A Viewing Package is a cached version of a document that the Viewer will use when viewing a document. Viewing a document from a Viewing Package significantly reduces the load on PrizmDoc Server and allows you to serve many more users per minute than you could otherwise. A Viewing Package can be created through Pre-Conversion or by using On-Demand Caching.

This topic provides information about the following:

Storage

Viewing Packages are stored in both the filesystem and configured database. If using multiple instances of PAS, you must use a shared database and NAS (Network Attached Storage). The Running PrizmDoc Application Services (PAS) on Multiple Servers topic can provide more information for configuring PAS in Cluster Mode.

By default, storage is configured in the following way:

Config Key	Storage Provider	Description
viewingPackagesData	database	Data about a Viewing Package. This is the data that can be retrieved from GET /v2/viewingPackages.
viewingPackagesProcesses	database	Data about a Viewing Package creator process. This is the data that can be retrieved from GET /v2/viewingPackageCreators.
viewingSessionsData	database	Data about a Viewing Session. When creating a Viewing Session, an entry is added to this table.
viewingSessionsProcessesMetadata	database	Data about processes for a Viewing Session. This is currently used for content conversion and markup burner processes.
viewingPackagesArtifactsMetadata	database	Metadata for a viewing package artifact. This is used to find specific artifacts for a package and contains the artifact type, the file name in the filesystem among other important information.
viewingPackagesArtifacts	filesystem	Artifacts for a Viewing Package. These include SVG and raster content for every page, the source document, and other artifacts the Viewer will likely request.

Configuration

Viewing Packages are opt-in and require special configuration to work properly. At a minimum, your configuration should include the following:

Pre-Conversion

Creating a Viewing Package through Pre-Conversion provides a way to generate packages whenever it makes the most sense for your application to do so. It allows you to make use of down-time for Pre-Conversion to reduce load in high traffic periods. Pre-Conversion does all the work of creating a Viewing Package whenever it is requested. It starts a process that will begin downloading content and allow you to poll for progress.

It is recommended that you maintain a queue of Pre-Conversions so you don’t overload the server and have faster turnaround time. We recommend a maximum of 5 Pre-Conversion processes at a time per PAS and PrizmDoc Server instance. This will allow packages to be created quickly while maintaining a sustainable load.

For a step-by-step process, go to the Pre-Converting Documents topic.

On-Demand Caching

Creating a Viewing Package through On-Demand Caching is a seamless process through the Viewing Session API. On-Demand Caching allows you to trigger a Viewing Package creation process in the background and use the resulting Viewing Package when it is ready. This feature is designed to allow immediate viewing using PrizmDoc Server while caching a package for subsequent views of the same document.

As an example, consider this request for a Viewing Session:

This request will always return a viewingSessionId regardless of the status of the matching Viewing Package. If a Viewing Package does not currently exist with the given documentId, PrizmDoc Server will handle document viewing while a background process creates a Viewing Package. Once the background process is complete, PAS will handle all further viewing sessions until the Viewing Package expires (24 hours by default).

Performance Considerations

When enabling the Viewing Packages feature, there are additional considerations for hardware requirements. See Sizing Servers for more information.

Known Issues

Document Comparison

Viewing packages are not supported for comparison viewing sessions.

Markup & Redaction

When creating a Viewing Package and pages of the source document have successfully converted to the intermediate PDF, but PAS fails to get one or more pages of viewable content, the Viewer will not have that failed page content available to markup (i.e., redact) . Because of this, when the user attempts to burn the markup into the output document, (which is generated from the intermediate PDF), it will have the page content and no markup.

If this occurs and the failed page contains content that must be redacted, a new viewing package should be created to retry the conversion.

Raster Content

Viewing Packages limit available raster content. During Viewing Package creation, raster content is requested only for 0.125 , 0.25, 0.5, and 1.0 scale. API requests for raster content with a scale that is not one of these values will return 404 Not Found. The Viewer will always request full scale raster content from a viewing package which may result in degraded fidelity since the full scale image will be returned and any additional scaling will occur within the browser. Additionally, requests for tiling and thumbnails are not supported and will return 501 NotImplemented.