2018-11-28

Owned by Eric Jansson
Last updated: Nov 29, 2018
2 min read

Participants

  • Brad Bannister
  • Matt Warden
  • Philip Heimes
  • Andrew Rice
  • Josh Allen
  • Erik Joranlien
  • Geoff McElhanon
  • Tarun Verma
  • Chris Moffatt
  • Eric Jansson
  • Cy Jones
  • Shannon Kerlick
  • Sayee Srinivasan

Notes

Gap analysis and development estimates topic: it was recommended that we put ODS API v2.x vs v3.x differences in scope as well (the gap analysis was done on v2.5)

Cost savings information and discussion

  • Several instances of cost analyses for cloud-based deployments were shared, ranging from costs that were 10x to 5x, especially when production instances of cloud-based deployments were necessary
  • It was also noted that collaboratives who operation with one database per ODS API can also experience linear scale costs as they add new LEAs
  • The workaround of running SQL on a VM was discussed, but it was noted that this incurs the management costs, has other less direct costs, mostly in terms of staffing. While LEAs often do not perceive these costs, the goal of cloud deployments is often to save these costs.
  • Consolidation of SQL Instances to meet license requirements can be an option, but introduces complexity

Urgency discussion

  • lack of open source or affordable option is actively suppressing adoption; but there may be workarounds - e.g. installing SQL Server on a VM, using the Cloud Formation Templates; possible to adopt Cloud Formation Templates to better support this

Version discussion

There was a long discussion on which version of the ODS/API is the priority. It was noted that adding support for 3.x is a carrot to move adoptions forward to new versions, but that there is pain for current 2.x implementations. There was seen a need to balance these, but also a need for more information on community needs (see survey action below).

There are also development options for adding support for multiple versions of the ODS /API

  • Modifications to code directly to support another open source database (2 versions of code)
  • One version of code that supports multiple databases via configuration

One estimate from the field placed the latter option as roughly 3X more expensive

Platform choice

Broad agreement on PostgreSQL and MariaDB as the most likely options.

Discussion on topic: are agencies ready for PostgreSQL?
Answer: it doesn't really matter, as there is broad movement away from proprietary technology solutions and we can expect the K12 sector to follow suit; without such support, the ODS API risks looking "archaic"

Actions: do a quick survey on scope of the issue. Key issues

  • Demand for 2.x or 3.x?
  • Interest in cloud/SAAS/IAAS options?
  • Proficiency of agencies with Postgres or MySQL?


Materials