| 7 | |
| 8 | Disaster (in)Tolerance discussion |
| 9 | |
| 10 | __Targets__ |
| 11 | * https://support.mayfirst.org |
| 12 | * offsite backups |
| 13 | * service advisories |
| 14 | * single point of sending/updating that is colo independent |
| 15 | * multple upstream routes fail over options |
| 16 | * bring back all servers without physical visit to the colo or data compromise |
| 17 | |
| 18 | __Solutions__ |
| 19 | * multiple upstream routes |
| 20 | * GSM (or other mobile) modem in each colo |
| 21 | * second uplink crossconnect |
| 22 | * direct interconnect between cabinets |
| 23 | * SMO |
| 24 | * db and config replication for SMO, with manual cutover and cutback scripts |
| 25 | * read-only, non-login copy of SMO with modified template that makes it clear that trouble is a-brewin' |
| 26 | * on-failure-display |
| 27 | * backups |
| 28 | * big beefy disk array servers |
| 29 | * member CPE to provide other backup services |
| 30 | * get mf/pl cable link at the lair |
| 31 | * move sittingbull to the lair |
| 32 | * UPS-backed low-power machine as console server, switch backed as well. |