Download PPT - Boston University

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Wake-on-LAN wikipedia , lookup

Piggybacking (Internet access) wikipedia , lookup

Computer network wikipedia , lookup

Distributed firewall wikipedia , lookup

Cracking of wireless networks wikipedia , lookup

Net bias wikipedia , lookup

Recursive InterNetwork Architecture (RINA) wikipedia , lookup

Remote Desktop Services wikipedia , lookup

Lag wikipedia , lookup

Network tap wikipedia , lookup

Airborne Networking wikipedia , lookup

Zero-configuration networking wikipedia , lookup

Quality of service wikipedia , lookup

Transcript
The Need for A Network Resource Status Service
IMIC Workshop 1999
Boston University
John Zinky
[email protected]
BBN Technologies
GTE
www.dist-systems.bbn.com/tech/QuO
August 30, 1999
1 8/99 IMIC Workshop 5/25/2017
New Network Services
John Zinky
QuO
BBN Technologies
Specifying QoS Parameters is a Familiar Concern
To Even the Youngest User
User-Level QoS
Size of Picture
Sample Rate of Sound
System-Level
Requirements
Mbytes of Movie File
Type of Movie Player
Access Speed of ISP
CPU/Memory
2 8/99 IMIC Workshop 5/25/2017
New Network Services
John Zinky
BBN Technologies
Distributed Objects with QoS Extensions is a Powerful
Abstraction Layer on Which to Build Applications
Collaborative Simulation
Planning
WorkFlow
CORBA + QuO
TCP/IP
QoS
ATM
New Network Services
John Zinky
Multicast
3 8/99 IMIC Workshop 5/25/2017
Distributed Objects are the
first abstraction layer that
unifies CPU, storage, and
communications
This interface needs to be
hidden from applications
• It is too complicated
• It is changing too quickly
RSVP
Ethernet
Applications for geographically
dispersed, heterogeneous
environments
BBN Technologies
Trading-Off Bandwidth, CPU and Storage is
the Key to Creating QoS Aware Applications.
• Storage is
moving into the
edge network
• The Bandwidth
bottleneck is at
the edge of the
network
• Java Applets
move CPU away
from the servers
to the client
Client
Server
ISP
ISP
Web Node
Web Farm
The backbone network is getting faster, but dumber
4 8/99 IMIC Workshop 5/25/2017
New Network Services
John Zinky
BBN Technologies
The QoS Problem:
Who is in Charge of Managing QOS
• QoS is well-controlled for some location of Client and Object
Client
Object
• QoS is unruly across the network
Client
Network
• How can we resolve this?
Object
Client
Object
Client
Object
5 8/99 IMIC Workshop 5/25/2017
New Network Services
John Zinky
BBN Technologies
Simple Client/Server-Base CORBA Implementation
Does Not Manage QoS in the Network
Client
Logical Method Call
ORB Proxy
Object
Application
Developer
ORB Proxy
Middleware
Developer
COTS ORB
COTS ORB
Network
Client
6 8/99 IMIC Workshop 5/25/2017
Network
New Network Services
Server
John Zinky
BBN Technologies
QuO Adds QoS Control and Measurement Into the
CORBA Remote Method Call
Logical Method Call
Client
SysCond
Delegate
Contract
SysCond
Object
SysCond
Contract
SysCond
SysCond
SysCond
Delegate
Qosketeer
SysCond
Mechanism/Property
Manager
ORB Proxy
Application
Developer
Specialized ORB
ORB Proxy
Mechanism
Developer
Specialized ORB
Network
Client
7 8/99 IMIC Workshop 5/25/2017
Network
New Network Services
Server
John Zinky
BBN Technologies
QuO Clients Connect to CORBA Objects using
Functional and Quality Interfaces
Functional
Client
Interface Description Language
Defines the Functional Interface
The Client sees NO change
in the functional interface,
I.e. the language bindings
are the same
IDL QDL
QuO
Delegate
Client ORB
Quality
Quality Description Languages
Define the System Interface
A QuO Connection
logically moves the object
into the client’s address
space
Network
QuO helps spread the object’s
functionality to the most
desirable location; extending
the strict Client-Server
implementation of the object
Server ORB
QuO
Delegate
The QuO Connection
integrates the individual
QoS agreements for the
Network, Client, and Server,
by using a contract
Remote
Object
8 8/99 IMIC Workshop 5/25/2017
New Network Services
John Zinky
BBN Technologies
CMU REMOS is an Example of Network
Configuration and Status Dissemination Service
9 8/99 IMIC Workshop 5/25/2017
New Network Services
John Zinky
BBN Technologies
QuO Instrumentation Collects Resource Status and
Measures Delivered QoS
Client
Measured
QoS
Correlate
Probes
Delegate
Probes
Piggybacked
Measurements
Probes
QuO
Gateway
10 8/99 IMIC Workshop 5/25/2017
Disseminate
Delegate
Infer
Integrate
Translate
Collect
Probes
Status
Resource
New Network Services
Object
Expected
QoS
Resource
John Zinky
Resource
Probes
QuO
Gateway
BBN Technologies
Applications Need to be Aware of Available Resources
• Configuration Information (Query-based)
• The expected Capacity between two IP addresses
• The expected MIP of the Client and Server computers
• Possible Caches at Client, Servers and Proxies
• Status Information (Subscription-based)
• Number Greedy Flow over bootleneck link
• Load Average of Hosts
• Available cache size
11 8/99 IMIC Workshop 5/25/2017
New Network Services
John Zinky
BBN Technologies
Conclusion: Resource Awareness is a Missing Service
that should be offered by Networks
• Distributed Objects infrastructure is the place to make trade-offs
between Bandwidth, CPU, Storage.
• A Resource Configuration and Status Service is needed to make
these trade-offs
• The Network infrastructure must be involved with creating this
service.
• Two levels of service are needed
– Query-based to get Configuration
– Subscription-based to get Status
• The service should be as Ubiquitous as DNS
– DNS is Administrative, not physical
12 8/99 IMIC Workshop 5/25/2017
New Network Services
John Zinky
BBN Technologies