Requirements, Trade-offs, and Challenges

Question 1

What are the types of web service availability?

Answer 1

Service availability

Data availability

Question 2

What are the types of consistency models?

Answer 2

Data - centric

Client - centric

Question 3

Types of Data-centric consistency that don’t use synchronization?

Answer 3

Strict consistency
Sequential
FIFO (PRAM)
Causal

Question 4

What is Strict consistency?

Answer 4

Uses absolute time ordering

Strongest Memory Coherence and strictest requirements.

Read operations on the memory address always returns the same value written by the most recent write operation to that address.

Requires absolute global time.
Deemed practically impossible for Distributed Systems

Question 5

Sequential Consistency

Answer 5

All processes see the same order of all access operations

No operations started until the previous one is completed

Question 6

Causal Consistency

Answer 6

Better concurrency than sequential
All processes see causally-related shared accesses in the same order.
Need to keep track of dependencies

Question 7

FIFO (PRAM) Consistency

Answer 7

PRAM (Pipelined Random Access Memory)
All processes see writes from each other in the order they were used.

Good performance
Simple and Easy

Question 8

What are the types of Data-centric consistency models that use synchronization?

Answer 8

Weak
Release
Enter

Question 9

Weak consistency

Answer 9

Synchro accesses are sequentially consistent

Before synchro access, all previous regular data accesses are to be completed.

Before regular access, all synchro data accesses are to be completed.

Burden on programmer
Good performance

Question 10

Release consistency

Answer 10

Relaxation of weak consistency

Made consistent when a critical section is exited

Burden on programmer
Explicit feedback on critical section
Good performance

Question 11

Entry consistency

Answer 11

Data on a critical region made consistent when a critical region is entered

Ordinary shared data associated with a synchronization variable

when acquire is done on a synchro variable only data guarded by that variable will be made consistent.

Question 12

User-centric consistency

Answer 12

To present individual apps with a view of the database that is consistent with their own actions

Uses Eventual consistency

Question 13

Eventual consistency

Answer 13

No updates occur for a while, then eventually all updates propagate through the system

Question 14

User-centric consistency models

Answer 14

Monotonic reads
Monotonic writes
Read your writes
Writes follow reads

Question 15

Latency

Answer 15

The time between a request and a response

Question 16

Tail latency

Answer 16

A small percentage of response times that take the longest in comparison to the bulk of its response times.

Question 17

What causes variability in response times?

Answer 17

Shared resources
Background services & activities 
Queueing 
Energy management
Garbage collection

Question 18

How to reduce system latency

Answer 18

Selective Replication
Parallelization of sub-operations
Reducing head of line blocking (HOL)
Managing background services

Question 19

Reducing database access latency

Answer 19

Replication & Sharding
Eventual consistency
Read replicas
Caching
Geo-Locating

Question 20

Resiliency

Answer 20

The ability of a system to recover from failures and continue to work with min downtime and data loss