Repository navigation
Capacity Profiles and Cloudflare Plans
The capacityProfile setting sizes the plugin to your Cloudflare plan and database: how much analytics data it keeps, how much work one background run may do, and how large the daily summaries get. Pick the profile that matches your plan, then grow along the scale-out ladder below.
- Find your plan: Workers Free or Workers Paid.
- Open Settings, section Data and capacity, and set Capacity profile:
- Workers Free: Small (
small). - Workers Paid, analytics in the EmDash database: Standard (
standard, the default). - Analytics in a database of its own (
databaseordatabasesoption): Dedicated (dedicated).
- Workers Free: Small (
- Optional: set Storage budget (MB) in the same section to give the analytics tables more or less room than the profile's default (0 keeps the default). See Storage budget.
- Save. Open the Data page, tab Storage and retention, to see the storage estimate against the budget and the database's size now.
small |
standard (default) |
dedicated |
|
|---|---|---|---|
| Default storage budget | 300 MB | 6 GB | 9 GB |
| Requests per background run | 40 | 400 | 400 |
| Scan budget (visits a detailed report may read) | 50,000 | 500,000 | 500,000 |
| Pair rows per site and day | 340 | 7,500 | 15,000 |
| Cube cells per site and day | 150 | 1,000 | 1,000 |
| Journey step rows and three step paths per day | 200 and 50 | 2,000 and 500 | 2,000 and 500 |
| Daily write guard | yes | no | no |
| Search rows written per UTC day | 21,250 | 500,000 | 1,000,000 |
A quarter of the budget is reserved for summaries (with the default budgets: small 75 MB, standard 1.5 GB, dedicated 2.25 GB); detailed visits use the rest. When detailed data passes its share, the detailed window shortens to the newest days that fit, and older days are answered from daily summaries. Nothing is sampled.
A report over a range that holds more visits than the scan budget is answered from the daily summaries when every filter applies there. Otherwise it reads detailed rows up to 4 times the budget, and beyond that it asks for a shorter range.
storageBudgetMb sets the space the analytics tables may use, in whole megabytes; 0 uses the profile's default. The range depends on the database of the analytics tables:
| Database | Default (0) | Allowed | Warning above | Strong warning above |
|---|---|---|---|---|
Any D1 database, small
|
300 MB | 50 to 450 MB | 300 MB | 400 MB; in the EmDash database also budget plus the rest of the database above 475 MB |
EmDash database (D1), standard or dedicated
|
6,000 MB (9,000 with dedicated) |
100 to 9,000 MB | 6,000 MB | 8,000 MB, or budget plus the rest of the database above 9,500 MB |
Own D1 database, standard or dedicated
|
6,000 MB (9,000 with dedicated) |
100 to 9,500 MB | 9,000 MB | none |
Why these maxima: a D1 database holds at most 10 GB, cannot be raised and refuses writes when full (500 MB on Workers Free). In the EmDash database that space also holds the site's content, so a full database stops the CMS from saving content too.
- The Settings page shows the allowed range under the field and the warning of the typed value as it changes. Saving a value outside the range is refused with the reason.
- The Data page lists the warning under Needs attention, with a link to the setting.
- The Data page shows the database's size now (the whole file, content included): an own D1 database reports it itself, the EmDash database through the Worker's one D1 binding that no database option names (
DBin the template); with none or several such bindings the page says the size is unknown. From 80 percent of 10 GB (500 MB onsmall) it shows a warning, from 90 percent a critical notice. - A new budget applies with the next retention run. When the detailed data does not fit, the detailed window shortens and older days are kept as daily summaries.
The warning texts and the estimate: Limits, Storage budget.
With small, recording stops for the rest of the UTC day once the estimated rows written of the day reach 85,000 (85 percent of the Free plan's 100,000), so the CMS can still save content. While blocked, page views and events get 503 without Retry-After and are counted as budget_blocked. The Data page says: "The daily write allowance of the small profile is nearly used. Recording resumes at the next UTC day." With the typical mix that is about 25,000 page views a day.
| Limit | Value |
|---|---|
| D1 database size | at most 10 GB, cannot be raised; writes are refused when full |
| D1 queries | one at a time per database |
| Workers Free: database size | 500 MB per database |
| Workers Free: rows written | 100,000 per day per account |
| Workers Free: rows read | 5 million per day per account |
| Workers Free: queries per invocation | 50 |
| Workers Free: subrequests per invocation | 50 (measured on a Free account) |
| Workers Free: CPU time per request | 10 ms; waiting on the database does not count |
| Workers Paid: D1 queries per invocation | 1,000 |
| Workers Paid: subrequests per invocation | 10,000 |
Rows written per page view: about 3.3 with a typical mix (3.9 with the later retention delete).
