPDATE, DELETE in SQL?
A
Types of statements you use in an app to interact with a database.
What is important in read replicas is to only interact with it with SELECT statements, which equal to read statements.
SELECT = read
insert means add, update means modify, and delete is remove. These make changes to the database, but with read replicas we only want to read them, since the changes come replicated from the original database.
16
Q
What SQL statements should be used to interact with RDS read replicas?
A
only SELECT, which equals to read.
17
Q
In AWS normally there is a cost when data goes from where to where?
A
From an AZ to another AZ.
From a region to another region
Or from aws to outside of aws
18
Q
Does RDS read replicas have any cost for the data replicated between AZs?
A
No. RDS is an exception and its free. (Only within same region)
19
Q
Does RDS read replicas have any cost for the data replicated between Regions?
A
YES.
20
Q
What RDS feature is for disaster recovery?
A
RDS Multi-AZ
21
Q
What RDS feature is for read scalability?
A
Read Replicas
22
Q
What RDS feature is for automated storage scalability?
A
Storage auto scaling
23
Q
How does RDS Multi-AZ work?
A
It creates a live or “SYNC” replica of your database in a different AZ.
This replica is in standby, and in case of AZ, network, storage, or any failure of main DB, it automatically failovers to the standby replica, which now becomes the master.
24
Q
Which RDS feature uses a SYNC replica?
A
RDS Multi-AZ
25
Which RDS feature uses an ASYNC replica?
RDS Read Replicas
26
Can Read Replicas be set up as Multi-AZ?
Yes. A read replica can have multi-az enabled and be highly available itself. meaning the read replica will have a replica in standby.
27
Describe turning on Multi-AZ for an RDS database.
Just click "Modify" on the database, and enable "Multi-AZ"
28
Does enabling Multi-AZ on a DB imply any downtime?
NO. Zero downtime.
29
How does enabling Multi-AZ for an RDS DB work behind the scenes?
AWS takes a snapshot of your DB instance, and then restores it to a new standby Database. Then syncronization between both databases is enabled. (Writes to the main database are also written to the standby database).
30
How do you create a database in RDS? (steps)
RDS > Create Database
Select database type: Aurora, mariadb, ibm db2, oracle, etc.
Select DB version
Enter a DB aws identifier
Enter new user and password for db
Select instance type "db.t3.micro" for example.
Select storage type, size, and iops.
configure storage auto scaling
configure networking options, VPC, subnet, SG, and port for Db.
setup authentication: user/pw, iam users and roles, or kerberos.
Enable monitoring settings
Enter the database name
enable backups and its settings
enable log export to cloudwatch
enable automatic minor version upgrades and configure maintenance window for such upgrades
31
What is RDS Custom?
A managed, Oracle or SQL Server only database.
It grants you OS and Database customizations.
It lets you access through SSH to the underlying EC2 instance of the database
32
What is a recommendation to perform before making changes to your RDS Custom Database or OS?
De-activate automation mode, and take a snapshot of the database before making changes
With automation mode enabled, rds still treats the db as a managed db, but you can still access the instance.
33
Can you take snapshots of your database instances? If so, how does it work?
Yes. You can take an RDS Database Snapshot. This snapshot is a backup can be restored to a new database.
34
What OS configurations do you need to perform in your RDS database?
None. AWS does everything.
RDS automates the setup, operation and scaling of the databases.
35
Whats the difference between RDS and RDS Custom?
RDS is managed by AWS. With RDS Custom you can manage the OS and database through SSH and configure the database..
36
Which types of databases can you use with RDS Custom?
Oracle and SQL Server.
37
Which DBs is Aurora driver compatible with?
MySQL or Postgre. You can choose one of these 2 on creation.
38
Which is faster, Aurora, Postgre or MySQL?
Aurora. Aurora is cloud optimized, meaning its x5 faster than MySQL in aws and x3 Faster than Postgre in aws.
39
How does Aurora manage storage?
It starts at 10GB and automatically grows up ti 128TB
40
How is replication in Aurora better than other DBs?
Aurora's replication is faster than other DBs, having sub 10ms replica lag.
41
How does Multi-AZ work with Aurora?
It doesn't. Aurora is HA native, and failover is instantaneous.
