A Python sample code demo how to collect Telemetry mesasge via UDP/gRPC and how to Parse message. Make it easy and fun to build your own telemetry recevier and agile handle telemetry message. Test bed : N9K and XRv9K N9K x.x.x.x version : nx-os 9.2.2 JSON / GPB-kv (Self-Description) xrv9k x.x.x.x version : xrv9k 6.4.1 JSON / GPB compact / GPB-kv (Self-Description) Telemetry Receiver UDP : port 57500 gRPC dial in : port 57400 gRPC dial out : port 50051 gRPC receiver working in dialin/dialout mode , passed testing with Cisco IOS XR Version 6.4.1 NOTE: max UDP protobuf length = 65535 ,header=28 ,real content = 65507 bytes Inside internal MDT header, IOS-XR has 12 bytes header, NX-OS has 6 bytes IOS-XR inside interanl header: +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | MSG TYPE | ENCODING_TYPE | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | MSG_VERSION | FLAGS | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | MSG_LENGTH | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ .. ~ ~ ~ PAYLOAD (MSG_LENGTH bytes) ~ +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ MSG TYPE (2 bytes) = 1 (for MDT) ENCODING_TYPE (2 bytes) = 1 (GPB), 2 (JSON) MSG_VERSION (2 bytes) = 1 FLAGS (2 bytes) = 0 MSG_LENGTH (4 bytes) 

use the follwoing command to make python proto output: protoc --python_out=. ./telemetry.proto //for GPB compact , demo message is system uptime ,use the following proto file generated by IOS-XR protoc --python_out=. ./uptime.proto to make gRPC service and service encode message pb2 with following command: python -m grpc_tools.protoc -I. --python_out=. --grpc_python_out=. cisco_grpc_dialout.proto 

Telemetry GPB message from NX OS Node :DUT02 IP Address (source port) :('10.75.58.198', 41532) Encodig Path :sys/procsys/sysmem GPB kv format dn:sys/procsys/sysmem free:26164608 memstatus:OK name:sysmem total:32828228 used:6663620 ================================================================== Telemetry GPB message from NX OS Node :DUT02 IP Address (source port) :('10.75.58.198', 41532) Encodig Path :show process cpu GPB kv format pid:1 runtime:16411 invoked:306652 usecs:53 onesec:0.00 process:init ================================================================= Telemetry GPB message from IOX Node :XTC IP Address (source port) :('10.75.58.60', 32898) Encodig Path :Cisco-IOS-XR-shellutil-oper:system-time/uptime GPB compact format Content decoded here : Host Name :XTC System Up Time :1660 Seconds ================================================================= 

================================================================= ('10.75.58.60', 33208) This is a System Uptime telemetry messgae Node =XTC Encoding Path =Cisco-IOS-XR-shellutil-oper:system-time/uptime Collection ID =5233 Host Name =XTC Uptime =58169 Recevice Count =1 Timestamp from Start =7.964596271514893 ================================================================= ('10.75.58.60', 33208) This is a System Uptime telemetry messgae Node =XTC Encoding Path =Cisco-IOS-XR-shellutil-oper:system-time/uptime Collection ID =5235 Host Name =XTC Uptime =58191 Recevice Count =2 Timestamp from Start =29.962813138961792 ================================================================= 

RP/0/RP0/CPU0:XTC#telemetry generate gpb-encoding path RootOper.QOS.Node.PolicyMap.Interface.Input.Statistics file disk0:/qos.proto Tue Jan 22 06:54:59.603 UTC Created /disk0:/qos.proto * NOTE For Telemetry generate protofile , path should be a yang to xml schema path, How to find yang to xml schema path? RP/0/RP0/CPU0:XTC#run Tue Jan 22 06:58:56.512 UTC [xr-vm_node0_RP0_CPU0:~]$cd /pkg/telemetry/mdt/protogen [xr-vm_node0_RP0_CPU0:/pkg/telemetry/mdt/protogen]$ls yang_to_schema.txt [xr-vm_node0_RP0_CPU0:/pkg/telemetry/mdt/protogen]$ yang_to_Schema.txt has all yang path to xml schema path maps. 

MDS SAN switch 32G line card push telemetry stream via gRPC GPB/GPB-kv encoding encoding path should be NX OS analytics:your_query_name dial-out only see sample code telemetry_grpc_dial_out_no_tls.py So far , NX OS 8.4.1 only support gRPC GPB/GPB-kv for GPB-kv encoding with fabrc_telemetry.proto file. MDS 9710 sample configuration: telemetry sensor-group 1 path analytics:test_query path show_stats_fc2/1 path show_stats_fc2/2 sensor-group 2 path analytics:dcnminitITL destination-group 1 ip address 10.79.98.77 port 50051 protocol gRPC encoding GPB-compact destination-group 2 ip address 10.124.2.116 port 57500 protocol gRPC encoding GPB-compact subscription 1 snsr-grp 1 sample-interval 30000 dst-grp 1 subscription 2 snsr-grp 2 sample-interval 30000 dst-grp 2 sw-core1-9710# sh analytics query all Total queries:2 ============================ Query Name :test_query Query String :select all from fc-scsi.port Query Type :periodic, interval 30 Query Name :dcnminitITL Query String :select port, vsan, app_id, initiator_id, target_id, lun, active_io_read_count, active_io_write_count, total_read_io_count, total_write _io_count, total_time_metric_based_read_io_count, total_time_metric_based_write_io_count,total_read_io_time, total_write_io_time, total_read_io_initiat ion_time, total_write_io_initiation_time,total_read_io_bytes, total_write_io_bytes, total_time_metric_based_read_io_bytes, total_time_metric_based_writ e_io_bytes, read_io_rate, write_io_rate, read_io_bandwidth, write_io_bandwidth,read_io_size_min, read_io_size_max, write_io_size_min, write_io_size_max ,read_io_completion_time_min, read_io_completion_time_max, write_io_completion_time_min, write_io_completion_time_max,read_io_initiation_time_max, writ e_io_initiation_time_max, read_io_aborts, write_io_aborts,read_io_failures, write_io_failures, read_io_timeouts, write_io_timeouts from fc-scsi.scsi_in itiator_itl_flow Query Type :periodic, interval 30 Query Options :differential gRPC decode sample ,please refer to san_analytics sub folder For SAN Analytics metrics detail please refer to san_analytics sub folder 

Sample telemetry machine learning engine code , LSTM prediction and Multivariate Gaussian Distribution Abnormal Detection. Non supervised learning , Affinity propagation clustering , for root cause detection and abnormal detection