Google Cloud TPU
Incident Management is now generally available! Incident Management is now generally available!

Google Cloud TPU

Crawler Crawler

Overview

Google Cloud TPU products make the benefits of Tensor Processing Units (TPUs) available via scalable and easy-to-use cloud computing resource for all ML researchers, ML engineers, developers, and data scientists running cutting-edge ML models.

Use the Datadog Google Cloud Platform integration to collect metrics from Google Cloud TPU.

Setup

Installation

If you haven’t already, set up the Google Cloud Platform integration first. There are no other installation steps.

Log collection

Google Cloud TPU logs are collected with Stackdriver and sent to a Cloud pub/sub with an HTTP push forwarder. If you haven’t already, set up a Cloud pub/sub with an HTTP push forwarder.

Once this is done, export your Google Cloud TPU logs from Stackdriver to the pub/sub:

  1. Go to the Stackdriver page and filter the Google Cloud TPU logs.
  2. Click Create Export and name the sink.
  3. Choose “Cloud Pub/Sub” as the destination and select the pub/sub that was created for that purpose. Note: The pub/sub can be located in a different project.
  4. Click Create and wait for the confirmation message to show up.

Data Collected

Metrics

gcp.tpu.cpu.utilization
(gauge)		Utilization of CPUs on the TPU Worker as a percent.
Shown as percent
gcp.tpu.memory.usage
(gauge)		Memory usage in bytes.
Shown as byte
gcp.tpu.network.received_bytes_count
(count)		Cumulative bytes of data this server has received over the network.
Shown as byte
gcp.tpu.network.sent_bytes_count
(count)		Cumulative bytes of data this server has sent over the network.
Shown as byte

Events

The Google Cloud TPU integration does not include any events.

Service Checks

The Google Cloud TPU integration does not include any service checks.

Troubleshooting

Need help? Contact Datadog support.

On this Page