Google App Engine Datastore

How is App Engine Datastore Health and Performance Determined?

The CloudStatus service reports Datastore health based on its availability to create, read, and delete records. Multiple HQ agents repeatedly perform create, read, and delete transactions. Based on transaction success and performance, the agents collaborate in determining service health and performance metrics.

CloudStatus uses a simple set of data to test create/delete/read times. We refer to the combination of all these entities as a 'User Group'.

• Root of the entity group, small record representing a 'user'
• Child of the root, large record (512KB) representing a 'profile picture'
• Multiple children of root, medium records (2KB) representing 'blog posts'

Datastore Metrics

• Create Time— is measured for a transaction that creates a User Group
• Delete Time— is measured for a transaction that deletes a User Group
• Read Time— is time it takes to find and read a User Group. Users are randomly selected, and the user key is used to look up the user and profile picture records. Posts are found via a GQL ancestor query.

What is Service Health?

• Healthy—indicates full health and satisfactory performance
• Issues—indicates partial health, instances of less than satisfactory performance, or both
• Service Failure—indicates service not available, or poor health

Other App Engine Services

• Engine
• memcache
• URLFetch

Close Window