AI appliances such as Huawei Ascend Claw and the Zhongke Controllable “Lobster Box” bring powerful automation to enterprises. Traditional monitoring can compete for host resources and lose visibility when an appliance's operating system hangs because of heat or hardware failure.

Use an independent out-of-band management channel

CloudSino DCOS communicates with dedicated management controllers such as iBMC, iLO and iDRAC. This creates a monitoring path independent of the appliance's operating system and production network.

Agentless monitoring

  • No monitoring software, drivers or scripts need to be installed on the appliance.
  • The management controller performs hardware data collection independently of business CPU, memory and disk I/O.
  • Avoid monitoring-agent compatibility issues that can destabilize the host environment.

Keep visibility when the operating system fails

  • Use the out-of-band channel to inspect temperature readings, fan failures and other evidence surrounding a crash.
  • Send a hardware reset command remotely when the operating system is unresponsive.

Understand temperature and energy use

  • Monitor appliance current and generate device and cluster energy-efficiency curves with DCOS analysis.
  • Track each fan's speed and health to identify signs of degradation.

As AI appliances become valuable business assets, their monitoring needs to remain available during failures. DCOS provides an independent hardware management path for routine monitoring and remote recovery.

Article image 1
Article image 2
Article image 3