(Mostly) unedited notes from Stephen Foskett’s Interop session. And yes, FCoTR was mentioned!

• Converging on convergence
  • Data centers rely more on standard ingredients
  • What will connect the systems together
• Drivers of convergence
  • Virtualization
    • Demanding greater nw & storage IO
    • I/O blender
    • Mobility & abstraction
  • Consolidation
    • reduce port count, combine LAN + SAN
    • NW abstraction
• Performance
  • Data-driven applications need massive I/O
  • Virtualization, VDI
• Whats in it for you?
  • Server managers
    • Flexibility/mobility
    • Better support for virt servers & blades
    • Increased overall performance
  • Network managers
    • According to Foskett, network managers get the worst of end of the changes in the data center (he put it a different way, but this is what he meant).
    • Wider sphere of influence – ethernet everyplace
    • More tools to control traffic
    • More to learn
    • New headaches frm storage protocols
  • Storage managers
    • Fewer esoteric storage protocols
    • New network protocols
    • Less responsibility for I/o
• Flexibility
  • No more cabling issues
  • Servers become interchangeable units
  • Less concern abt i/o slots, cards & ports
• Changing data center
  • Placements & cabling of SAN switches & adapters dictates where to install servers
  • No more north south conversations (traditional client server applications), more east west.
• Why choose a protocol?
  • It depends (typical consultant answer 😉 )
  • Strategy
  • Performance
  • Compatibility (lots of FC arrays, FCoE helps fan the arrays inexpensively by bridging old FC infrastructure)
    • If you have no FC, why go to FCoE?
  • Cost

We are looking at next evolution of scsi, this is not something really radical

• Why go iSCSI?
  • Targets are robust & mature. Almost storage vendor offers iSCSI arrays
  • Lots of software targets
    • Lots of client-side adabters/HBAs and adapters
    • Smooth transition (plug&play ) into 10G
  • Stephen also gave a nice shout out to a white paper by Dell on iSCSI performance
    
• Why goFCoE?
  • Lots of arrays, switches, tools, skillsets to support FC, easily transferable
  • Provides a way for incremental adoption
• Stephen also predicted mass adoption of FCoE by 2014. You heard it here first. 🙂