Introduce ChaosEngineering for ShardingSphere

[mlycore]

As a critical components for any production use, ShardingSphere plays a very important role which have a high level availability requirements.

[mlycore]

At first we plan to have the DatabaseChaos, providing some tests on high availability of ShardingSphere. DatabaseChaos is currently a CRD of DatabaseMesh supporting AWS RDS Instance rebooting and AWS RDS Cluster automatically failover.

[mlycore]

Currently we plan to support two types of chaos for ShardingSphere:

• Infra: like Kubernetes and AWS chaos
• JVM: general JVM chaos

[mlycore]

Here are the issues:

• Introduce general infra chaos to ShardingSphere #253
• Introduce JVM chaos to ShardingSphere #254

[mlycore]

Recent thoughts on this, propose a new CRD called ShardingSphereChaos that contains two different kinds of Chaos:

• General Chaos: including kubernetes, network, disk, time, etc.
• JVMChaos: including byteman/bytebuddy scripts.

We have to specify the detail of one Chaos, the execution requirement, the explosive radius, the load generator, the observed result and its regarding quality gate. Then the chaos would be a full lifecycle management.