A county keeps a few bastion hosts that admins use rarely overnight, but each host must preserve in-memory session state when brought back. Which cost-saving compute action fits?
Select an answer to reveal the explanation.
Short Explanation
Bastions that nap all night shouldn't burn full EC2 rates. Hibernate (when the instance type supports it) parks the memory state, cuts the overnight bill, and wakes cleaner than a full terminate/rebuild.
Full Explanation
Hibernating eligible EC2 bastion hosts overnight is the correct cost-saving action when county admins use the hosts rarely at night but must preserve in-memory session state on resume, because hibernation stops compute billing while writing RAM contents to the root volume for later restore. That fits rare-use administrative jump boxes better than leaving them fully running. Terminating bastions nightly and rebuilding from scratch without state considerations fails because termination discards in-memory session state the stem requires to survive, forcing a redesign the city did not ask for. Converting bastions to Spot without interruption notices fails conceptually: Spot can interrupt at any time, and skipping interruption handling is operationally unsafe for administrative access paths even before cost claims. Leaving bastions running 24/7 on the largest memory-optimized size fails by maximizing idle compute spend contrary to rare overnight use. Exam caveat: when state must persist across idle periods, prefer hibernation on supported instance types over terminate/rebuild, unmanaged Spot, or oversized always-on hosts. Operational check: confirm instance types and AMIs support hibernation; enable hibernation; schedule overnight hibernate/resume aligned to admin windows; verify sessions restore; and monitor that stopped/hibernated hours actually reduce the county’s bastion bill.