Topology¶

LAVA's distributed architecture enables horizontal scaling and flexible deployment across geographically distributed and network-restricted environments.

A LAVA server can be deployed in an isolated lab or in the cloud. LAVA workers can be deployed from any location with network connectivity to the LAVA server.

Local LAVA deployment¶

LAVA server and workers can be deployed on local machines. Several workers can be connected to a LAVA server. Multiple DUTs can be connected to a worker.

graph TD
    server([LAVA Server])
    worker1([Worker1])
    dut1([DUT1])
    dut2([DUT2])
    worker2([Worker2])
    dut3([DUT3])
    dut4([DUT4])

    server <--> worker1
    server <--> worker2

    worker1 <--> dut1
    worker1 <--> dut2

    worker2 <--> dut3
    worker2 <--> dut4

Cloud-hosted Services¶

LAVA server can be hosted in the cloud using docker-composer or K8s. Some companies provide managed LAVA instances.

For example, Linaro offers both cloud-hosted LAVA instances and Linaro Automation Appliance (LAA) that managed by the LAVA Managed Service (LMS) at lavacloud.io. This integrated solution significantly simplifies and accelerates LAVA deployment and device enablement in LAVA.

While the LAVA server and LMS are cloud-hosted, the LAA can be deployed from any location. LMS provides seamless LAVA integration, secure remote access and Over-The-Air (OTA) upgrades for your LAAs.

graph TD
    server([cloud-hosted LAVA Server])
    lms([cloud-hosted LMS])
    laa1([LAA1 + DUT1])
    laa2([LAA2 + DUT2])

    server <--> laa1
    server <--> laa2
    laa1 <-.-> lms
    server <-.-> lms
    laa2 <-.-> lms

Each LAA serves as a pre-configured LAVA worker. It provides everything needed by most hardware automation tasks. The DUT is attached to the LAA via the MIB directly. Simply plug in the LAA power and Ethernet cables, then it will automatically connect to the LMS. You can connect it to your LAVA server and add device configurations with just a few clicks on the LMS web UI.