October 10 2017
Last Meeting Notes: September 26 2017
AGENDA
Item | Lead (Time) | Notes |
---|---|---|
Sneak Peak: OpenLMIS pitchfest for TechNet Conference in Portugal. It is a five minute pitch, Vidya Sampath has graciously agreed to give us a preview. If you join the call on time we will kick it off with a quick OpenLMIS pitch. We will share the script and visuals with the community incase others want to leverage for other similar opportunities | Vidya Sampath (5 minutes) | The goal of the pitchfest is to share the aspirational vision of OpenLMIS and spark interest within the EPI participants. We want to increase awareness of the OpenLMIS software/community and the upcoming features to support managing immunization supply chains. |
Software Development Update
Overall updates and topics under discussion:
| Mary Jo Kochendorfer (Deactivated) (15 minutes) | To achieve the short term roadmap (to demo initial features by Jan and release by March) we:
|
OpenLMIS Reporting for the vaccine DISC indicators We need a reporting solution to demo and meet the needs for DISC indicators. Leveraging what the community has done for defining the reporting needs here: Reporting and Analytics, we believe the following proposal is a solid approach and foundation for future needs (i.e. the gap analysis and more). Proposal is here: Reporting and Data Warehouse Strategy | Matt Berg and Peter Lubell-Doughtie (Unlicensed) (10 min) | Committee members were asked to pre-read the proposal and come with questions. OpenLMIS Reporting for the vaccine DISC indicators: Matt Berg: We have realized we need a better strategy for reporting. We have been researching to implement industry standards. There have been a lot of advancements following this reporting and DW strategy approach. The main idea is “can we have a way of taking data out of systems like OpenLMIS” and have a strategy for feeding the data into data warehouse.
Any questions: Ashraf: This architecture looks like a reasonable approach has been applied. Is there any evaluation that you have for each of the items, that explains the decision making process? Matt B: Kafka and Hadoop are the standard. Everyone uses it for the data pipeline and data storage. Druid has emerged as the OLAP db. But there are alternatives like NIFI. We can share analysis. Superset is the best open source equivalent to Tableau that we’ve found. It integrates with OLAP and ODBC databases. These tools are designed to work together so we can leverage the integrations that we don’t have to build. Ashraf: Is there going to be some time lag when extracting data, has this been considered? Matt: This is dependent on the business requirements. This is designed to support true real-time. As soon as you ingest the data, you can set it up to reflect what the entire aggregate could be. We can extract it out or push it into it, we just have to define how real time we want these tools to be. Peter: They can be run at a lower cost if they are run at batch intervals. Matt: If we want to be able to respond to a stock out alert or poor performance, we have to track down at different levels based on the trends of data. Ashraf: For the visualization of Superset, OpenLMIS has its own way of garnering the permissions. How do we avoid recreating a different set? Matt: We will spend some time next week discussing security schemes. The hope is that we can integrate with it so that we don’t have multiple sets of verification. Parambir: It would help if in the field there was some kind of assessment where we asked what kind of resources were needed (hardware, software). Additionally for the reporting analytics, what are the additional resources required? Even though a lot of this is open source, what is required to keep this going so that people can start putting a monetary aspect to it and assign it a budget. Matt: What are the requirements needed for this (server resourcing, devops), this is packaged to run on one machine. Peter: On the devops side we are building it with as much automated and consistent as possible. Parambir: The adhoc reports and skills necessary to build them will be a challenge. Matt: We are excited by Superset because of the ability to explore the data and generate new visualizations, with training, is more accessible. Once you’ve defined the data set. These tools are designed to run on a massive scale, but you could combine them down if you are ok with giving up the service level requirement. Ashraf: About the interactive maps; was there other industry standards that more widely use interactive mapping tools before deciding on Ona specific one? Matt: Action Item: I can send some slides to share. We are leveraging mapbox which provides a tool creating visual maps. It’s the industry standard for creating web maps that are open source. |
Upcoming UI Changes | 10 minutes | |
Share our published documentation on the current Release Candidate Process. We anticipate this evolving when more teams are working in parallel. | ||
Upcoming conferences:
| TechNet Details
|
ACTION ITEMS:
- Mary Jo Kochendorfer (Deactivated) Schedule a product committee call to discuss the tools used for tracking tests on OpenLMIS
- Mary Jo Kochendorfer (Deactivated): Schedule time for PC to define what metrics we want to see, and this will support the team in prioritizing what to focus on next for performance testing.
ATTENDANCE:
RECORDING:
Video: OpenLMIS Product Committee Meeting-20171010 1552-1.arf (download the WebEx video player here)
ADDITIONAL READING:
Long-term Roadmap
Thoughts? What's missing?
- Some of the items are overlapping but trying to find a way to demonstrate the longer term goals within features, performance, standards and integrations.
OpenLMIS: the global initiative for powerful LMIS software