This blog post discusses how to store large binary data, such as PDFs and PowerPoint presentations, in Couchbase using a custom chunk manager. The goal is to reduce the number of systems and licenses needed, while benefiting from Couchbase's replication features. The architecture involves two buckets: `cmismeta` for metadata and `cmisstore` for blobs. When a document is inserted into a folder, metadata is stored in `cmismeta`, and the blob is chunked and stored in `cmisstore`. Each chunk has a unique key, and the number of chunks depends on a parameter set by the project. The post also describes how to retrieve the file from Couchbase by concatenating each part and sending it to a stream. The workbench tool provided by Apache Chemistry can display the document content, including its mime type. The architecture and implementation are designed to be scalable and efficient, allowing for large binary data storage in Couchbase.