VSS
Distributed Systems @ HSR
Distributed Systems @ HSR
Fichier Détails
| Cartes-fiches | 61 |
|---|---|
| Langue | Deutsch |
| Catégorie | Informatique |
| Niveau | Université |
| Crée / Actualisé | 06.08.2020 / 10.08.2020 |
| Lien de web |
https://card2brain.ch/box/20200806_vss
|
| Intégrer |
<iframe src="https://card2brain.ch/box/20200806_vss/embed" width="780" height="150" scrolling="no" frameborder="0"></iframe>
|
Which parts does the CAP theorem consist of?
Consistency
Availability
Partition-tolerance
What does consistency (CAP Theorem) mean?
every node has the same consistent state
What does availability (CAP Theorem) mean?
Every non-failing node always returns a response
What does partition-tolerance (CAP Theorem) mean?
The system continues to be consistent even when network paritions
What is vertical scaling?
More hardware in the same machine
What does horizontal scaling mean?
multiple machines
What are the advantages of vertical scaling?
- lower cost at small scale
- no adaption of software required
- less administrative effort
- less power consumption
What are disadvantages of vertical scaling?
- risk of HW failure causing outage / more difficult to add fault tolerance
- HW limits for scaling / limited by moores law
What are the advantages of horizontal scaling?
- lower cost at massive scale
- easier to add fault tolerance
- higher availability
What are the disadvantages of horizontal scaling?
- adaption of software required
- more complex systems and more components are involved
- more power consumtion
Name 3 reasons for distributed system
- scaling
- location
- fault tolerance
What are the two categories of distributed systems?
- controlled distributed systems
- fully decentralized systems
What are some aspects of controlled distributed systems?
- 1 responsible organization
- low churn
- secure environment
- high availibility
- can be homogeneous / heterogeneous
- consistency
What are some aspects of fully decentralized systems?
- N responsible organizations
- High churn
- hostile environment
- unpredictable availability
- is heterogeneous
- weak consistency
List the 7 types of transparency in the context of distributed systems
- location transparency
- access transparency
- migration, relocation transparency
- replication transparency
- concurrent transparency
- failure transparency
- security transparency
What is location transparency?
users should not be aware of physical location
What is access transparency?
users should access in a single, uniform way
What is migration, relocation transparency?
users should not be aware resources have moved
What is replication transparency?
users should not be aware of replicas, it should appear as a single resource
What is concurrent transparency?
users should not be aware of other users
What is failure transparency?
users should not be aware of recovery mechanisms
What is security transparency?
users should be minimally aware of security mechanisms
What is moores law?
number of transistors doubles every 2 years, doubling chip performance every 18 months
What is nielsens law?
a high-end users connection speed grows by 50% per year
What is kryders law?
disk density doubles every 13 months
What is the cause of bit-flips?
cosmic rays
What is Bitsquatting?
= DNS hijacking without exploitation
which means registering domains with a single bit error
e.g. aeazon.com -> amazon.com
works because of possible bit flips
Is load balancing vertical or horizontal scaling?
horizontal scaling
Name the 3 types of load balancers
- hardware load balancer
- software load balancer
- cloud-based load balancer
Which goals do load balancers achieve?
- high availability and reliability by requesting only servers that are online
- flexibility to add or subtract servers as demand dictates
Which messages are sent if a TCP connection is established?
1. SYN
2. SYN-ACK
3. ACK
Which messages are sent if a TCP connection is terminated?
1. FIN
2. ACK + FIN
3. ACK
How does a DDoS Amplification attack work?
- attacker sends ping with spoofed IP (10bytes)
- server responds to victim (100bytes)
--> allows attacker to amplify attack with factor 10
Name the 8 load balancing algorithms
- round robin
- weigthed round robin
- least connections
- least time
- least pending requests
- agent based
- hash
- random
Name 3 arguments for using queues in distributed systems
- communicating without continous connection
- decoupling (hardware is not reliable)
- systems can be heterogeneous
Name 3 standards used for message queues
- AMQP
- STOMP
- MQTT
What does AMQP stand for?
Advanced Message Queing Protocol
What does STOMP stand for?
Streaming Text Oriented Messaging Protocol
What does MQTT stand for?
Message Queuing Telemetry Transport
What topologies for messages queues do you know?
- Point-to-Point, Producer-Consumer
- Publish-Subscribe
- Bidirectional Queues
