Rapid Cache recommendations and insights

The Rapid Cache recommender analyzes your data usage and storage to provide recommendations that help you identify bucket-zone pairs in which it might be beneficial to create a cache using Rapid Cache. Rapid Cache also provides insights to show evidence for the recommendations. This page provides an overview of the Rapid Cache recommender and describes how to view recommendations and view insights.

Overview

Rapid Cache provides an SSD-backed zonal read cache for Cloud Storage buckets. To help you decide where to set up a cache and what time to live (TTL) to use, the Rapid Cache recommender provides recommendations for creating caches in bucket-zone pairs.

Reviewing and taking actions on cache recommendations can help you save on network data transfer fees, improve latencies, and avoid exceeding bandwidth quotas.

How it works

The Rapid Cache recommender simulates the existence of a cache and analyzes the cacheability and cost of your data over the past seven days by using the following insights:

• Cache hit rate: the percentage of bytes that would have been served from the simulated cache.

• Peak cache throughput: the amount of throughput that the simulated cache would serve, which represents extra bandwidth over what would be allowed without the simulated cache.

• Net cache cost: the cost of the simulated cache, which is the sum of the cache cost, Class B operations cost difference, network data transfer cost difference, and data retrieval cost difference. A negative value for net cache cost indicates that using caching would save money.

  This value is calculated using the following data:

  • Cache cost: the cost of using the simulated cache, which includes cache ingestion fees, storage fees, data transfer out fees, and read operation fees.

  • Class B operation difference: the cost of performing Class B operations on objects within the cache, minus the cost of performing the same Class B operations without caching.

  • Network data transfer difference: the cost of transferring data out of a multi-region bucket with caching, minus the cost of performing the same data transfer out of a multi-region bucket without caching.

  • Data retrieval cost difference: the cost of retrieving data from the cache, minus the cost of retrieving the same data from a non-cached bucket.

The Rapid Cache recommender recommends a zone-bucket pair when the maximum cache size exceeds 100 GiB and one or more of the following three criteria are met:

• The cache hit rate exceeds 80%

• Net savings on multi-region data transfer out fees exceed $700 per week, based on negotiated price

• The peak cache throughput limit exceeds 800 Gbps

These criteria are set so you only get recommendations for creating caches that are likely to provide significant benefit. Workloads that don't meet these criteria might also have significant benefits from having a cache, so you might want to set your own criteria. You can set your own criteria by exporting the recommendations to BigQuery and then querying the data using the values you want.

Recommendations include the zone name, the recommended time to live (TTL) of the cache, and the data ingestion behavior of the cache.

Pricing

Refer to Recommender pricing.

Before you begin

Before you begin, complete the following steps.

Enable the Recommender API

Enable the Recommender API.

Get required roles

To view recommendations and insights for Rapid Cache, ask your administrator to grant you the Storage Admin (roles/storage.admin) role on the bucket or project.

This role provides a set of permissions that let you view recommendations and insights for Rapid Cache. To see the exact permissions that are required, expand the Required permissions section:

For instructions on granting roles, see Use IAM with buckets or Managing access to projects.

View Rapid Cache recommendations

  curl \
-H "Authorization: Bearer $(gcloud auth print-access-token)" \
"https://recommender.googleapis.com/v1/projects/PROJECT_ID/locations/LOCATION/\
recommenders/google.storage.bucket.AnywhereCacheRecommender/recommendations"

gcloud recommender insights list --project=PROJECT_ID \
--location=LOCATION --insight-type=INSIGHT_TYPE

  curl \
-H "Authorization: Bearer $(gcloud auth print-access-token)" \
"https://recommender.googleapis.com/v1/projects/PROJECT_ID/locations/LOCATION/\
insightTypes/google.storage.bucket.AnywhereCacheSimulationInsight/insights"