Meeting Minutes 26th April 2013, Approved on 3rd May 2013
- Roll: Martin, Patrick, Chet, Greg, Robin, Zhexuan, Jacques, Scott Agenda: V2 Approved as posted.
12th April 2013: https://www.oasis-open.org/apps/org/workgroup/tab/email/archives/201304/msg00016.html Motion: Patrick, 2nd Jacques approve the minutes of 12 April 2013 passed w/o Agreed to post to wiki
AI: All outstanding
Report to Board on Spec Quality:
- Zhexuan has prepared some slides to cover the action item from the board to improve quality of specifications
- - TAB review check list by Jacques (July 2012) - TAB rotation to review specifications restarted (April 2012)
- Future items: Guidelines for chairs, Guidelines of using JIRA
Cloud Migration Plans:
https://www.oasis-open.org/apps/org/workgroup/tab/email/archives/201304/msg00023.html Couldn't hire a sys admin. Hybrid was not feasible public cloud with persistence - auto fail over recovery Cost is a driving factor with limited budget Significant project to get away from home grown infrastructure No new features, just a migration, some apps may be upgraded (e.g. new JIRA) some legacy apps can be kept just migrated, and after migration can be updated. Jacques: what about lock-in - porting to a new cloud Scott: not a prime concern porting will have a cost, but trying to stick to open non-proprietary technology Patrick: do we need 11vms for 1.5GB Greg: probably a mistake in the spreadsheet - should be TBs Backup will be to offsite which incurs an outbound cost. Patrick: how much outbound bandwidth do we use at the moment Greg: whatever the cap is! Benefit of multiple Vms is each can have its own bandwidth which can scale up if necessary. Patrick: how much traffic. Can we look at the apache logs re: traffic? Greg: its so old probably not providing reliable data, but can make approximation Oasis not a high volume operation. Patrick: knowing the numbers would help better understand what high and low means and helps to give opinions on the architecture wrt load and demand. Scott: the info was given to contractors and they took this into account. Greg: better to get real accurate data once migrated and then look at optimal deployment. Scott: have struggled in the past, esp with old apps and monitoring and data, so a new scalable environment is better. Robin: more important things to worry about than how many vms. Jacques (Fujitsu): I can conceive that 1 app per VM makes the independent management of these apps more convenient
- (scale the apps independently, recover without affecting each other etc.)
- circulate to the Board process SC as a precursor to getting feedback from the chairs list.
- Next meeting next week.