Operating System Architecture for Incredibly Diverse Devices

1
Operating System Architecture for Incredibly
Diverse Devices

• David Culler
• http//www.cs.berkeley.edu/culler
• U.C. Berkeley
• Expeditions Meeting
• 8/16/1999

2
Away from the average Device
3
Convergence at the Extremes

• Powerful Services on Small Devices
• massive computing and storage in the
  infrastructure
• active adaptation of form and content along the
  way
• Extremes more alike that either is to the middle
• More specialized in function
• Communication centric design
• wide range of networking options
• Federated System of Many Many Systems
• Hands-off operation, mgmt, development
• High Reliability, Availability
• Scalability
• Power and space limited
• simplicity
• They have to work or die!

4
Body of Work in the Very Large

• Academic metacomputing, computational grid
• Glunix (UCB), Globus(ANL), Legion (UVA), IPG
  (NASA), Harness, NetSolve, Snipe (UTK), EMOP,
  Apples, Cactus, Meta-Chaos, ...
• Commercial
• LSF

5
Dominant thrust Glue over big OS

• Nodes a full OS and Institutional structure
• accounts, authentication, resources, execution,
  storage, policy
• User constructs a personal virtual machine
  spanning numerous, potentially diverse resources
• mapping from uniform metasystem mechanisms to
  site specific mechanism
• naming, authentication, transparent execution,
  storage, scheduling
• uniform, multiprotocol communication mechanism
• Many nodes in a net, not Unix on Steroids
• mgmt is fundamental problem, but still unadressed

6
And in the small...

• Academic
• microkernels, Exokernel, OSKit, ucLinux, ELKS,
• Commercial
• PalmOS, PSION, GeoWorks, WinCE, Inferno, QNX,
  VxWorks, javaos, chorusOS, BeOS,
• jini, corba, dcom, ...
• gt tracks the 80386
• when it becomes 1990 PC Unix will run on it

7
Small OS Dominant thrust

• Unix on a diet real time seasoning
• Microkernel finally works on small devices
• ability to remove components (modularity) fault
  boundaries more important than performance
• legacy applications less dominant

8
Design Issues for Small Device OS

• Managing address spaces
• Thread scheduling
• IP stack
• Windowing System
• Device drivers
• File system
• Applications Programming Interface
• Power management

9
Core Questions

• What are the principles of design for tiny
  operating systems?
• How are they different from a desktop or server?
• Where should we look for ideas and experience?
• How can operating systems be made radically
  simpler?
• How should we proceed with the investigation?

10
1st Stab at Principles for Simple OS

• Communication is fundamental
• treated as part of the hardware. No comm is like
  no power
• you dont bring up the device then configure
  comm.
• hands off a direct user interface is the
  exception not the norm
• typical device has a network on one side
  sensor/actuators on the other
• buttons and display a special case
• all deployment, development, configuration, mgmt,
  programming, is through the communication
  interface
• schedule data movement, not arbitrary threads of
  computation
• constant self-checking and telemetry
• rely on the infrastructure for hard stuff

11
other places to look for ideas

• Operating systems that are not called operating
  systems
• eg modern disk controller
• event scheduler handling stream of commands from
  network link, controlling complex array of
  sensors and actuators, performing sophisticated
  calculations to determine what and when
  (scheduling and caching) as well as transforming
  data on the fly
• automatic connection, enumeration, configuration
• but several simplifying assumptions must be
  removed

Complex array of Sensors and actuators
Network link - EIDE, SCSI - FCAL, SSA - USB,
1394 - ???
12
OS as little more than FSM

• Primary focus is scheduling discrete chunks of
  data movement
• not general thread scheduling and unlimited
  memory management
• there may be a bounded amount of work to xform or
  check data
• Commands are an event stream merged with
  sensor/actuator events
• General thread must be compiled to sequence of
  bounded atomic transactions
• spagetti part of an application is configuring
  the flows
• steady-state is straight-forward event processing
  signaling unusual events
• Simplify network-based debug and mgmt

13
Adaptation in Flows

• View data transfers as continuous flows
• plumbing as programming model
• reservoirs provide slack
• trade bandwidth for robustness
• Natural form of adaptation
• ex faster consumer gets more data
• flow equations provide goal, simple error bounds,
  and react
• performance availability

14
Example streaming output
High water mark gt stop input
low water mark gt start input

• Events associated with particular configuration
  of reservoirs and flows drive operation
  scheduling
• hard to effect through queue management and
  priorities of threads

15
Availability gt Performance availability
16
Alternative Comm Location independent Access
to shared storage

• Key Concepts from Parallel Architecture
• hierarchical composition of cache-coherence
  protocols consistency models
• natural framework for adaptation (pull what you
  touch)

17
How to proceed?

• A lot of experience to be gained from the chopped
  desktop OS efforts
• same time-to-mkt benefit research prototype
• brings many small, interesting devices on line
• bring Universal Computing Lab on-line (IRDA
  802.11)
• and from meta-OS efforts
• Build simulation environment for bottom-up
  development
• currently exploring UCLA GloMo simulator
• Develop always-on networking component for
  expedition testbed
• Deploy for unusual new devices
• smart boulders -gt pebbles -gt dust