Development process of distributed architecture

Load balancing algorithm

1. Polling
2. Weighted polling
3. Random algorithm
4. Minimum number of connections
5. hash calculates the hash value according to the client ip, and hash%2 is modulated to determine the distribution target

Database evolution

1. Database performance improvement

• How to synchronize the database
• How to route the data source

< /ul>

1.1 Option 1: Introduce a search engine:

• Improve retrieval speed
• Reduce the pressure of reading the database
• lucence, solr, elk
• Index construction:
  Build an index based on raw data
  Full build -> First build
  Incremental build -> Incremental synchronization
  Non-real-time (asynchronous/timed task), real-time

1.2 Option 2: Caching

• Caching: In order to reduce low-speed devices And the difference between high-speed equipment.
• Nosql storage or other storage methods
  Nosql: hbase, mongoDB, Redis…
  Other caches: browser cache, application cache, db cache, cpu cache, file storage, etc

---

2. Distributed database

2.1 What is the role of transaction?

2.2 Distributed Database:

Realize database splitting by sub-database and table:
According to service characteristics (business volume, type etc. ) Configure database performance separately

---

Application evolution

Problems of traditional single application:

1. Deployment and Maintenance difficulties
2. Business coupling is very high
3. Performance bottlenecks

4. Test

   1. The application is split according to functional modules:

   1.1 such as: member service, transaction service, commodity service

The role of RPC in distributed?

1.2 Service segmentation

Advantages:

• Special personnel to maintain special services

  li>

• According to different throughput requirements, make targeted expansions

Disadvantages:

• Operation and maintenance costs Increase
• Increased team members

---

High Availability Plan

1 . Single point of failure

1.1 Multi-room deployment

---

monitoring

1. Link monitoring

• zipkin
• traceID

2. Hardware monitoring
cpu, internal memory, disk

---

architecture development process

1. SOA-ESB< /p>

• Client 1, Client 2….
• ESB:
  Communication and discovery of services
  Protocol conversion
  Security
  Limits Flow
• Server cluster

2. Microservices
Registration center: eureka, zookeeper, consoul

3. Container + k8s scheduling

4. serviceMesh (service mesh) The sidecar solves: service fusing, invocation, discovery, load balancing< /p>