Category: machine-learning

machine-learning

Extract Product Lines

To do effective legacy displacement, we need to figure out how to break down the problem into manageable pieces. Extract Product Lines does this by identifying product lines and using…

machine-learning

Patterns of Legacy Displacement

My colleagues do a lot of work that involves replacing legacy software systems. Three of my colleagues – Ian Cartwright, Rob Horn, and James Lewis – have started to capture…

machine-learning

photostream 126

Bacharach, Rhineland-Palatinate, Germany (2016) Source: martinfowler

machine-learning

Follower Reads

When you are using Leaders and Followers, you can use Follower Reads to achieve better throughput and lower latency. However there is always a replication lag between the leader and…

machine-learning

Stepping Back from Speaking

A personal note on why I wish to give up doing talks more… Source: martinfowler

machine-learning

Version Vector

A Lamport Clock is effective to order events on a single server, but with multiple servers we need a vector of version numbers to detect concurrent updates. more… Source: martinfowler

machine-learning

Lamport Clock

Next distributed system pattern from Unmesh is the Lamport Clock. Distributed systems can’t assume that time is consistent between servers, and sometimes time goes backwards. Yet it’s important to have…

machine-learning

Hybrid Clock

While a Lamport Clock is often necessary to order events in a distributed system, clients often need to know the actual times too. The Hybrid Clock, Unmesh’s next pattern, combines…

machine-learning

Versioned Value

When a distributed system has mutable data, nodes need to know which is the most recent value, so a versioned value stores a version number with every value. more… Source:…

machine-learning

Gossip Dissemination

Unmesh Joshi is completing another batch of his series on Patterns of Distributed Systems. First of these is Gossip Dissemination, which uses a random selection of nodes to pass on…