A new service description format in generator module

[salehsedghpour]

Currently, the representation of chains and clusters is somehow complex/hard/time consuming. Unified, prettier and simpler

[salehsedghpour]

The initial idea for a new service description format could be something like:

{
  "latencies": [
    {
      "src" : "cluster_1",
      "dest": "cluster_2",
      "latency": 0.002
    }
  ],
  "clusters": [
    {
      "name": "cluster_1",
      "namespace": "ns_1",
      "services": [
        {
          "name": "service_1",
          "node": "node-1",
          "resources": {
            "limits": {
              "memory": 1024,
              "cpu": 1000
            },
            "requests": {
              "memory": 512,
              "cpu": 500
            }
          },
          "endpoints": [
            {
              "name": "/end_1",
              "protocol": "http",
              "cpu_consumption": 0.003,
              "network_consumption": 0.002,
              "memory_consumption": 0.003,
              "calledservices": [
                {
                  "cluster": "cluster_1",
                  "service": "service_2",
                  "endpoint": "/end_2",
                  "protocol": "grpc",
                  "traffic_ratio": 1,
                  "requests": "asynchronous"
                }
              ]
            }
          ]
        }
      ]
    }
  ]
}

It could help to solve #2 , #3, #4, #5, #8, #9, #12, #13, #14, #16, #18.
@alekodu do you have any comment on this?

[salehsedghpour]

As we discussed, we decided to:

• split the services for name services simply by having a list of dictionaries for each namespace.
• Add a random option for both cluster or namespace placement
• protocol could be deleted for each called service
• traffic ratio should be renamed to traffic forward ratio.

Also, we decided to wait for merging this PR until the services are ready.
Another point is to fork the repo on our own account to make us able to use every possible feature of Github.

[salehsedghpour]

Based on our discussion and to support everything we needed in the previous comment, we decided to go for:

{
  "latencies": [
    {
      "src" : "cluster_1",
      "dest": "cluster_2",
      "latency": 0.002
    }
  ],
  "services": [
    {
      "name": "service_1",
      "clusters": [
        {
          "cluster": "cluster_1", // Random
          "namespace": "ns_1", // Random
          "node": "node_1"// Random
        }
      ],
      "resources": {
        "limits": {
          "memory": "1024",
          "cpu": "1000"
        },
        "requests": {
          "memory": "512",
          "cpu": "500"
        }
      },
      "endpoints": [
        {
          "name": "/end_1",
          "protocol": "http",
          "cpu_consumption": 0.003,
          "network_consumption": 0.002,
          "memory_consumption": 0.003,
          "calledservices": [
            {
              "cluster": "cluster_1",
              "service": "service_2",
              "endpoint": "/end_2",
              "traffic_forward_ratio": 1,
              "requests": "asynchronous"
            }
          ]
        }
      ]
    }
  ]
}