SAHARA Reference Architecture and Summer Plans 14 May 2001

1
SAHARA Reference Architecture and Summer Plans14
May 2001

• Anthony Joseph, Randy H. Katz, Ion Stoica
• Computer Science Division
• Electrical Engineering and Computer Science
  Department
• University of California, Berkeley
• Berkeley, CA 94720-1776

2
Project Goals

• Delivery of end-to-end services with desirable
  properties (e.g., performance, reliability,
  qualities), provided by multiple potentially
  distrusting service providers
• Architectural framework for
• Economics-based resource allocation
• Third-party mediators, such as Clearinghouses
• Dynamic formation of service confederations
• Support for diverse business models

3
Overlay Services Networks

• E.g., solve the multicast management and peering
  problems by moving up the protocol stack

Steve McCanne
4
McCannes Service Stack
Applications
End Host
End host Services
TCP service
DNS stub
redirection
Infrastructure Services
Overlay
Cache Services
Proxy Services
DNS
Network Services
IP service
Router
Steve McCanne
5
Randys Service Architecture
Prototype Applications Universal In-Box,
Context-Aware UI, Group Collaboration
Context-Awareness Services Activity
Tracking/Coordination,Preferences
Specification/Interpretation
Adaptation Services Introspection, Tacit
Information Extraction/Organization
Wide-Area Services Discovery, Mobility, Trust,
Availability
Performance Measurement and Monitoring
Core Wide-Area Network
6
Bhaskars Reference Model
Composed services
Application plane
Service location
Service-level path creation
Peering relations, Overlay network
Network performance
Logical platform
Detection
Handling failures
Service clusters
Recovery
Hardware platform
7
Ideas for Driving Applications

• Location-aware/context-aware information delivery
  and presentation
• Extends UniIn-Box loc-based, exploits calendar
  info
• Mediation to translate formats
• IP Telephony, Packet VoD, Teleconferencing
• Streaming media, multicast-based
• Bandwidth, latency, jitter, lose rate constraints
• Clearinghouse provisioning
• Event Delivery for Distributed Applications
• Performance/reliability constrained messaging
• Interactive Games? Storage/OceanStore? Telemetry?

8
Ideas for Operational Environment

• Virtual Operators/Service Provider (VOSP)
• Provide service to end users with no
  server/network infrastructure of own
• Independent Path providers (e.g., ISPs) and
  Server providers (e.g., Internet Data Centers)
• Many-to-many relationship between VOSP and
  Path/Server Providers
• Confederated Service Provider
• Service-level peering sharing of paths and
  servers to deploy end-to-end service with
  performance and reliability constraints

9
Alternative Operational Environments

• Confederation Model
• Providers share (limited) information about
  topology, server location, path performance
• Cooperatively collect internal information and
  share
• Overlay Model
• Reverse-engineer topology and intra-cloud
  performance
• Collection done by brokers outside of the cloud
• SLAs, Verification, Maintenance of Trust
  Relationships different in the two models

10
Open Issues/Questions

• Traditional Overlay Networks
• Server (Application Level Router) Placement
• For scaling, reliability, load balancing, latency
• Where? Network topology discovery WAN Core,
  Metro/Regional, Access Networks
• Choice of Inter-Server Paths
• For server-to-server latency/bandwidth/loss rate
• Predictable/verifiable network performance
  (intra-ISP SLA)
• Redirection Mechanisms
• Random, round-robin, load-informed redirection
• Net vs. server as bottleneck

11
Open Issues/Questions

• Performance-constrained Service Placement
• Separation of Service, Server, Service Path
• Assume Server Centers known, can be
  discovered (how does OceanStore deal with
  this?), or register with a Service Placement
  Service (SPS)
• How is Service named, described, performance
  constraints expressed, and registered?
• How is app/service-specific performance measured
  and made known to Service Placement Service?
• Brokering between Server Centers and Service
  Creator, Path Provider and Service Creator

12
Reference Architecture
Distributed Application
Constraint Specification
Marshal Resources Based on Economic Constraints
Adapt
Service Redirection
Service Registration
Service Placement Service
Pricing Service
Path Broker
Server Broker
Perf Measurement Service
Server Registration
Advertisement Registration
SLAs
Verify
Path Provider (ISP Cloud)
Path Provider (ISP Cloud)
Path Provider (ISP Cloud)
Path Provider (ISP Cloud)
Path Provider (ISP Cloud)
Server Center Provider
13
Methodological Framework

• Wide-area Network Emulation
• How is OceanStore tackling this problem?
• How is Walrand group tackling this problem?
• Virtual WAN on Millennium Testbed
• Build operational model on top of VWAN
• Traffic generation and measurement infrastructure
• Build Confederation and Overlay operational
  models
• What part of mechanisms for measurement,
  negotiation, registration, redirection, etc. the
  same and which are different?