Skip to main content

Topology

A topology defines the hardware that is simulated. A topology consists of one or more data centers. Each data center contains a power source, an optional battery, and one or more clusters of hosts.

topology
└── datacenters
├── powerSource
├── battery
└── clusters
└── hosts
├── cpu, memory, gpu
├── cpuPowerModel, gpuPowerModel
└── cpuDistribution, gpuDistribution

A topology can be written directly in the experiment, or in a separate file that the experiment imports with importFrom (see Experiment).

Topology​

FieldTypeDefaultDescription
datacenterslist of data centersrequiredThe data centers in the topology.

Data center​

FieldTypeDefaultDescription
namestring"DC"Name of the data center.
clusterslist of clustersrequiredThe clusters in the data center.
powerSourcePower sourceunlimited power, no carbon traceThe power source used by the data center and its connected carbon model.
batteryBatteryno batteryA battery that can store energy for the data center.

Cluster​

FieldTypeDefaultDescription
namestring"Cluster"Name of the cluster.
countinteger1The number of identical clusters to create.
hostslist of hostsrequiredThe hosts in the cluster.
Older topology files

Topologies that list clusters at the top level, with a power source and battery per cluster, can still be read. OpenDC puts the clusters in a single data center and logs a warning. Only the power source and battery of the first cluster are kept.

Regarding naming

Users can provide names for Data Centers, Clusters, and Hosts.

This can be done to better understand the generated output. If a name is used multiple times, OpenDC will add an increasing suffix in the form name-i.

Example​

A data center with two clusters. The first has 16 identical hosts, the second has 4 hosts with a GPU:

{
"datacenters": [
{
"name": "DC1",
"powerSource": {
"carbon": { "type": "named", "name": "carbon_traces/NL_2021-2024.parquet" }
},
"clusters": [
{
"name": "cpu-cluster",
"hosts": [
{
"name": "cpu-host",
"count": 16,
"cpu": { "coreCount": 32, "coreSpeed": "2.1 GHz" },
"memory": { "size": "256 GiB" },
"cpuPowerModel": { "type": "linear", "idlePower": "100 W", "maxPower": "350 W" }
}
]
},
{
"name": "gpu-cluster",
"hosts": [
{
"name": "gpu-host",
"count": 4,
"cpu": { "coreCount": 16, "coreSpeed": "2.4 GHz" },
"memory": { "size": "128 GiB" },
"gpu": { "coreCount": 5120, "coreSpeed": "1.5 GHz", "memory": "32 GiB" }
}
]
}
]
}
]
}