Architecture

Question 1

What is a distributed monolith architecture?

Answer 1

It is an anti-pattern in which a monolith is separated into multiple services that are highly dependent and do not adopt the best practices of miscroservices architecture.

Question 2

What is a cell based architecture?

Answer 2

Multiple isolated and independent instances of a workload, called a cell, handling a subset of the overall workload.

Question 3

What are the layers of a cell based architecture?

Answer 3

Cell router, Cell, and Control plane

Question 4

What is a microservice based architecture?

Answer 4

A collection of small autonomous services that work together that are bounded by context, autonomously developed, independently deployable, decentralized and built and released with automated processes.

Question 5

What kind of architecture is diagrammed in the attached image?

Answer 5

Cell based architecture

Question 6

What is an example of a distributed monolith architecture?

Answer 6

Any instance in which two or more services cannot be deployed in isolation. One cannot be changed without the other also being changed.

Question 7

What is an ETL pipeline?

Answer 7

A series of processes in which data is extracted from one or more sources, transformed - usually into a standardized format - and then loaded into a target repository, usually a database or data warehouse.

Question 8

What is the purpose of cell based architecture?

Answer 8

Cell based architecture is aimed at predictable scaling and limiting impact of failures.

Question 9

What is an example of a cell based architecture?

Answer 9

When the entire production environment is replicated. Or a logical subset.

Question 10

How does a cell based architecture reduce the impact of failures?

Answer 10

By isolating failures to a single cell and avoiding single points of failure.

Question 11

How does cell based architecture allow for predicable scaling?

Answer 11

By capping each cell to a fixed maximum size.

Question 12

What does SOLID stand for?

Answer 12

SOLID stands for:

S - Single-responsibility Principle
O - Open-closed Principle
L - Liskov Substitution Principle
I - Interface Segregation Principle
D - Dependency Inversion Principle

Question 13

What does single responsibility mean?

Answer 13

A class should have one and only one reason to change, meaning that a class should have only one job.

Question 14

What is the open closed principle?

Answer 14

Objects or entities should be open for extension but closed for modification.

Question 15

What is the liskov substitution principle?

Answer 15

Let q(x) be a property provable about objects of x of type T. Then q(y) should be provable for objects y of type S where S is a subtype of T.

This means that every subclass or derived class should be substitutable for their base or parent class.

Question 16

What is interface segregation?

Answer 16

A client should never be forced to implement an interface that it doesn’t use, or clients shouldn’t be forced to depend on methods they do not use.

Question 17

What is the dependency inversion principle?

Answer 17

Entities must depend on abstractions, not on concretions. It states that the high-level module must not depend on the low-level module, but they should depend on abstractions.

Question 18

What are the typical building blocks of data intensive applications?

Answer 18

1. Data storage (databases)
2. Remembering the results of expensive operations to speed up reads. (caching)
3. Allowing users to filter or search data (search indexes).
4. Asynchronously send messages to be handled by another process. (stream processing)
5. Periodically crunch large amounts of data (batch processing)

Question 19

How do you make systems reliable?

Answer 19

1. Designs systems in a way that minimizes opportunities for error.
2. Decouple the places where people make the most mistakes from the places where they can cause errors.
3. Test thoroughly.
4. Allow quick and easy recovery from human error.
5. Set up clear and detailed monitoring, such as performance metrics and error rates (telemetry).
6. Implement good management practices.

Question 20

What is scalability?

Answer 20

Scalability describes a systems ability to handle increased load.

Question 21

How do you describe load?

Answer 21

Through load parameters specific to your system (cache hits, simultaneous people in a chat room, db transactions, requests).

Question 22

Latency vs response time

Answer 22

The response time is what the user sees, latency is the duration that a request is waiting to be handled.

Question 23

What’s the best way to measure performance?

Answer 23

By using percentiles and a median value.

Question 24

What’s the best way to measure performance?

Answer 24

By using percentiles and a median value.

Question 25

How do you express percentiles?

Answer 25

P values. For example, p50 is 50th percentile, or the median.

Question 26

What is head of line blocking?

Answer 26

When a request takes a long time, preventing subsequent requests from going through.

Question 27

What is tail latency amplification?

Answer 27

When users require multiple backend calls thus increasing the proportion of slow end user requests (response times).

Question 28

What is the wrong and right way of aggregating response times?

Answer 28

The wrong way is by averaging percentiles (which is mathematically meaningless. The right way is by using a histogram.

Question 29

What is the wrong and right way of aggregating response times?

Answer 29

The wrong way is by averaging percentiles (which is mathematically meaningless). The right way is by using a histogram.

Question 30

What three factors influence the maintainability of a system?

Answer 30

Operability - make it easy for the operations team to keep the system running smoothly Simplicity - make it easy to understand the system by removing unnecessary complexity Elvolvability - make your system extensible and easy to evolve.

Question 31

What is normalization?

Answer 31

The process of structuring a data model to reduce data redundancy and improve database integrity.

Question 32

Give an example where document based databases might not perform well.

Answer 32

When there are multiple kinds of relationships between data objects and queries to retrieve them: Many to one relationships are an example of this. This can be compensated for by denormalization, but this will require additional work to ensure the denormalized data is consistent.

Question 33

What is an example of normalization.

Answer 33

Given a field school, with a value string with potential multiple repetitions of the same value, updating the field to a reference integer would be normalizing.

Question 34

What are document databases good at?

Answer 34

One to many relationships. Storing huge values of data - having flexible schemas, performing better due to locality, more fidelity to the data model used in the application layer, ability to be sharded for quicker lookup.

Question 35

What are relational databases good at?

Answer 35

Better Support for joins, many to one and many to many relationships. Basically, highly interconnected data.

Question 36

What is data locality?

Answer 36

Data that’s all in one place as opposed to being dispersed among tables.

Question 37

What performance limitations significantly reduce the set of situations in which document databases are useful?

Answer 37

Updated to documents - if the encoded size of the document changes on update, it requires an entire rewrite as opposed to a change in place.

Question 38

What performance limitations significantly reduce the set of situations in which document databases are useful?

Answer 38

Updated to documents - if the encoded size of the document changes on update, it requires an entire rewrite as opposed to a change in place.

Question 39

What is an example of resilient routing in AWS?

Answer 39

Route 53 as DNS routes requests according to its own routing records - latency based for example - and those end requests end up at in-region load balancers which further distribute load as configured.

Question 40

Application load balancer vs network load balancer

Answer 40

1. 7th layer path based routing, web sockets, and health checks for web apps. 2. 4th layer tcp routing for low latency IoT applications, gaming serves, db connections

Question 41

What is a target group?

Answer 41

Target groups group ec2 instances or ECS cluster tasks and are associated with load balancers so that LB’s can distribute traffic to them.