Hardware Management/SpecsAndDesigns: Difference between revisions
mNo edit summary |
|||
| Line 82: | Line 82: | ||
[http://files.opencompute.org/oc/public.php?service=files&t=c3ee8db5c9e64c1ddecc71fe084afaf0 Firmware Update Requirements version 0.2b PDF] | [http://files.opencompute.org/oc/public.php?service=files&t=c3ee8db5c9e64c1ddecc71fe084afaf0 Firmware Update Requirements version 0.2b PDF] | ||
===Cloud Server Multi Node System Specification=== | |||
A description and requirements of a Cloud Server which is either a server, enclosure or rack that contains other uniquley addressable and managable devices. | |||
The document's purpose is to have a single Ethernet connection to the Cloud Server and document the possible bus architectures, managment protocols and minimum expected functionality of the Nodes contained in the Cloud Server. | |||
The Clould Server's BMC with a Ethernet connection to the outside world is the called the Spine BMC and both IPMI RMCP or RESTfull interface are allowed. Internally the Spine BMC communicates using | |||
The Node requirements include a subset of the IPMI 2.0 commands. The Spine BMC may have a Serial Console session to each Node. | |||
[http://files.opencompute.org/oc/public.php?service=files&t=1eb58f54e79a48f0007de6b7379a093f Cloud Server Multi Node System Specification version V0.7.5, Date Augest 19,2015 PDF] | |||
Revision as of 00:58, 20 August 2015
Specs and Designs
This page contains links to the Specs ad Designs for the OCP Hardware Management Project
LAVA LMP
More information about LAVA LMP can be found at:
- https://github.com/opencomputeproject/lava-lmp-firmware/subscription
- https://github.com/opencomputeproject/lava-lmp-lmpd/subscription
DRAFT of the Open Hardware Machine Management Specifications v1.01 (PDF)
OCP Hardware Management Specification: Remote Machine Management v1.01 Draft: 02-25-2014
Charter for the Open Hardware Management Track (PDF)
Charter for Open Hardware Management Track Draft: 12-28-2011
DRAFT Charter for Open Hardware Management Track w/Japanese Translation (PDF)
Proposed track for Open Hardware Management
DRAFT Hardware Management Specifications for IPMI
Proposed specifications for the IPMI implementation on any device using IPMI. This includes PDUs, Servers, Switches and Storage devices. The specifications allow a Data Center System Manager uniform remote access to the hardware in the rack. All the specifications have questions in red.
| Specification | Version | Date | Description |
|---|---|---|---|
| OCP ID | 0.04 | June 9, 2014 | Identify all OCP specifications implemented by a single IPM Controller |
| ICAP Base | 0.02 | June 9, 2014 | The basic functions of an OCP IPM Controller that a Data Center System Manager depends on |
| ICAP OPTI | 0.02 | June 9, 2014 | The identification of XFSP+/QSFP bays and protocol independent monitoring of optical links for XSFP+/QSFP modules |
| ICAP DRAM | 0.03 | June 9, 2014 | The identification and temperature monitoring of DDR3/DDR4 modules |
As Data Centers raise room temperatures, potentially on a daily basis, the ICAP OPTI and ICAP DRAM provide feedback on how close the hardware is to the manufacturer define thermal limits.
DRAFT Requirements for Firmware Update
The draft specification identifies the firmware update requirements that all OCP compliant platforms and devices must adhere to. The Firmware Update solution is for the application the user interfaces and the capabilities of the embedded firmware on the servers, switches and storage devices. The requirements are broken into four sections:
- General: general firmware update requirements
- Security: requirements related to the integrity and authenticity of the image
- Control: requirements related to how firmware updates are controlled
- Status: requirements related to how/what status is available
Firmware Update Requirements version 0.2b PDF
Cloud Server Multi Node System Specification
A description and requirements of a Cloud Server which is either a server, enclosure or rack that contains other uniquley addressable and managable devices. The document's purpose is to have a single Ethernet connection to the Cloud Server and document the possible bus architectures, managment protocols and minimum expected functionality of the Nodes contained in the Cloud Server.
The Clould Server's BMC with a Ethernet connection to the outside world is the called the Spine BMC and both IPMI RMCP or RESTfull interface are allowed. Internally the Spine BMC communicates using
The Node requirements include a subset of the IPMI 2.0 commands. The Spine BMC may have a Serial Console session to each Node.
Cloud Server Multi Node System Specification version V0.7.5, Date Augest 19,2015 PDF