wrench::ServerlessComputeServiceProperty

class ServerlessComputeServiceProperty : public wrench::ComputeServiceProperty

Configurable properties for a ServerlessComputeService.

Public Static Attributes

static const std::string CACHE_IMAGES_ON_HEAD_NODE

Whether remote images (i.e., layers) should be first downloaded to the head node’s storage, and cached there for fast copies to compute nodes, or whether compute nodes should download remote images themselves. (“true” (default) or “false”).

static const std::string COMPUTE_NODE_ACTIVATION_OVERHEAD

The compute node activation overhead; activating multiple nodes at once incurs this overhead only once (default value: “0s”) Examples: “5s”, “5000ms”, etc.

static const std::string CONTAINER_IDLE_TIMEOUT

The timeout before an idle container is terminated and its disk/resources reclaimed.

static const std::string CONTAINER_STARTUP_OVERHEAD

The overhead to start a container after the image has been loaded into RAM (default value: “0s”, default unit: seconds): Examples: “5s”, “5000ms”, etc.

static const std::string FUNCTION_STARTUP_OVERHEAD

The overhead to spawn a function execution on a container (default value: “0s”, default unit: seconds): Examples: “5s”, “5000ms”, etc.

static const std::string INVOCATION_PROCESSING_OVERHEAD

The overhead for the compute service to process an incoming request (default value: “0s”, default unit: seconds): Examples: “5s”, “5000ms”, etc.

static const std::string MAX_NUM_CONCURRENT_INVOCATIONS_PER_CONTAINER

The maximum number of function executions that can execute concurrently in a container (default value: 1).

static const std::string MAX_NUM_CONCURRENT_INVOCATIONS_PER_NODE

The maximum number of function executions that can be started/running per compute node (default value: “0”, which means: same as number of cores per compute node).

static const std::string MAX_NUM_CONTAINERS_PER_NODE

The maximum number of containers that can be starting/running/idling per compute node (default value: “0”, which means: same as number of cores per compute node).

static const std::string MAX_OUTSTANDING_INVOCATIONS

The maximum number of outstanding invocations, i.e., invocations that have not yet completed or failed, including those waiting for image preparation or execution. (must be > 0; default value: “infinity”).

static const std::string MAX_OUTSTANDING_INVOCATIONS_PER_FUNCTION

The maximum number of outstanding per-function invocations, i.e., invocations that have not yet completed or failed, including those waiting for image preparation or execution for a particular function. (default value: “infinity”).

static const std::string NUM_COMPUTE_NODES_INITIALLY_ACTIVATED

The number of compute nodes are initially all activated (“all” means all are activated) (e.g., “0”, “all”, “12”).

static const std::string SIMULATE_REMOTE_IMAGE_DOWNLOADS

Whether the download of remote images should be simulated or not (if not, then zero time is assumed). Setting this to false can create strange re-ordering of initial requests at time zero (because so much happens exactly at time zero), which likely doesn’t matter a lot, but be warned. It’s likely better to simulate remote image downloads and just make the remove bandwidth very high when wanted to essentially ignore remote downloads.

static const std::string STORAGE_SERVICES_BUFFER_SIZE

The buffer size used by all internal storage services.