0224 How to face high concurrency? Cache? Why would the fire station?

Change of scene

Scenes Feature Change point
old Complex business logic
new A large number of concurrent users to access high Software development methodology, process management, organizational structure change

Vertical scaling

Improve server hardware configuration to support;


two questions:

  1. The upper limit of the physical server hardware performance;
  2. The cost of inputs with the performance increase is not linear;

Horizontal scaling

Use more servers to support the formation of clusters

Evolution distributed architecture

image.png

image.png

image.png

image.png

image.png

image.png

22 Cache

background

Distributed architecture, using most of the cache can optimize performance.

classification

Who operation according to the data source can be divided into two categories cache;

Access cache

Direct connection to the data source assembly cache and update the cache

scenes to be used:

  1. CDN deal with static resources;
  2. Reverse proxy caching dynamic resource

Bypass cache

Application directly connected to a data source to update the cache

scenes to be used:

  1.  Local cache: jvm present in the stack data;
  2. Distributed Cache: routing algorithm, hash (key) / length, primarily through hash table;

Advantages and disadvantages

advantage:

  1. Respond faster, much faster than disk memory and remote access;
  2. Saving computing resources is a direct result of the cache;
  3. Can reduce the pressure, cdn web access can reduce the pressure, the reverse proxy can reduce the pressure on the database access

Disadvantages:

  1. Dirty read data; (expired and a failure notification process)
  2. Hot data difficult to recognize;

Clients can also use the cache, reducing access to static resources;

Concept stage

Development in Taiwan

time Development Description
2008 Taobao and Tmall duplication and waste of resources problem - "share a common part, platform
2015 supercell advanced mode (precipitated a universal material, algorithm)
Ali began large and medium units, small reception organizational and operational mechanisms of transformation
2017 Ali, drops began to share experience in the construction stage, small and medium enterprises have begun, but encountered many problems
2018 baj started in Taiwan-based organizational structure change, the station completely hot up
2019 SMEs continue to build in Taiwan, but a lot of confusion, not to determine the theoretical system of knowledge, only some practical experience

Why in the table so the fire?

I think the two big reasons:
First, the model benefits baj large enterprises, do, and have achieved good results;
second, change and development of SMEs encountered problems; summed up in three little points ;
pain point 1. SME information process, redundant construction (chimneys, silos of data), the station can solve the pain points;
2. SMEs in the face of economic development downstream, hope to their core competencies and maximize precipitation reuse, enterprise core competencies uncertainty facing market uncertainties;
3. SMEs in order to pursue sustained business growth, focus on the B-side (supply side), hoping to give their ability to assign more traditional industries.

summary

image.png

image.png

image.png

image.png32

By manual carefully you can learn:

1. Understand the evolution of distributed architecture;

2. Common classification of cache and the advantages and disadvantages;

Development and popular in Taiwan reasons 3. Analysis;

The original is not easy, please indicate the source, more than welcome to communicate

Published 110 original articles · won praise 10 · views 20000 +

Guess you like

Origin blog.csdn.net/tian583391571/article/details/104482918