chore: add two test schemas in data generator examples

One schema is a test for many small databases like a bunch of free tier users. The other is for a database with 2k measurments getting moderate ingest load (4.1 MB/sec).

iox_data_generator/schemas/many_dbs.toml

@@ -0,0 +1,80 @@
+# This schema is meant to test out many databases writing data in like a bunch of free tier users.
+# Start with a database_list of 10k to make things interesting. This will send on average of
+# 208 requests/sec and 6.25 MB/sec across the 10k databases. The top 10 will have 60 requests/min
+# and 1.8MB/min.
+name = "many_dbs"
+
+[[values]]
+name = "some_tag_10"
+cardinality = 2
+template = "id_{{id}}_{{random 15}}"
+has_one = ["extra_static"]
+
+[[values]]
+name = "child_tag"
+cardinality = 3
+belongs_to = "some_tag_10"
+has_one = ["rotation"]
+template = "id_{{id}}_{{random 10}}"
+
+[[values]]
+name = "rotation"
+cardinality = 4
+template = "id_{{id}}_{{guid}}"
+
+[[values]]
+name = "extra_static"
+cardinality = 1
+template = "whatever-constant-value"
+
+[[tag_sets]]
+name = "first_set"
+for_each = [
+    "some_tag_10",
+    "some_tag_10.extra_static",
+    "some_tag_10.child_tag",
+    "child_tag.rotation",
+]
+
+# each sampling from this agent generates 32,465 bytes of LP, first few lines look like:
+# main_measurement_1,agent_id=1,child_tag=id_1_T6iJnnBTE3,extra_static=whatever-constant-value,measurement_tag=1,rotation=id_1_de4ddb8c-31a6-440f-a273-7132bdd43bd7,some_tag_10=id_1_rWtIkI26LTlfu0J intfield=71334i,floatfield=0.7934452557768101 1639151629935287000
+# main_measurement_1,agent_id=1,child_tag=id_2_VsiUF2xVuz,extra_static=whatever-constant-value,measurement_tag=1,rotation=id_2_890145b3-8157-4d6f-ac02-1fe37584190f,some_tag_10=id_1_rWtIkI26LTlfu0J intfield=64582i,floatfield=0.0957134480635704 1639151629935287000
+# main_measurement_1,agent_id=1,child_tag=id_3_XNL51f1NdT,extra_static=whatever-constant-value,measurement_tag=1,rotation=id_3_8bcf7547-06e9-4033-9ffb-e00ac4e6c5a9,some_tag_10=id_1_rWtIkI26LTlfu0J intfield=26179i,floatfield=0.09993902612184669 1639151629935287000
+# main_measurement_1,agent_id=1,child_tag=id_4_mqCyprcTDQ,extra_static=whatever-constant-value,measurement_tag=1,rotation=id_4_f465d43e-f1ab-4250-99ac-67af7c1d4c72,some_tag_10=id_2_X4eWjH9ImjTeta2 intfield=16511i,floatfield=0.033060266070114475 1639151629935287000
+[[agents]]
+name = "first_agent"
+tag_pairs = [
+    {key = "agent_id", template = "{{agent.id}}"}
+]
+
+[[agents.measurements]]
+name = "main_measurement_{{measurement.id}}"
+count = 20
+tag_set = "first_set"
+tag_pairs = [
+    {key = "measurement_tag", template = "{{measurement.id}}"}
+]
+
+[[agents.measurements.fields]]
+name = "intfield"
+i64_range = [1, 100000]
+
+[[agents.measurements.fields]]
+name = "floatfield"
+f64_range = [0.0, 1.0]
+
+[[database_writers]]
+database_ratio = 0.001
+agents = [{name = "first_agent", sampling_interval = "1s"}]
+
+[[database_writers]]
+database_ratio = 0.01
+agents = [{name = "first_agent", sampling_interval = "10s"}]
+
+[[database_writers]]
+database_ratio = 0.1
+agents = [{name = "first_agent", sampling_interval = "30s"}]
+
+[[database_writers]]
+database_ratio = 1.0
+agents = [{name = "first_agent", sampling_interval = "60s"}]

iox_data_generator/schemas/many_measurements.toml

@@ -0,0 +1,61 @@
+# This schema tests what load looks like with many measurements (2,000). If pointed at a single database
+# with the configured 20 agents at 10s sampling, it will send an average of 2 requests/second (representing
+# 16k rows) with 4.1MB/second of LP being written. Each agent writes 8k lines per request.
+name = "many_measurements"
+
+[[values]]
+name = "some_tag"
+cardinality = 2
+template = "id_{{id}}_{{random 15}}"
+has_one = ["extra_static"]
+
+[[values]]
+name = "child_tag"
+cardinality = 2
+belongs_to = "some_tag"
+has_one = ["rotation"]
+template = "id_{{id}}_{{random 10}}"
+
+[[values]]
+name = "rotation"
+cardinality = 3
+template = "id_{{id}}_{{guid}}"
+
+[[values]]
+name = "extra_static"
+cardinality = 1
+template = "whatever-constant-value"
+
+[[tag_sets]]
+name = "first_set"
+for_each = [
+    "some_tag",
+    "some_tag.extra_static",
+    "some_tag.child_tag",
+    "child_tag.rotation",
+]
+
+[[agents]]
+name = "first_agent"
+tag_pairs = [
+    {key = "agent_id", template = "{{agent.id}}"}
+]
+
+[[agents.measurements]]
+name = "main_measurement_{{measurement.id}}"
+count = 2000
+tag_set = "first_set"
+tag_pairs = [
+    {key = "measurement_id", template = "{{measurement.id}}"}
+]
+
+[[agents.measurements.fields]]
+name = "intfield"
+i64_range = [1, 100000]
+
+[[agents.measurements.fields]]
+name = "floatfield"
+f64_range = [0.0, 1.0]
+
+[[database_writers]]
+agents = [{name = "first_agent", sampling_interval = "10s", count = 20}]