Sample configuration for AWS EC2 handling 25k Obcerv entities and 10k metrics/sec (small) with NGINX Ingress controller

Download this sample AWS EC2 handling 25k Obcerv entities and 10k metrics/sec (small) configuration provided by ITRS.

# Example Obcerv configuration for AWS EC2 handling 25k Obcerv entities and 10k metrics/sec.
#
# Nodes: (4) c5.4xlarge (16CPU, 32GB)
#
# The resource requests total ~33 cores and ~79GiB memory (assuming collection-agent DaemonSet runs on 4 nodes)
# and includes Linkerd resources.
#
# Disk requirements:
# - Timescale:
#   - 4 x 512 GiB timeseries data disk for each replica (x3)
#   - 50 GiB data disk for each replica (x3)
#   - 50 GiB WAL disk for each replica (x3)
# - Kafka: 100 GiB for each replica (x3)
# - Loki: 30 GiB for each replica (x1)
# - Zookeeper: 1 GiB for each replica (x3)
# - etcd: 1 GiB for each replica (x3)
# - Downsampled Metrics:
#   - Raw: 5 GiB for each replica (x3)
#   - Bucketed: 5 GiB for each replica (x3)
#
# The configuration references a default storage class named `gp3` which uses EBS gp3 volumes. This storage class should
# be configured with the default minimum gp3 settings of 3000 IOPS and 125 MiB/s throughput - you can create
# this class or change the config to use a class of your own, but it should be similar in performance.
#
# This configuration is based upon a certain number of Obcerv entities, average metrics per entity, and
# average metrics collection interval. The following function can be used to figure out what type of load to expect:
#
# metrics/sec = (Obcerv entities * metrics/entity) / average metrics collection interval
#
# In this example configuration, we have the following:
#
# 10,000 metrics/sec = (25,000 Obcerv entities * 4 metrics/entity) / 10 seconds average metrics collection interval
#

defaultStorageClass: "gp3"
apps:
  externalHostname: "obcerv.mydomain.internal"
  ingress:
    annotations:
      kubernetes.io/ingress.class: "nginx"
      nginx.org/mergeable-ingress-type: "master"
ingestion:
  externalHostname: "obcerv-ingestion.mydomain.internal"
  replicas: 2
  ingress:
    annotations:
      kubernetes.io/ingress.class: "nginx"
      nginx.ingress.kubernetes.io/backend-protocol: "GRPC"
  resources:
    requests:
      memory: "512Mi"
      cpu: "500m"
    limits:
      memory: "512Mi"
      cpu: "500m"
iam:
  ingress:
    annotations:
      kubernetes.io/ingress.class: "nginx"
      nginx.org/mergeable-ingress-type: "minion"
zookeeper:
  replicas: 3
  resources:
    requests:
      memory: "256Mi"
      cpu: "200m"
    limits:
      memory: "512Mi"
      cpu: "200m"
kafka:
  replicas: 3
  diskSize: "100Gi"
  defaultPartitions: 24
  consumer:
    fetchMaxWaitMs: 250
    fetchMinBytes: 524288
  resources:
    requests:
      memory: "3Gi"
      cpu: "1"
    limits:
      memory: "3Gi"
      cpu: "2"
timescale:
  clusterSize: 3
  dataDiskSize: "50Gi"
  timeseriesDiskCount: 4
  timeseriesDiskSize: "512Gi"
  walDiskSize: "50Gi"
  resources:
    requests:
      memory: "14Gi"
      cpu: "2"
    limits:
      memory: "14Gi"
      cpu: "4"
  retention:
    entity_attributes:
      chunkSize: 2d
      retention: 1y
    metrics:
      chunkSize: 8h
      retention: 30d
    metrics_5m:
      chunkSize: 1d
      retention: 90d
    metrics_1h:
      chunkSize: 5d
      retention: 180d
    metrics_1d:
      chunkSize: 20d
      retention: 1y
    statuses:
      chunkSize: 7d
      retention: 1y
    signal_details:
      chunkSize: 1d
      retention: 30d
loki:
  diskSize: "30Gi"
sinkd:
  jvmOpts: "-XX:MaxDirectMemorySize=100M"
  resources:
    requests:
      memory: "1280Mi"
      cpu: "250m"
    limits:
      memory: "1280Mi"
      cpu: "400m"
  rawResources:
    requests:
      memory: "1Gi"
      cpu: "250m"
    limits:
      memory: "1Gi"
      cpu: "400m"
platformd:
  replicas: 2
  resources:
    requests:
      memory: "1536Mi"
      cpu: "1"
    limits:
      memory: "2Gi"
      cpu: "1500m"
dpd:
  jvmOpts: "-Xmx4G"
  maxEntitySerdeCacheEntries: 25000
  metricsMultiplexer:
    maxFilterResultCacheSize: 200000
    maxConcurrentOps: 100
    localParallelism: 6
  selfMonitoringThresholds:
    metrics_partition_lag_warn: 100000
    metrics_partition_lag_critical: 500000
  resources:
    requests:
      memory: "5Gi"
      cpu: "2"
    limits:
      memory: "5500Mi"
      cpu: "3"
metricForecastd:
  resources:
    requests:
      memory: "512Mi"
      cpu: "250m"
    limits:
      memory: "768Mi"
      cpu: "500m"
downsampledMetricsStream:
  resources:
    requests:
      memory: "3Gi"
      cpu: "1"
    limits:
      memory: "3Gi"
      cpu: "1500m"
  bucketedResources:
    requests:
      memory: "3Gi"
      cpu: "1"
    limits:
      memory: "3Gi"
      cpu: "3"
  rocksdb:
    raw:
      indexAndFilterRatio: 0.5
      memoryMib: 100
      writeBufferMib: 8
      writeBufferRatio: 0.25
    bucketed:
      indexAndFilterRatio: 0.5
      memoryMib: 100
      writeBufferMib: 8
      writeBufferRatio: 0.25
entityStream:
  intermediate:
    storedEntitiesCacheSize: 1000
    resources:
      requests:
        memory: "1Gi"
        cpu: "300m"
      limits:
        memory: "2Gi"
        cpu: "500m"
  final:
    resources:
      requests:
        memory: "1Gi"
        cpu: "300m"
      limits:
        memory: "2Gi"
        cpu: "500m"
signalsStream:
  resources:
    requests:
      memory: "768Mi"
      cpu: "150m"
    limits:
      memory: "1536Mi"
      cpu: "300m"
etcd:
  replicas: 3
collection:
  metrics:
    resources:
      requests:
        memory: "768Mi"
        cpu: "200m"
      limits:
        memory: "1Gi"
        cpu: "250m"
["Obcerv"] ["User Guide", "Technical Reference"]

Was this topic helpful?