Disaster Recovery

Servers:

  • Domain Controllers
  • Domain File Servers
  • Domain DFS Servers
  • Domain Cron Servers

Server Locations:

  • DC1 –
  • DC2 –
  • Poe –
  • VMware –
  • EB2 –

Backups:

  • AD Database – Frequency/Retention
  • Group Policies – Frequency/Retention
  • Central Packages – Frequency/Retention

Dependencies:

  • SAN –
  • Celerra –
  • VMware –
  • Network –
  • Unity Account Provisioning –
  • Data Feeds –

Disaster Scenarios: Due to the nature of the services provided by AD Domain Administrators, the three typical disaster scenarios (loss of facility <30 days, power outage <72 hours, and university technology infrastructure outage <30 days) are not specific enough to adequately explain workaround procedures as AD Domain Administrators works out of 3 server facilities and have UPS/backup generator power, and the infrastructure is very tightly coupled with OIT’s technology infrastructure.

Therefore it was split into 15 scenarios:

  • Loss of Physical Access to Campus: Not Affected
  • Campus Power Outage: Server Infrastructure unaffected, but on-campus client machines and labs would be unavailable.
  • Campus Network Outage Week+: Staff can likely log into workstations and laptops using cached credentials and work on local data; sneaker net data that users need that resides on servers. Without network access, the usefulness of the lab computers and similar would be extremely limited.
  • Poe – Total Loss: Not Affected
  • Poe – Facility Infrastructure Outage <72 hours: Not Affected
  • Poe – Facility Infrastructure Outage 72+ hours: Not Affected
  • EB2 – Total Loss: Not Affected
  • EB2 – Facility Infrastructure Outage <72 hours: Not Affected
  • EB2 – Facility Infrastructure Outage 72+ hours: Not Affected
  • DC1 – Total Loss
  • DC1 – Facility Infrastructure Outage <72 hours
  • DC1 – Facility Infrastructure Outage 72+ hours
  • DC2 – Total Loss
  • DC2 – Facility Infrastructure Outage <72 hours
  • DC2 – Facility Infrastructure Outage 72+ hours

Out of band access (ILOM, DRAC, Cyclades) will be used for hardware management in the event of a loss of physical access to campus.

Communications:

  • AIM will be used for Instant Messaging in the event of a loss of Jabber services.
  • Gmail will be used for Email in the event of a loss of NCSU-provided mail services.
  • Ventrilo (running on a VM in VMware) is a free (up to 8 people) voice chat server that will be used for meetings/discussion in the event of a loss of physical access to campus.

Technical Notes:

  • SAN connectivity exists only in DC1, DC2, and Poe.
  • References of things “going to tape” implies going from the server, to the VTL in DC2, and then to the tape library in MCNC.
  • All instances referring to “Network” implies DNS/DHCP as well.