This means that since Aurora is cloud native, you get HA by default.
42
What is a cost of using Aurora DB?
It's 20% more expensive than other RDSs, but it's more efficient.
43
How does Aurora store its data?
It makes 6 copies of all the data across 3 AZs.
The database has one shared logical volume spread across volumes in the 3 AZs.
Each time data is written, it goes to the 3 AZs, to 2 different physical volumes per AZ.
But it's not the same 2 volumes in each AZ. Each write goes to 2 different volumes in each AZ, meaning the whole database is spread across hundreds (100s) of volumes.
44
What happens to Aurora if 1 AZ goes down?
It can still operate normally with 2 AZs
45
What are the main features of Aurora Storage?
It does replication of data across 3 AZs in the region. To 100s of volumes.
It does constant self healing of the data with peer to peer replication, to keep 6 copies of it across the 3 AZs.
It automatically expands storage with use.
All of this is by default.
46
How many copies of data do you need for writes and reads in Aurora?
You need 4 copies for writes, and 3 copies for reads. This for every piece of data.
47
How do read replicas work in Aurora?
In aurora, you have 1 master instance that makes writes to the storage, and up to 15 reader instances that take reads from the storage.
A reader instance can become the master in under 30 seconds if the master fails.
48
What is the writer endpoint in Aurora?
It's a DNS name that always points to the master, even if the master failovers to a reader instance
49
What is Aurora Auto Scaling?
It's the automated scaling of the number of reader instances depending of the need of the database. This feature of aurora is combined with the use of a Reader Endpoint.
50
What is a Reader Endpoint in Aurora?
Since we use many reader instances, and auto scaling. For the app to be able to connect to these automatically created/deleted reader instances, it uses the Reader Endpoint. Now the app only needs to connect to the reader endpoint, and it connects the app to one of the reader instances.
It's basically a load balancer at the connection level between the app and the reader instances.
51
How do you increase storage in Aurora?
You don't. Storage Auto expands between 10GB and 128TB.
52
What is Backtrack?
An Aurora feature and technology, that lets you restore your database to any point in time, without the use of backups.
53
What are the 2 performance options when creating Aurora?
Aurora standard and Aurora I/O-Optimized
54
What is a Regional Cluster?
It's the conjunction of instances of the Aurora Database. The Writer instance + all the Reader instances.
55
What do the reader and writer endpoints point to?
The writer endpoint will always point the the Writer instance.
The reader endpoint will always point to one of the reader instances.
56
What are reader instances in Aurora?
The aurora version of read replicas
57
Can you have an aurora database be cross-region?
Yes. You can create a read replica for an aurora database in another region. This is a "Cross Region Read Replica"
58
How do you configure auto scaling in Aurora?
You create a policy in the DB, that says based in average TARGET CPU utilization, or average TARGET connections, create more read replicas, to keep as close to that target as possible.
You can also set minimum and maximum of read replicas.
You can also configure scale in, so that it will remove read replicas after a period, if your usage is lower than the configured target.
59
What is Aurora Global Databases?
You can add regions to your Aurora Database, ideal for global apps.
It consists of one region with a writer instance. And up to 5 regions with reader instances.
Up to 16 reader instances per region.
60
What are custom endpoints?
A feature that lets you use different custom reader endpoints.
This lets you redirect traffic to different groups of reader instances depending on the custom endpoint used.
It's useful, for example, when you need a subset of reader instances of a more powerful kind, for running more analytical queries.
You set a custom endpoint, and this new endpoint will only forward queries to these powerful instances, and then a different custom endpoint will redirect read requests to other reader instances.
61
What is Aurora Serverless?
It's a more automated version of Aurora. It does auto scaling based on the actual workload.
It's more granular in its scalability.
You don't need to do capacity planning.
For example, you can have a serverless aurora database consisting of a single DB instance. And this instance will automatically scale up and down its ACU (CPU and Memory), between the minimum and maximum you set up.
This is automated vertical scaling.
Then you can add reader instances just like any other aurora database. These instances will also be serverless with capabilities for automated vertical scaling. Adding instances is horizontal scaling.
Then you can enable auto scaling of serverless instances. This is automated horizontal scaling.
