Docs » Splunk Infrastructure Monitoring のメトリクス パイプライン管理 » Scenario: Improve storage use and costs by routing and archiving your data

Scenario: Improve storage use and costs by routing and archiving your data 🔗

Available in Enterprise Edition. For more information, see Subscription types, expansions, renewals, and terminations.

以下のシナリオは、架空のeコマース企業であるButtercup Gamesの例です。

背景 🔗

SkylerはButtercup Gamesの中央observabilityチームの管理者です。Skylerは、会社の予算内に収まるように、様々なチームのobservabilityの使用状況を監視しています。

Lately, Skyler has been noticing an increase in their metric time series (MTS) usage. With the help of the Splunk Observability Cloud account team, Skyler obtains a detailed metrics usage report. The report gives Skyler insights into their MTS volume, use of dimensions with high cardinality, use of those MTS in charts and detectors, and distribution of MTS across different teams.

Skyler は、あるチームが割り当てられた使用量の上限に近づいていることに気づきます。Skyler はそのチームのサイト信頼性エンジニア（SRE）リードである Kai に連絡を取り、チームの使用量を最適化するよう依頼します。Skyler は、カーディナリティの高いディメンションの MTS、MTS の総使用量、チームの MTS 使用量を Kai に伝えました。

調査結果 🔗

The metrics usage report shows that Kai’s team sends about 50,000 MTS for the service.latency metric to Splunk Observability Cloud, but not all the MTS at full granularity are essential. Kai looks at the report to understand more about the cardinality of different dimensions.

Kai knows that their team cares only about service latency performance for data centers in Europe, so they only filter data where data_center_region = Europe. But, they also want to make sure they have access to recent data in case they want to dig deeper into any other data.

アクション 🔗

Kai decides to use Archived Metrics to control how Splunk Observability Cloud stores their team’s data.

  1. In Splunk Observability Cloud, Kai goes into Metrics Pipeline Management, searches for the metric service.latency and configures the ingestion route to Archived Metrics. Kai can now see all the MTS as Archived MTS.

  2. Kai creates a route exception rule and specifies a filter where data_center_region = Europe. This gives them the estimate of 2,497 Real-Time MTS. Kai also restores the previous hour data to make sure they don’t have gaps.

  3. Now, Kai views the list of charts and detectors that use service.latency. To learn more about viewing or downloading the list, see Understand your metrics usage with the metrics usage report.

  4. Kai already had a filter set up on the charts and detectors for data_center_region = Europe. Kai verifies the data is visible in one of the charts.

  5. Kai revisits the metric service.latency in Metric Pipeline Management to see the MTS estimates again. The estimates now show a 95% reduction in the Real-time MTS count, from 50,000 to 2,497.

概要 🔗

By archiving and routing a portion of MTS to real-time, Kai and Skyler have successfully reduced their overall MTS usage, staying below their usage limits while lowering storage costs for Buttercup Games.

さらに詳しく 🔗

To learn more, see the following docs:

This page was last updated on 2024年11月13日.