Managing servers is hard enough. Doing it with a rack of servers is more difficult. Maintaining a data centre full of servers can provide a significant challenge.
There are lots of different things that need to be provided to manage servers. Providing remote access, changing the boot order in the BIOS settings, and managing the assets in each box.
Now there is an API that will allow you to programmatically inventory your server environment. An extensible API that will support your environment no matter who your server vendor is (or are). Let’s see what this could look like.
Traditionally each server manufacturer has provided a unique and different way of delivering this, despite the IPMI standard. But IPMI was not extensible and provided a lowest-common-denominator feature set, which meant that each and every manufacturer added their own specific functionality using proprietary mechanisms.
Released back in August 2015, Redfish is a more extensible and secure replacement for IPMI. Both are defined by the DMTF (Distributed Management Task Force) It has extended beyond the essential server needs to include PCI switches and storage environments.
Redfish builds on three fundamental architectures that modern coders should know. These are a RESTful API which uses HTTPS for the security and communication and JSON for transmission of data between systems. Using these capabilities means that most developers can already code to extract data from Redfish enabled platforms.
The initial capabilities of the Redfish API provide health data from servers, including state and asset information, as well as temperature, fan and power readings. It gives the ability to reboot and power cycle capability, as well as event log access and notifications.
To get more people involved in Redfish, the DMTF has released a video showing a high-level overview of the Redfish capabilities. The video is alongside an updated Redfish API Schema. The documents located on the DMTF Redfish subsection and the video is embedded below.
So from this, it is theoretically possible to scan each server for it’s installed cards and create a repository of currently connected devices, including serial numbers. With only a few lines of code and the API definitions, your server management headaches may be a thing of the past.
Let’s see if the server manufacturers start adopting Redfish in their baseboard management controllers and push this standard forward. With HP, Dell, Cisco and Lenovo behind the standard, we should shortly see it in the wild.