For each serverless writer or reader instance, you only pay for what you use. The minimum possible is 0.5 ACUs.
Note: Here instance means of the database. Not a VM.
62
When is Aurora Serverless more useful?
When you have infrequent, intermittent, or unpredictable workloads.
63
What are ACUs?
ACU: Aurora Capacity Units: These are the units that scale the serverless version of Aurora. They are equivalent to cpu memory and networking.
64
What is the vertical scaling unit for aurora serverless?
ACUs: Aurora Capacity Units.
65
When is RDS serverless and not serverless?
Aurora is the only option of RDS that can be serverless. Every other option including "normal" Aurora is not serverless.
Serverless is when you dont provision VM instances. Serverless Aurora uses ACUs for vertical scalability, which is automated, so it can be considered serverless. With aurora if you have no workloads your usage can be a minimum of 0.5 ACUs, which equates to 1BG of RAM and 1 CPU, and you just pay for that half of an ACU. This is considered serverless, since you are not paying for example for a t2 VM instance, you are paying for the current usage in ACUs of one or more serverless DB instances.
If in change if you use a t2 instance, you pay for the whole instance even if you currently have no workloads.
66
When is serverless Aurora cheaper than normal Aurora?
When there is no usage of the database. For Aurora with VMs, you pay for the whole VM instance, no matter if theres no traffic.
For serverless Aurora, you only pay for what you use. And if there is no usage it automatically scales down the ACUs to the minimum, thus saving you money in those hours of inactivity.
This is why its best used for intermittent, infrequent or unpredictable workloads.
67
What is promoting a Region?
It's a functionality in Global Aurora. In case of an outage in the primary aurora region, you can promote a read only region to primary.
This has an RTO of less than 1 minute.
68
What are global aurora use cases?
For global apps. Read only regions work as cache for accessing the database faster from different places in the world.
69
What is Aurora Machine Learning?
It's an integration between Aurora and ML services, like Amazon SageMaker and Amazon Comprehend.
The ML services integrate with the database to analyze the data and return query results depending on the query.
69
How long does it take for aurora to replicate to another region?
Less than 1 second
70
How are RDS automated backups performed? How do they work?
Daily full backup of the database. (During the backup window).
Transaction logs are backed up every 5 minutes. This means you have a backup at the database level every 5 minutes. You can restore from the oldest backup to 5 minutes ago.
The oldest backup depends on the retention policy. This can be between 1 to 35 days.
70
What are transaction logs in a database? What are they useful for?
Database transaction logs record all changes and modifications, offering a detailed history of processed data. These logs are crucial for recovery, enabling tracking and restoring specific transactions in case of errors. Every SQL database maintains a transaction log that documents all transactions and their associated changes.
71
What are good use cases for snapshots in RDS?
For automated backups the retention limit is 35 days. For snapshots there is no limit for retention.
You can also snapshot an RDS database and delete the RDS database to save costs, then recreate it from the snapshot when you need to use it. This is a trick that can come up in the exam. This would be useful for a database that you only use for 2 hours per month for example.
Since the snapshot storage is several times cheper than the database storage, which you still have to pay for if you stop the database. You can instead snapshot the database and delete it to save money.
72
How do Aurora Backups work? Whats the difference with other RDS backups?
Aurora is 1 to 35 days same as RDS. But in Aurora these backups can't be disabled.
You have a recovery timeframe of any point in time within that retention period. In RDS it's every five minutes.
73
What happens when you restore an RDS /Aurora backup or snapshot?
It creates a new database
74
How do you restore an On-Premises MySQL Database to RDS or Aurora?
Create a Backup of the database
Store it in S3
Restore it from S3 into a new RDS instance
75
Why is Cloning an Aurora Database so fast and cost effective compared to taking a snapshot and restoring it to a new database?
Because it doesn't actually copy any data until it's re-written:
Cloning an Aurora Cluster, creates a new cluster in AWS, but this cluster shares the data with the original cluster, it doesn't actually copy it until there is a write operation. Writes are processed to new volumes corresponding to the new database. While reads are still performed to the volumes of the original database, until these are re-written.
76
What is Aurora Cloning? What is a use case?
Aurora cloning is a feature that allows you to create a new Aurora Cluster from a pre existing one. It's faster than snapshot and restore.
