Sync stats data

info.md

Data

To produce the statistics that we're currently interested in, here are the core data points I've identified for each category.

MAU / Device Usage / Retention

• account_id (uid)
• account_created_at
• device_id
• device_created_at
• device_deleted_at
• device_type
• device_last_active_at

device_last_active_at can be derived from the Sync log timestamps.

Example Stats

• Sync MAU by device count
• Sync retention
• % of Sync users by active device count
• % of Sync users by device types
• % of lapsed users by device count
• number of users whose device count changed
• number of users by device count
• time to create second device

Sync Activity

• uid

• device_id

• device_type

• sync_api_method

• sync_api_status

• sync_api_errno

  Example Stats

  • % of users who sync each category
  • time until second device fetches updates
  • number of users affected by critical errors

Logs

The relevant log lines for sourcing this data come from the FxA auth-server and Sync storage-server.

FxA

• Account created event
  • uid
  • created_at
• Device created event
  • uid
  • device_id
  • created_at
• Device deleted event
  • uid
  • device_id
  • deleted_at

Sync

All sync app requests signal activity

• uid
• device_id
• device_type
• path
• status
• timestamp
• errno
• ...

Tables

All the proposed metrics could be produced from a relational model of the log data and standard SQL aggregate functions. I'm not saying we should do it this way, but its the model I'm most comfortable with. The table structure could be something like this:

Users

• uid
• createdAt

Devices

• device_id
• uid
• type
• added_at
• removed_at

SyncActivity

• uid
• device_id
• request_path
• request_started_at
• request_completed_at
• request_size
• response_started_at
• response_completed_at
• response_size
• response_status