This documentation and the Cisco Observability Platform functionalities it describes are subject to change. Data saved on the platform may disappear and APIs may change without notice.

MELT Data

It is important to understand how a domain uses telemetry signals to represent aspects like performance, availability, functionality, security of any given entity. MELT (Metrics, Events, Logs, and Traces) represents different types of telemetry signals that are used to support the different aspects of observability in the Cisco Observability Platform.

The Cisco Observability Platform accepts data ingested in OTLP format only. However, it converts the OTLP data to an internal MELT format right after the initial authentication and authorization is complete. Further in the Cisco Observability Platform processing pipelines, MELT data is represented by a domain model defined by the application or enrichment developers. This format represents a contract for the developers because it is used by platform extension points, such as, Resource Mapping, Metric Mapping, Tag Enrichment, Metric Derivation, and Workflows.

To ensure that your solution is able to ingest telemetry data, include the following MELT signals in your data model:

SignalDescriptionMandatoryExample
MetricsAggregated set of measurements grouped or collected at regular intervals or a given time span; not discrete.YesMetric Definition Example
Expand source 
 {
  "namespace": {
    "name": "common",
    "version": 1
  },
  "kind": "metric",
  "name": "average_response_time",
  "displayName": "Average Response Time",
  "category": "meter_legacy",
  "contentType": "distribution",
  "aggregationTemporality": "delta",
  "isMonotonic": false,
  "type": "double",
  "unit": "ms",
  "ingestGranularities": [
    60
  ]
}
EventsA discrete action happening at a moment in time.  YesEvent Type Definition Example
Expand source 
[
 {
   "kind": "event",
   "namespace": {
     "name": "common",
     "version": 1
   },
   "name": "logs",
   "displayName": "Log events",
   "description" : "These are log events coming from customers",
   "attributeDefinitions": {
     "attributes": {
       "severity": {
         "type": "string"
       }
     }
   }
 }
]
LogsStrings of text with a discrete timestamp associated with them; unstructured or structured.No
The FMM model treats logs as events; type is a mandatory field. The default type for logs is logs:generic_record

Log Format Type Definition Example

Expand source 
{
 "types" : [
   {
     "kind": "event",
     "namespace": {
       "name": "logs",
       "version": 1
     },
     "name": "generic_record",
     "displayName": "Generic log record",
     "description" : "Generic log record, where attributes are discovered dynamically based on the customer supplied parsing logic",
     "attributeDefinitions": {
       "attributes": {
         "severity": {
           "type": "string"
         }
       }
     }
   }
 ]
}

Sample Log Record

Expand source 
{
   "entities": [{
       "type": "pod",
       "namespace": "k8s",
       "id": "84647373"
   }],
   "source": "LCA",
   "timestamp": 0,
   "types": ["logs:generic_record"],
   "raw": "[DEBUG] 2022-09-26 23:08:06 MSExchangeServer HandleDBRequest - Server=10.31.65.1 received request for lookup from [client_ip=112.21.231.28] and processing took 34 ms \n",
   "appd.log.format": "microsoft:exchange_server_logs"
}
TracesChains of events (or transactions) between different components in an application. Traces are discrete and irregular in occurrence.No
Trace Record
Expand source 
{
 "name": "RawTraceDto",
 "type": "record",
 "namespace": "com.appdynamics.avro.data.traces.v1",
 "doc": "Record to describe a raw telemetry distributed trace.",
 "fields": [
   {
     "name": "traceId",
     "type": "bytes",
     "doc": "Represents unique identifier of the trace."
   },
   {
     "name": "spans",
     "type": {
       "type": "array",
       "items": "com.appdynamics.avro.data.traces.v1.RawSpanDto"
     },
     "doc": "Collection of one or more raw spans related to this trace."
   }
 ]
}

It is also important to understand the entity characteristics, properties and its relationship with other entities.

Note: Entity Centric Pages created by this model displays the right metrics for the entity you have identified and the events associated with it. This helps monitor its performance.