<< Click to Display Table of Contents >> Raynet One Data Hub > 2026.2 > Connectors > Alphabetic Connector List > Microsoft 365 > Connector Parameters  Sign-ins Configuration

The Microsoft 365 connector supports the extraction of sign-in activity data from the Microsoft Graph API. Because sign-in data volumes vary greatly between tenants — from a few thousand records to tens of millions — the sign-ins feature provides a rich set of parameters for tuning extraction performance and reliability.

 

Sign-ins data is written incrementally to a CSV file using a batched approach to keep memory consumption within acceptable bounds, even for very large data sets.

 

How Sign-ins Extraction Works

When sign-ins collection is enabled, the connector divides the configured time range into smaller time chunks (intervals). Each chunk is queried against the Microsoft Graph API independently. This chunked strategy offers two benefits:

 

•The connector can issue multiple chunk queries in parallel, reducing total extraction time.

•Individual chunks are small enough to stay within the API rate limits.

 

As records arrive, they are accumulated in memory. Once the in-memory batch reaches 100,000 records, the batch is written to a temporary CSV file and cleared from memory. Duplicate records are detected and discarded using a hash set that persists for the entire run. This deduplication mechanism is essential when parallel requests overlap at chunk boundaries.

 

If the Microsoft Graph API returns a throttling response (HTTP 429) or another transient error during a chunk query, the connector automatically retries the chunk after a configurable delay. The number of retry attempts is bounded by the signInsMaxResumes parameter.

 

Small Data vs. Big Data Scenarios

The appropriate configuration depends heavily on the size of the tenant's sign-in history.

 

Small Data (up to ~500,000 sign-ins)

For tenants with a relatively small number of sign-ins, the default parameter values are sufficient. No special tuning is required. The extraction will complete quickly and consume minimal memory.

 

Recommended settings for small data scenarios:

•signInsInterval: 60 (default)

•signInsParallelRequest: 1 (default)

•signInsPageSize: 1000 (default)

•signInsIntervalMinDuration: 2000 (default)

 

Big Data (more than ~500,000 sign-ins)

For tenants with millions of sign-ins, the connector must be tuned carefully to avoid excessive memory usage, API rate-limit errors, and very long runtimes. The chunked batch-writing mechanism ensures that memory peaks remain manageable, but the interval size and parallelism must be balanced against the API's throttling limits.

 

Recommended adjustments for big data scenarios:

•Reduce signInsInterval to a smaller value (e.g., 15 or 30 minutes) so that each chunk covers a narrower time window and returns fewer records per request.

•Increase signInsIntervalMinDuration (e.g., 3000–5000 ms) to introduce more spacing between parallel requests and reduce throttling risk.

•Keep signInsParallelRequest at 1 initially and increase it only if the extraction is too slow and no throttling errors occur.

•Increase signInsMaxResumes to handle the higher probability of transient errors on large data sets.

 

 

Configuring Test and Production Tasks

It is recommended to maintain separate task configurations for testing and production environments.

 

Test Task

A test task is used to verify that authentication is working correctly and that sign-ins data is being returned in the expected format. Configure the test task to extract only a short time window to keep the runtime short and minimize API load.

 

Recommended test task settings:

•signInsRange: si_period

•signInsPeriod: 1 (one day)

•signInsParallelRequest: 1

•ignoreSignInsErrors: false (so errors are visible immediately)

 

Production Task

A production task is scheduled to run regularly and extract the full sign-in history for the configured time range. Once the test task confirms that data is returned correctly, the production task can be configured with the appropriate range and performance settings for the tenant's data volume.

 

Recommended production task settings (adjust based on data volume):

•signInsRange: si_period with a period matching the scheduled run interval (e.g., 7 days for a weekly job)

•ignoreSignInsErrors: true (allows partial results to be written even if some chunks fail)

•signInsInterval: tune based on data volume (see Small Data vs. Big Data section above)

 

Sign-ins Parameters

 

Fetch Sign-ins

Enables or disables the collection of sign-in activity data. When set to false, all other sign-ins parameters are ignored and no sign-in data is extracted.

 

 

Ignore Sign-ins Errors

Controls whether errors encountered during sign-ins extraction cause the entire connector run to fail. When set to true, the connector logs the error and continues, writing any successfully extracted records. When set to false, any error during sign-ins extraction causes the run to fail immediately. It is recommended to use false during initial setup and testing, and true in production to ensure partial data is always written.

 

 

Sign-ins Range

Defines the time range strategy for sign-ins extraction. Three modes are available:

 

•si_all — Fetches all sign-in data available via the API. This mode can take a very long time on large tenants and should only be used for an initial full load.

•si_startend — Fetches sign-ins within an explicit start and end date/time, specified by the signInsStartDateTime and signInsEndDateTime parameters.

•si_period — Fetches sign-ins for the last signInsPeriod days, counted from the time of the connector run. This mode is recommended for recurring scheduled tasks.

 

 

Sign-ins Start Date/Time

The start of the extraction window when signInsRange is set to si_startend. The value must be prefixed with the letter D followed by a UTC date/time in the format yyyy-MM-ddTHH:mm:ssZ. This parameter is ignored when signInsRange is not si_startend.

 

 

Sign-ins End Date/Time

The end of the extraction window when signInsRange is set to si_startend. The format is identical to signInsStartDateTime. This parameter is ignored when signInsRange is not si_startend.

 

 

Sign-ins Period (in days)

The number of days back from the current time to use as the extraction window when signInsRange is set to si_period. Valid range: 1 to 365. This parameter is ignored for other range modes.

 

 

Sign-ins Interval (in minutes)

The length of each time chunk in minutes. The connector divides the total time range into chunks of this size and queries each chunk separately. A smaller interval means more API requests but fewer records per request, which reduces the risk of hitting API limits for high-volume tenants. Valid values: 1440, 720, 480, 360, 240, 120, 60, 30, 20, 15, 12, 10, 6, 5, 4, 3, 2, 1.

 

 

Sign-ins Minimum Interval Duration (in ms)

The minimum time in milliseconds that each chunk query must take. If a query completes faster, the connector waits for the remaining time before issuing the next query. This parameter controls the effective request rate and helps prevent throttling. The Microsoft Graph API has an observed limit of approximately 2 requests per second and 30 requests per minute for sign-ins. Valid range: 50 to 10,000.

 

 

Sign-ins Parallel Requests

The number of time-chunk queries to execute in parallel. Increasing this value reduces total extraction time but also increases the request rate, which raises the risk of throttling. It is recommended to start with the default value of 1 and increase only after verifying that the extraction completes without throttling errors. Valid range: 1 to 8.

 

 

Sign-ins Page Size

The number of records returned per API page within a single chunk query. A larger page size means fewer round trips per chunk but larger individual responses. Valid range: 100 to 1000.

 

 

Sign-ins Page Delay (in ms)

The delay in milliseconds between successive page requests within a single chunk query. Increasing this value reduces the request rate for multi-page chunks, which can help avoid throttling on densely populated time windows. Valid range: 50 to 1000.

 

 

Sign-ins Maximum Resumes

The maximum number of times the connector will retry a failed chunk query before giving up on that chunk. Transient API errors and throttling responses are retried automatically. On large tenants, up to four retries per chunk have been observed. Valid range: 2 to 16.

 

 

Sign-ins Resume Delay (in seconds)

The time in seconds the connector waits before retrying a failed chunk query. Increasing this value gives the API more time to recover from transient overload before the next attempt. Valid range: 1 to 16.