For Data Center and AI Infrastructure Ops

Details

How do you keep hundreds of console servers on one baseline?

Group them by policy, then schedule the work. Percepxion Smart Groups pair a membership policy (which console servers belong) with an assignment policy (what firmware and configuration they get). Build groups around however you run the floor: by hall, by row, by customer, by rack design.

Rollouts become jobs. Percepxion schedules operations across the fleet, manages firmware and applies config templates, and it backs up and restores each console server’s own configuration. A firmware campaign across a hall is one scheduled job, scoped to the group that needs it.

New capacity joins the baseline without hand configuration. When a new SLC 9000 calls home over its outbound-only MQTT connection, Percepxion evaluates the Smart Groups and applies the matching policy. That’s zero-touch provisioning, and it means the four-hundredth rack matches the first.

What does a dense AI rack need from the console layer?

Port density, fast uplinks and redundant power in one rack unit. The SLC 9000 ships in 16, 24, 32 and 48 serial port configurations, in RJ45 or USB serial, with Ethernet switch-port options up to 32 ports. Every unit has two 1GbE uplinks and two 1/10G SFP+ uplinks with failover, and dual AC or dual DC supplies. Configurations are factory-fit, so pick a standard SKU per rack design and order to it.

Console access is vendor-neutral, so the switches, servers and appliances in the rack all land on the same ports. The SLC 9000 logs port data in real time to local storage, NFS, syslog or a USB drive, which gives you the console history for a postmortem without anyone having had a session open. And when a node or switch drops off the fabric, the console path doesn’t depend on it.

How does a large ops team work it day to day?

From a browser, scoped to what each person owns. Percepxion opens a console to any attached device with no VPN and without knowing the device IP. Access runs through SAML 2.0 SSO with Okta, Azure AD or any SAML 2.0 provider, plus RBAC and MFA. Projects, portals and organizations split the fleet between teams or tenants, with per-operation permissions.

Scripts still get direct paths. Console port N answers SSH on TCP 3000+N. And for questions across the fleet, the Percepxion MCP server gives an MCP-compatible AI client 38 tools to query and work the console layer at scale.