Usage considerations

Consider the items in this section when using Semaphore for SharePoint Online.

Platform limitations

• Microsoft throttling limits - From time to time, Microsoft imposes throttling limits on Office 365 tenancies. This can cause the Semaphore classification process to classify documents incorrectly or not classify documents at all. During normal operation, Semaphore makes calls to SharePoint Online to do things like classify content that has recently been uploaded/updated. This may increase the usage of the Office 365 tenancy and result in throttling. If you have a large amount of content to upload for Semaphore to classify, consider spreading the upload over a series of days. This may prevent throttling. See Avoid getting throttled or blocked in SharePoint Online from Microsoft's guidelines.

• SharePoint Online storage quotas - Every Office 365 subscription has a storage quota. The quota is based on the license purchased. You can also specify a quota for each Site Collection. Each quota may limit the amount of additional classification information added to content. In most cases, this will not be a problem. See Manage site storage limits for Microsoft's guidelines.

• Limit the number of library columns configured to use Semaphore - Each column configured to use Semaphore can result in a separate classification request to Semaphore. This can impact performance. In addition, there are also practical limits to the number of columns in a SharePoint library. For example, if there are more than 12 lookup or managed metadata column types in the current library view, you may experience issues attempting to add or display content.

  Note:

  Different column types may be considered lookup columns. The Person or Group column type is one example. See the Microsoft documentation for details.

  In general, keep the number of columns in a document library or list to a minimum. Consider consolidating columns if they share the same Semaphore model. See Manage large lists and libraries for Microsoft's guidelines.

• Semaphore log retention - Semaphore logging information is retained for 30 days. Any log information older than 30 days is automatically removed.

  Note:

  This policy applies to exception information and to standard SharePoint Online term store synchronization, classify list, and deployment log information.

• Progress Data Cloud and Azure Government - Regardless of whether or not it is hosted in Azure Government, Progress Data Cloud supports Microsoft 365 environments but does not support Microsoft 365 Government environments.

• SharePoint Online "Location (Geolocation)" columns - Progress Data Cloud does not support SharePoint Online "Location (Geolocation)" columns.

• File name and path lengths - Semaphore for SharePoint Online restricts the lengths of file names and paths based on existing SharePoint Online limitations. For example, the length of the classification file name must fall within SharePoint Online length limits. Refer to Restrictions and limitations in OneDrive and SharePoint: File name and path lengths in Microsoft support documentation for details.

Classification performance

The Progress Data Cloud tenancy configuration determines the quantity of documents classified within a time period. This is based on the Classification & Language Service (CLS) allowance specified in the Semaphore license. For example, if this is set to 2 cores, then your tenancy can process up to two concurrent requests simultaneously. If 25 simultaneous Classify List operations are submitted in SharePoint Online, only two of those can classify at any one time. The remaining requests are stalled until a classification core is available.

In addition, the time to classify a document depends on a number of other factors:

• How busy your environment is - Activity in other Semaphore services (for example Studio or Semantic Enhancement Server) may impact the performance of classification.

• How busy SharePoint Online is - Because SharePoint Online is a shared service, performance may be impacted by your activities and other Microsoft clients.

• How big or complicated the document is - Every document has to be transferred to Semaphore for processing. The transfer time effects how quickly the document is classified. if a document is large, it takes additional time to extract the textual content before actual classification. A complicated document has multiple sections or partitioned text that will slow down the text extraction process.

• How big or complicated your published classification rules are - More complex information science strategies in model publication result in documents taking longer to classify. Similarly, the more rules (generally directly related to the number of concepts in the model), the loner classification takes.