A use case could be for staging. Staging is a process in which you store modify and then move data. It's a temporary database.
You need to do this to your database, but you don't want to affect production performance? Create a clone of the database and do it there.
77
What can you do to secure your RDS Database?
Encryption: You can encrypt all data at rest of your database instances using AWS KMS. This includes snapshots and auto backups.
This feature MUST be defined AT LAUNCH.
In flight encryption: It's enabled by defaulton RDS side. Clients will need to use the aws root certs to encrypt data in flight.
Use IAM Roles to connec to your database instead of user and pw.
Control network access to your database with security groups.
No SSH Access available except RDS Custom.
Enable Audit Logs and send them to CloudWatch for longer retention.
78
What is a workaround for encrypting an already existing RDS Database?
You can snapshot the database, and restore to a new encrypted database
79
What is RDS Proxy?
A fully aws managed database proxy.
80
Why use RDS Proxy?
Apps connect to the proxy instead of the database. Then the proxy pools these connections together reducing the cpu and memory required for the database to process connections.
81
What are RDS Proxy use cases?
It's for improving database efficiency by reducing resource usage in the database. It also minimizes open connections and timeouts to the database.
Its useful for when you have a lot of connections to your database instance. For example when connecting from lambda functions to your RDS database.
82
How does RDS Proxy work?
It's serverless, autoscalable, and highly available across AZs.
83
How does using an RDS Proxy affect a master instance failover of RDS/Aurora?
It reduces the time it takes to failover by up to 66%
84
How does RDS Proxy improves your DB Security?
It enforces IAM authentication for the DB. With RDS Proxy you can make people only connect through IAM to your database.
This IAM secret can then be stored in AWS SECRETS MANAGER service.
85
How do you connect through the internet to your RDS Proxy?
You can't. You need to configure access through networking in the VPC service.
86
What is ElastiCache?
A service that manages in memory data or cache.
Compatible with Redis and Memcached.
These 2 are cache storaging systems.
87
What is a Cache in elasticache?
It's an in memory database with very high performance, and low latency.
88
How can elasticache help your RDS Database?
It reduces the load of the database for read intensive workloads, by caching the most queried data, and reducing the queries received by the database.
89
What does AWS manage in Elasticache?
OS Maintenance, patching, setup, configuration, monitoring, failure recovery and backups.
Just like with RDS.
90
How does implementing Elasticache affect your app?
It requires heavy application code changes.
Your app will need to be configured to query the cache before or after querying the RDS DB.
91
Explain a Cache Hit.
The app queries the elasticache, if the query has already been made and is stored in elasticache, then you get a Cache Hit.
In this case you are saving a trip to the database.
92
Explain a Cache Miss
The app queries the elasticache, if the query hasn't been made yet then it's not stored in elasticache, then you get a Cache Miss.
In this case you need to fetch the data from the database.
Then the data is written to the cache, so next time the same query is made, it is in the elasticache and you get a cache hit.
93
What is an ElastiCache invalidation?
Its how you invalidate cached data when it becomes obsolete. With this you make sure the cached data is updated to the actual data behind the cache.
94
How can ElastiCache help with app user sessions?
To make the app stateless. Meaning that if you log in to an app, this session data will be stored in cache instead of the app database.
Then if you join or are redirected to a new instance of this app you are not asked to login again since the login data is in cache.
95
What are ElastiCache common use cases?
For caching a databases most queried data.
For storing app user session data in cache. (Stateless apps)
96
What are the 2 types of ElastiCaches?
Redis and Memcached
97
What are the main characteristics of Redis ElastiCache?
Multi AZ with Auto-Failover
Read replicas for scaling and HA
AOF Persistence for data durability
Backup and Restore Features
It divides the cache in sets and sorted sets
98
Which ElastiCache system has the most similar features to RDS?
Redis
99
Which ElastiCache system's features are more durability focused?
Redis
100
What are the main characteristics of Memcached ElastiCache?
Multi node for partitioning data (Sharding)
No replication
Non persistent
No backup and restore
Multi threaded architecture (multiple instances of the cache system)
101
Which ElastiCache option is more simple?
Memcached
102
Which ElastiCache option is best suited for High Availability, Disaster Recovery, etc?
