RackHD Architecture Forum



Vision

A vision statement is meant to be forward looking, strategic and a source of inspiration and motivation for a team.

“Be the hardware M&O that runs the worlds datacenters and products.”

Mission

A mission statement is meant to concentrate on the present and define a process and an expected level of performance of a team.

“Produce and document a software architecture that provides world-class hardware M&O services that are lightweight and modular.”




Team

  • Joe Golio
  • Byron Blunk
  • Tim Larson
  • Felix Yuan
  • Andrew Hou
  • Brian Parry
  • James King
  • Jeanne Ohren
  • Michael Regert
  • Maithri Ajjampura
  • Geoff Reid
  • Torrent Glenn

Upcoming Discussion Topics - Not yet stack ranked

  • Footprint reduction
    • Expectation of maintained performance
    • Relationship to scaling - a footprint relative to a certain scale
  • Unified Data Model
    • Redfish schema alignment – RI-123
    • IPMI/microkernel based catalogs - RackHD "native"
    • WS-MAN based catalogs - Advanced PE Services (see also RI-87 and RI-88)
    • Graph-based topology - Voyager
    • Grooming and removal of stale information - Telemetry produced by pollers vs. initial discovery/catalog info
    • Cache coherency - ensuring consistency across all consumers and consumption channels (i.e. REST, AMQP, etc.) (see also RI-116 and a future RI-TBD)
  • Brownfield vs. Greenfield deployment
    • Intrusive vs. non-intrusive discovery and cataloging (see also RI-88)
    • IP scanning vs. PXE boot of microkernel (see also RI-114)
    • Use case examples
      • Advanced PE Services (see also RI-85RI-86 and RI-115)
      • Service Now (TaaS, LaaS) (see also RI-123 and RI-127)
      • Dell Server team - OS install w/o discovery
  • Scaling
    • Related to "Footprint Reduction"
    • Scaling up - No. of nodes within a single RackHD neighborhood - Single RackHD instance / L2 broadcast domain
    • Scaling out - No. of RackHD neighborhoods that can be aggregated
    • Vision - 10Ks of nodes to 100Ks of nodes
    • Service distribution 
    • How to leverage Jon Hass' TTU student resource, along with DellEMC @scale testing, to help identify bottlenecks and hotspots in the architecture
  • Workflow Engine As A Service (WFEaaS) (see also RI-82)
    • Joe Golio's made up term (smile)
    • Use cases from past engagements - CoprHD, Project-C
    • Generic - standalone WFE

Potential future governance topics:

1) Day 0/1/2 vs. Tiering ("what does it mean to "support" a hardware platform?)

2) RackHD repo organization (more hierarchical with separation of prime time code, WIP code and PoC / Art of the Possible code.



Attachments

  File Modified

Microsoft Powerpoint Presentation RackHD Architecture Guiding Principles v0.6.pptx

Feb 17, 2017 by JosephG

Contributors