Distributed system architectural styles are formulated in terms of:
- Components
- Connectors
- Data
- Configuration
Connector:
A mechanism that mediates coordination between components
Components:
Connected objects through calls
Middleware contains:
- Commonly used components
- Functions
RESTful:
A view of distributed systems as a collection of resources that are managed by components
RESTful architecture:
- Resources identified through a single naming scheme
- All services use same interface
- Messages are fully-described
Flooding search:
A node passes a request to all neighbors
Centralized system characteristics:
- Servers offer service
- Clients use services
- Clients and servers can be on different machines
- Clients follow a request/reply model
Wrapper solutions:
- 1 on 1
- broker
FTP:
Upload/Download model
Vertical distribution:
Dividing applications into three layers and running each layer on a different server
Horizontal distribution:
Dividing a client or a server into equivalent parts and each part runs its own data
Peer-to-peer architecture:
All processors are equal and they act as both client and server
Random walk search:
A node passes a request to a random neighbor
Super-peer network:
Breaks the symmetry in pure peer-to-peer networks by making one of the peers a super peer that can be used as an index and a broker
Edge of the network:
The boundary between enterprise networks and the Internet
BitTorrent collaboration steps:
- Lookup a torrent file in the global directory
- Join a swarm
- Get a chunk of the file for free
- Trade copies of your chunk with other peers for copies of theirs
Reasons for having an edge infrastructure:
- Latency and bandwidth
- Reliability
- Security and privacy
Blockchain principles:
- Blocks organized into append-only chains
- Blocks are immutable
