data store scale-out

Google Groups archive
1

hello,

I’m interested in learning if and how the Orthanc can be “scalable” and particularly how it can scale-out its data store

the following questions are not restrictive, they are just for helping to define the area of my concern about the Orthanc ecosystem (DICOM server and plugins):

  • does Orthanc extract info from the DICOM files (meta-data) and store it in order to handle requests about the DICOM files?
  • is it able to store the meta-data into relational databases or into clustered RDBMs or into NoSQL databases?
  • can Orthanc handle partitioning/sharding of meta-data across multiple nodes?
  • can Orthanc be optimized for big data paradigms?
  • adding store capacity for the meta-data is as simple as adding a node and gaining proportional performance gain?
    best regards,
    Nikos
2

Hello,

There are already many threads on this forum about the scalability of Orthanc, please make a search.

An optimization related to C-FIND querying is currently in progress, but handling C-STORE and C-MOVE is already fine in the presence of large amount of data, as long as you use the PostgreSQL plugin:

http://book.orthanc-server.com/faq/troubleshooting.html#performance-issues

Regarding the other questions:

  • Yes, Orthanc extracts so-called a subset of so-called “main DICOM tags” (such as the Patient Name or Study Description), and indexes them into its database, in order to speed up querying.
  • Orthanc Plugins can monitor the arrival of DICOM files, and send information to NoSQL database. For instance, check out the contributed MongoDB plugin by Doc Cirrus: http://book.orthanc-server.com/plugins.html#index-of-the-contributed-plugins
  • Yes, partitioning/sharding can be achieved by using the official PostgreSQL plugin, and properly configuring the underlying PostgreSQL server: http://book.orthanc-server.com/plugins/postgresql.html
  • Yes, use the PostgreSQL and/or S3 plugins: http://book.orthanc-server.com/plugins.html
  • I don’t understand the question “adding store capacity for the meta-data is as simple as adding a node and gaining proportional performance gain?”
    HTH,

Sébastien-