Post 2016 Election Infrastructure
Cluster Teardown
We will be tearing down our elections infrastructure to go light-weight for post election season and maybe through mid-term elections.
- mothball whitefront(stop, do not terminate but make an AMI)
- Move whitefront schema off of multi0 to va-db0
- Decomission multi0
- Remove VA cluster Route 53 entries to prepare for cluster shutdown
- Tear down entire VA cluster (va-www*, va-db*, va-vpn0, util1)
- Clean up salt of VA systems
- Move django schema master to or-db0 from db0
- Setup or-db0 and or-db1 to become new main servers
- Restrict postgres to more limited resources. Use less RAM, mostly
- Install setup web server(s) for handling pypvs, web_api, pvsapi, voteeasy
- Install memcached
- Tune for proper balance between all services
- move balancer to point to new servers
- Move api balancer to point to new servers
- Shut down and terminate or-www*
- Shutdown and terminate api1
Decentralization
We will be moving some key Web apps and other services to AWS from the MT office.
Considerations
Some things we need to keep in mind while coming up with a plan:
- Security - Firewalling needs to be strong, but flexible. We need to consider some form of SSO and integrating a soft of 'knocking' where people can sign in to update our firewall rules to allow access. This would give us the greatest protection and flexibility in case we need to support many offices or remote employees. Plan for flexibility but in the short term we can work with something a little more rigid.
- Portability - what can actually move and what can't. We need to come up with key services and figure out what needs to move to whatever office the new departments are at, and what can be moved to a remote location(AWS).
CategoryITDoc
