Ceph and Apache Hudi Integration

Input and output integration overview

The Ceph plugin for Telegraf helps in gathering performance metrics from both MON and OSD nodes in a Ceph storage cluster for effective monitoring and management.

Writes metrics to Parquet files via Telegraf’s Parquet output plugin, preparing them for ingestion into Apache Hudi’s lakehouse architecture.

Integration details

Ceph

The Ceph Storage Telegraf plugin is designed to collect performance metrics from Monitor (MON) and Object Storage Daemon (OSD) nodes within a Ceph storage cluster. Ceph, a highly scalable storage system, integrates its metrics collection through this plugin, facilitating easy monitoring of its components. With the introduction of this plugin in the 13.x Mimic release, users can effectively gather detailed insights into the performance and health of their Ceph infrastructure. It functions by scanning configured socket directories for specific Ceph service socket files, executing commands via the Ceph administrative interface, and parsing the returned JSON data for metrics. The metrics are organized based on top-level keys, allowing for efficient monitoring and analysis of cluster performance. This plugin provides valuable capabilities for managing and maintaining the performance of a Ceph cluster by allowing administrators to understand system behavior and identify potential issues proactively.

Apache Hudi

This configuration leverages Telegraf’s Parquet plugin to serialize metrics into columnar Parquet files suitable for downstream ingestion by Apache Hudi. The plugin writes metrics grouped by metric name into files in a specified directory, buffering writes for efficiency and optionally rotating files on timers. It considers schema compatibility—metrics with incompatible schemas are dropped—ensuring consistency. Apache Hudi can then consume these Parquet files via tools like DeltaStreamer or Spark jobs, enabling transactional ingestion, time-travel queries, and upserts on your time series data.

Configuration

Ceph

[[inputs.ceph]]
  ## This is the recommended interval to poll. Too frequent and you
  ## will lose data points due to timeouts during rebalancing and recovery
  interval = '1m'

  ## All configuration values are optional, defaults are shown below

  ## location of ceph binary
  ceph_binary = "/usr/bin/ceph"

  ## directory in which to look for socket files
  socket_dir = "/var/run/ceph"

  ## prefix of MON and OSD socket files, used to determine socket type
  mon_prefix = "ceph-mon"
  osd_prefix = "ceph-osd"
  mds_prefix = "ceph-mds"
  rgw_prefix = "ceph-client"

  ## suffix used to identify socket files
  socket_suffix = "asok"

  ## Ceph user to authenticate as, ceph will search for the corresponding
  ## keyring e.g. client.admin.keyring in /etc/ceph, or the explicit path
  ## defined in the client section of ceph.conf for example:
  ##
  ##     [client.telegraf]
  ##         keyring = /etc/ceph/client.telegraf.keyring
  ##
  ## Consult the ceph documentation for more detail on keyring generation.
  ceph_user = "client.admin"

  ## Ceph configuration to use to locate the cluster
  ceph_config = "/etc/ceph/ceph.conf"

  ## Whether to gather statistics via the admin socket
  gather_admin_socket_stats = true

  ## Whether to gather statistics via ceph commands, requires ceph_user
  ## and ceph_config to be specified
  gather_cluster_stats = false

Apache Hudi

[[outputs.parquet]]
  ## Directory to write parquet files in. If a file already exists the output
  ## will attempt to continue using the existing file.
  directory = "/var/lib/telegraf/hudi_metrics"

  ## File rotation interval (default is no rotation)
  # rotation_interval = "1h"

  ## Buffer size before writing (default is 1000 metrics)
  # buffer_size = 1000

  ## Optional: compression codec (snappy, gzip, etc.)
  # compression_codec = "snappy"

  ## When grouping metrics, each metric name goes to its own file
  ## If a metric’s schema doesn’t match the existing schema, it will be dropped

Input and output integration examples

Ceph

1. Dynamic Monitoring Dashboard: Utilize the Ceph plugin to create a real-time monitoring dashboard that visually represents the performance metrics of your Ceph cluster. By integrating these metrics into a centralized dashboard, system administrators can gain immediate insights into the health of the storage infrastructure, which aids in quickly identifying and addressing potential issues before they escalate.

2. Automated Alerting System: Implement the Ceph plugin in conjunction with an alerting solution to automatically notify administrators of performance degradation or operational issues within the Ceph cluster. By defining thresholds for key metrics, organizations can ensure prompt response actions, thereby improving overall system reliability and performance.

3. Performance Benchmarking: Use the metrics collected by this plugin to conduct performance benchmarking tests across different configurations or hardware setups of your Ceph storage cluster. This process can assist organizations in identifying optimal configurations that enhance performance and resource utilization, promoting a more efficient storage environment.

4. Capacity Planning and Forecasting: Integrate the metrics gathered from the Ceph storage plugin into broader data analytics and reporting tools to facilitate capacity planning. By analyzing historical metrics, organizations can forecast future utilization trends, enabling informed decisions about scaling storage resources effectively.

Apache Hudi

1. Transactional Lakehouse Metrics: Buffer and write Web service metrics as Parquet files for DeltaStreamer to ingest into Hudi, enabling upserts, ACID compliance, and time-travel on historical performance data.

2. Edge Device Batch Analytics: Telegraf running on IoT gateways writes metrics to Parquet locally, where periodic Spark jobs ingest them into Hudi for long-term analytics and traceability.

3. Schema-Enforced Abnormal Metric Handling: Use Parquet plugin’s strict schema-dropping behavior to prevent malformed or unexpected metric changes. Hudi ingestion then guarantees consistent schema and data quality in downstream datasets.

4. Data Platform Integration: Store Telegraf metrics as Parquet files in an S3/ADLS landing zone. Hudi’s Spark-based ingestion pipeline then loads them into a unified, queryable lakehouse with business events and logs.

Feedback

Thank you for being part of our community! If you have any general feedback or found any bugs on these pages, we welcome and encourage your input. Please submit your feedback in the InfluxDB community Slack.