Shulou(Shulou.com)06/02 Report--

What problem does 1 SET solve?

1.1 disaster recovery problem

The core service is down, and the main database is concentrated in an IDC,

1.2 Resource expansion problem

The expansion of a single IDC is limited (for example, the computer room is fully loaded and the bandwidth is insufficient). Expanding IDC across computer rooms and remote locations will lead to delay problems caused by cross-computer rooms.

In the case of single data, the number of connections is limited, and unlimited expansion of applications is not supported.

1.3 the problem of splitting large clusters

The large scale of distributed cluster will bring the problems of resource expansion, large cluster split and disaster recovery.

From a business point of view, there will be disaster recovery and remote requirements, which cannot be realized at the bottom.

The unitary architecture arises at the historic moment

2. Evolution of SET solution

2.1 dual active users in the same city, remote master-slave architecture for storage, mq, cache, and scheduled tasks, writing across data rooms; an exception in the data center can be manually switched.

2.2 two places and three centres

On the basis of dual active users in the same city, data centers are deployed in different locations, and only disaster-level failures are enabled (remote cold backup)

Cold backup dare not cut, waste resources, the essence is still a single point, resources, disaster recovery, expansion is still unsolved 2.2.3.The SET traffic is routed to the central cluster or the unit cluster Route By UserId.

3 SET architecture

3.1 Unitary cluster, each cluster can independently process the traffic in the unit, realizing traffic separation and fault isolation.

3.2 only the data of the cluster is stored in each unit, and then the data is synchronized in both directions to achieve disaster recovery switching requirements.

3.3 Central cluster: services that have not been uniformized, usually non-core services

3.4 SET strategy:

Middleware: MQ, RPC, KV, etc.

RPC--- makes closed calls to SET services and invokes the central cluster to non-SET services

MQ and KV--- support the production, query and consumption of SET.

Data synchronization: full data: deploy the central cluster for the data with small amount of data, and route other data to the cell cluster for future planning: for the data of each unit, they need to synchronize with each other. SET capability for disaster recovery requirements: traffic scheduling and disaster recovery efficient localization service: according to the information from the front end, ip, territory, traffic routing to the most recent SET O2O scenario requires efficient localization service capability container deployment: SET one-click creation and release

4 what are the principles of SET architecture:

4.1Transparency to business systems. Business does not need to care about SET rules and deployment issues.

Sharding principle: split according to the business level, give priority to the largest business dimension, split a large number of O2O users according to the local proximity, a single SET is not limited to a single computer room, but the cluster should not be too large

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.