Distributed Systems

• Distributed system is a software system in which components located on networked computers communicate and coordinate their actions by passing messages.

Scalability

• Scalability is the ability of a system, network, or process, to handle a growing amount of work in a capable manner or its ability to be enlarged to accommodate that growth.Two particularly relevant aspects
  • Performance
  • Availability

Permformance aspects

• Short response time/low latency for a given piece of
work
• High throughput (rate of processing work)
• Low utilization of computing resource(s)

Availability

• Availability = uptime / (uptime + downtime)
• Availability % Downtime per year
• 90 % ("one nine") More than one month
• 99.9 % ("three nines") Less than 9 hours
• 99.9999% ("six nines") 31 seconds

Fault tolerance

• Fault tolerance ­ability of a system to behave in a well-defined manner once faults occur.
• Failures are norm.

Replication

• Replication is making copies of the same data on
  multiple machines.
• image.png
• image.png

Consistency

• image.png

Why strong consistency is hard to achieve

Nodes
• each node executes a program concurrently
• knowledge is local
• global state is potentially out of date
• nodes can fail and recover from failure independently
• messages can be delayed or lost
• clocks are not synchronized across nodes
Links
• Asynchronous system model.
• No timing assumptions.
• No bound on message transmission delay
• Useful clocks do not exist.

ACID

• Atomic
• Consistent
• Isolated
• Durable

BASE

• Basic Availability
• Soft­state
• Eventual consistency

CAP theorem

• It is impossible for a distributed computer system to simultaneously provide all three of the following guarantees:
  • Consistency (all nodes see the same data at the same time)
  • Availability (every request received by a non­failing
  [database] node in the system must result in a [non­error]
  response)
  • Partition tolerance (the system continues to operate despite
  arbitrary partitioning due to network failures)
• image.png
• image.png
• image.png
• image.png

CP systems

• Protocols:
  • Strict quorum protocols (paxos, raft, zab)
  • 2PC
• Storages:
  • MongoDB
  • HBase
  • Zookeeper

AP System

• image.png
• image.png

AP systems

• Protocols:
  • Partial quorum protocols
• Storages:
  • Couch DB
  • Cassandra
  • Amazon Dynamo

CAP fifteen years later

●Partitions are rare, there is little reason to forfeit C or
A when the system is not partitioned.
●Choice between C and A can occur many times
within the same system at very fine granularity.
●All three properties are more continuous than binary.
●Most software doesn’t neatly fit CP/AP definition.

参考来源

• cap-theorem-and-distributed-systems ppt
• CAP理论十二年回顾："规则"变了