Redis
103
Which ElastiCache option is best suited for complex data types?
Redis
104
Which ElastiCache option is best suited for speed?
Memcached
105
Which ElastiCache option is best for when you don't care to lose data?
Memcached
106
Which ElastiCache option partitions the data in multiple nodes?
Memcached
107
Which ElastiCache option has better performance? Why?
Memcached. It has a sharding feature that divides the data in multiple nodes, thus improving read speeds.
108
Which ElastiCache option supports auto-scaling? How does it work?
Both Redis and Memcached. You have a serverless option at creation, which you configure to scale the speed of your cache instances depending on usage. It is more expensive
109
What is easycreate?
A simplified method or creating your elasticaches
110
Can you run elasticache onpremises? How?
Yes. With AWS Outposts.
111
How does Multi AZ work for ElastiCache?
Same as RDS. It auto failovers to a read replica on a different AZ on failure.
112
Which ElastiCache option is the only one that supports Multi AZ?
Redis.
113
How does ElastiCache work?
Once created, the elasticache gives you the names for the writer endpoint and reader endpoint, which you configure in your app to write and read to the cache.
114
Does ElastiCache support encryption?
Yes encryption at rest and in transit
115
Which ElastiCache option is great for Leaderboards?
Redis. It has a feature called sorted sets that is useful for element ordering.
116
Which port does FTP use?
21
117
Which port does SFTP use?
22
118
Which protocol uses same port as SSH?
SFTP
119
What do the ports the databases use for communication look like?
They usually use one port, and they are in the 1000s, 3000s, 5000s, and stuff like that.
120
What is a popular database RDS does not support? Why?
MongoDB. Cause it's noSQL.
121
You're planning for a new solution that requires a MySQL database that must be available even in case of a disaster in one of the Availability Zones. What should you use?
Multi AZ
122
We have an RDS database that struggles to keep up with the demand of requests from our website. Our million users mostly read news, and we don't post news very often. Which solution is NOT adapted to this problem?
An elasticache cluster?
RDS Multi AZ?
RDS Read Replicas?
RDS Multi AZ
Elasticache reduces load of the RDS Database because the cached data is not queried from the database. The most frequently read data is read from cache.
123
Which RDS (NOT Aurora) feature when used does not require you to change the SQL connection string?
Multi AZ. Its a standby syncronous replica instance of the writer and has same dns name as the writer.
124
Your application running on a fleet of EC2 instances managed by an Auto Scaling Group behind an Application Load Balancer. Users have to constantly log back in and you don't want to enable Sticky Sessions on your ALB as you fear it will overload some EC2 instances. What should you do?
Store session data in ElastiCache
125
How can you enhance the security of your ElastiCache Redis Cluster by allowing users to access your ElastiCache Redis Cluster using their IAM Identities (e.g., Users, Roles)?
Using Redis Authentication?
Using IAM Authentication?
Using Security Groups?
Using IAM Authentication (Only for Redis)
126
Which ElastiCache supports IAM Authentication to the system?
Redis.
127
You would like to create a disaster recovery strategy for your RDS PostgreSQL database so that in case of a regional outage the database can be quickly made available for both read and write workloads in another AWS Region. The DR database must be highly available. What do you recommend?
Create a Read Replica in the same region and enable Multi-AZ on the main database
Create a Read Replica in a different region and enable Multi-AZ on the Read Replica
Create a Read Replica in the same region and enable Multi-AZ on the Read Replica
Enable Multi-Region option on the main database
Create a Read Replica in a different region and enable Multi-AZ on the Read Replica
128
Which RDS database technology does NOT support IAM Database Authentication?
Postgre
Oracle
MySQL
Oracle
129
You need to store long-term backups for your Aurora database for disaster recovery and audit purposes. What do you recommend?
Enable Automated Backups
Perform On Demand Backups
Use Aurora Database Cloning
Perform On Demand Backups
130
Your development team would like to perform a suite of read and write tests against your production Aurora database because they need access to production data as soon as possible. What do you advise?
Make a DB snapshot and restore it into a new database
Create an Aurora Read Replica for them
Do the teste against the production database
Use aurora cloning feature
Use aurora cloning feature. Faster than snapshot.
