Since the advent of the word "automatic mode", people seem to be proud of the definition mode and use mode as the trend.
When the halo of the standalone mode gradually exits and the distributed mode is blank, although it is just a few short words, I believe we will introduce it in detail in the future. We can also write it.
The following text from: http://srinathsview.blogspot.com/2011/10/list-of-known-scalable-architecture.html
- lb (load balancers) + Shared nothing units -this mode is generally available for cache services, such as CDN
- LB + stateless nodes + Scalable storage .
- Peer to Peer ubuntures (Distributed Hash table (DHT) and Content Addressable networks (CAN) -P2P was once popular, however, I personally feel that the complexity of copyrights and protocols will block its development.
- distributed queues -
- Publish/Subscribe paradigm -4 and 5 are close to each other. The queue-based subscription service can design highly scalable systems, but its complexity cannot be ignored. At the same time, I think this kind of service can be intended for end users, such as RSS news subscription.
- gossip and nature-converted red ubuntures -similar to P2P, this type of system assumes that most people are good people, therefore, information can be shared throughout the community .
- map reduce/data flows -Classic Mode of large-scale data processing.
- tree of responsibility -similar to the global network architecture, physical data centers can be designed as well.
- stream processing -complements MAP/reduce to process very short-lived jobs.
- scalable storages -in fact, this is similar to 2nd entries. You may like to add an integer.