Estimates for 25 months of detail at about 340 bytes per page view, with the default budgets:
| Profile (detail share) | Fits |
|---|---|
standard (4.5 GB) |
about 17,500 page views a day |
dedicated (6.75 GB) |
about 26,200 page views a day |
small (225 MB) |
about 660,000 page views in total (about 22 months at 1,000 a day) |
These are estimates. The Data page reports the estimate for your own traffic.
The analytics tables can live in a second D1 database. That doubles the storage ceiling and keeps analytics writes from queueing behind the CMS.
-
Create the database:
npx wrangler d1 create site-analytics
-
Add it to the site's wrangler config next to the EmDash database:
-
Name the binding in the plugin options and deploy:
emdash({ plugins: [cookielessAnalytics({ database: { binding: "CA_DB" } })] })
-
Set
capacityProfiletodedicated.
What moves: every analytics table (visits, events, error pages, drop counters, summaries, revenue and Search data). What stays in the EmDash database: settings, cursors, salts, Search credentials, goals, funnels, dashboards, API keys and other configuration.
For a Postgres database instead of D1, see Postgres through Hyperdrive.
With client sites, the databases option gives chosen sites a D1 database each:
cookielessAnalytics({
databases: {
default: "CA_DB", // every site not listed below; leave it out to keep them in the EmDash database
sites: { shop: "CA_DB2" }, // site id: D1 binding
},
})Each database has its own 10 GB, writer, storage budget and write guard. Background tasks run once per database and split the profile's requests between them (two databases on small: 20 each). databases cannot be combined with database, and holds at most 500 entries.
-
smallprofile on Workers Free, with the write guard. -
standardprofile on Workers Paid. - The queue tier, for bursts and overload (Queue-Worker).
- An own database (
database) with thededicatedprofile. - Site routing (
databases): one database per busy site. - A second install.
| Key | Default | What it does |
|---|---|---|
capacityProfile |
standard |
Storage budget, requests per background run, summary caps, write guard. |
storageBudgetMb |
0 |
Space for the analytics tables in MB; 0 is the profile's default. See Storage budget. |
rawRetentionDays |
760 |
Days of detailed rows kept; shortens by itself when the budget is reached. See Retention. |
syncSubrequestBudget |
40 |
Requests one Search sync run may make. Keep 40 on Free; about 900 on Paid. |
- The default is Standard, which Workers Free cannot carry. On Free, set Small. The Data page warns ("This install appears to run on the Workers Free plan") after a background run hits the request limit.
- The Free plan's daily row limits are per account. More databases do not raise them, and each write guard counts only its own database. On the Free plan keep one database.
-
dedicatedin the EmDash database defaults to 9,000 MB and shows the strong warning: that profile is meant for an own database. - The storage budget is one setting for the install. While any site's analytics tables are in the EmDash database, its range (up to 9,000 MB) applies when saving; each own database clamps the budget to its own range.
- Every site of a database shares its storage budget. Many client sites in one database shorten the detailed window for all of them.
- A binding named in the options that does not exist at runtime is never replaced by another database: writes are answered 503 and counted as
write_error. - Moving a site to another database does not move its data.
- A dev server outside the Workers runtime has no bindings, so with
databaseset it records nothing. Leave the option out there. - With the queue tier, pass the same
databaseordatabasesoption to the consumer.
Quick-Install, Postgres-Hyperdrive, Queue-Worker, Retention, Data-Page, Multi-Site-and-Sites-View, Settings-Reference, Troubleshooting.
Full reference: https://github.com/infinitnet/cookieless-analytics/blob/main/docs/limits.md
Cookieless Analytics by Infinitnet, Apache-2.0. These pages are generated from docs/wiki in the repository; edit them there, not here.
Start
Reports
- Reports
- Filters and segments
- Search Satisfaction
- Goals and funnels
- Revenue and refunds
- Links report
- Search integrations
- Content and site search
Sending data
Sharing and context
Sites
Admin
Running it