Spot Instance requests can be either one-time (request only) or persistent (maintain capacity), with options including maximum price, interruption behavior, allocation strategy, launch specifications, and health check replacement.
Spot Instance requests define how EC2 should provision and manage Spot capacity. A request can be a one-time request (type: request) which launches the specified instances once and then the request expires, or a persistent request (type: maintain) which continuously maintains target capacity, replacing instances that are interrupted or terminated [citation:8]. The request includes a launch specification (AMI, instance type, subnet, security groups), a maximum price you're willing to pay, and optional parameters for allocation strategy and interruption behavior.
Type:
one-time(request expires after fulfillment) orpersistent(maintains capacity, replaces interrupted instances) [citation:8]Instance Interruption Behavior:
terminate(default),stop, orhibernate– determines what happens when Spot capacity is reclaimedSpot Price: Maximum price willing to pay; if not specified, defaults to On-Demand price
Valid From/Until: Time window during which the request can be fulfilled
Launch Group: Ensures instances launch together or not at all (similar to cluster placement)
Availability Zone Group: Instances launch in a single Availability Zone
Block Duration: (Deprecated) Formerly allowed guaranteed runtime of 1-6 hours
Tag Specifications: Apply tags to Spot Instances at launch
For Spot Fleets (which are a different resource type), additional configuration options include allocation strategy (lowestPrice, diversified, capacityOptimized), On-Demand portion (mix of On-Demand capacity), health check replacement, and target capacity modification [citation:2][citation:5]. Spot Fleets also support launch template overrides with weighted capacity, priority ordering, and instance requirements for flexible instance selection [citation:3].
Target Capacity: Desired total capacity (units, not necessarily instance count) [citation:2]
On-Demand Target Capacity: Portion of target capacity to fulfill with On-Demand instances [citation:5]
Allocation Strategy: Controls which pools are used for Spot fulfillment [citation:3][citation:5]
Replace Unhealthy Instances: Automatically replace instances that fail EC2 status checks [citation:2]
Excess Capacity Termination Policy:
noTerminationordefault– controls behavior when scaling down [citation:3]Valid Until: Timestamp when the fleet request expires if not fulfilled
IAM Roles:
AWSServiceRoleForEC2SpotFleetrequired for fleet permissions [citation:2]