Our Group currently has the following version of Healthshare, and Ensemble installed:
Cache for UNIX (Red Hat Enterprise Linux for x86-64) 2015.2.1 (Build 705U) Mon Aug 31 2015 16:53:38 EDT [HealthShare Modules:Core:14.01.7913 + Linkage Engine:14.0.7913 + Patient Index:14.0.7913 + Clinical Viewer:14.0.7913 + Active Analytics:14.01.7913]
We are planning on upgrading all of our instances from 2015.2.1 to the latest version (currently 2018) starting in May.
These instances are responsible for receiving data (i.e., Dicoms, CDAs, ADTs, and a custom JSON format) from site source systems, checking them for consent, extracting whitelisted elements, de-identifying the data, and passing that data downstream to a client database. We store bare minimum patient information on the MPI to check consent. We mostly use custom code for the data normalization and processing. For communication between servers, we use SOAP classes generated in studio. For consent, we use the consent API in HSPI.
The main reason we are upgrading is to get Atelier, the latest vulnerability fixes for HSPI as well as the latest MPIID Data Integrity and configuration Check utilities. There are a few issues involved with the upgrade that we are having trouble reconciling.
Our first issue is that we can only do one upgrade at a time. We could spend several days upgrading all the sites. If we were to perform our updates one at a time, Could the Edge Gateways on the older versions of Ensemble still interact with the newer versions of Healthshare on the shared Registry and MPI instances? The InterSystems documentation seems to suggest that communication over SOAP should not be affected (see Web Services and SOAP section at https://docs.intersystems.com/latest/csp/docbook/DocBook.UI.Page.cls?KEY...). We are worried about communication between the old and new Healthshare out of the box components.
Another potential approach was to do a parallel fresh install of 2018 and to import all custom code to that version. That way we could migrate everything without having to bring the previous instances down until the new ones are ready to go. The issue with this approach is that we are unsure if we would be able to use the database files containing the Audit logs as well as the patient information in the MPI. Will those same databases work across different versions?
There are a few more issues, but those are our primary concerns. Has anyone done a parallel install vs. an upgrade? What are the advantages/ disadvantages to a parallel install? Are there any other considerations?
We believe that...
The advantages of a parallel upgrade seem to be:
- Less downtime since production systems can run until their replacements are ready.
- Easier rollback (if the new instance doesn’t work just turn it off and switch back to the old instances)
- We would need to copy the databases the old instances were using and remap them.
- We would need to reschedule the tasks we had running in the new instances.
- Not sure if the New versions of Ensemble/ Healthshare can use the older database files.
- Not sure if using the old database files will get us the Audit logs
- Not sure if any additional configurations are necessary for a parallel install
Any feedback on this matter would be greatly appreciated.