330 likes | 548 Views
Policy-Based Networking. 指導教授: 鍾添曜教授 主講人: 李孝治 12/01/99. Policy Framework Architecture. Internet Draft: Waters et al., February 1999. Policy The combination of rules and services where rules define the criteria for resource access and usage. An abstraction
E N D
Policy-Based Networking 指導教授: 鍾添曜教授 主講人: 李孝治 12/01/99
Policy Framework Architecture • Internet Draft: Waters et al., February 1999. • Policy • The combination of rules and serviceswhere rules define the criteria for resource access and usage. • An abstraction • of general configuration and operational characteristics of resources in a policy domain(Autonomous System/Administrative Domain) & • of service-level objects (SLO) on these resources(derived from service-level Agreements, SLA) • Components: Condition/Action pairs • Policy Management Tool - define and update policy rules • Policy Repository - retain and retrieve rules • Policy Decision Point (PDP) - access the conditional criteria of a rule • Policy Enforcement Point (PEP)- execute the actions of a rule when the conditions evaluate to “true”
Policy prescriptions | V +--------------------+ | Policy | Functions: Policy Editor | Management Tool | Rule Translation +---------+----------+ Rule Validation | | <--- Repository Access Protocol | (e.g., LDAP) +---------+----------+ | Policy Repository | | (Directory Server, | Functions: Storage and Retrieval | Database, etc. ) | of Rules +---------+----------+ | | <--- Repository Access Protocol | (e.g., LDAP) +---------+----------+ | Policy Decision | Functions: Policy Trigger and | Point (PDP) | rule locator. +---------+----------+ Device Adapter | State and Resource Validation | | <--- Policy Protocol for policy mechanisms | +---------+----------+ | Policy Enforcement | Function: Execution of specified Actions | Point (PEP) | Device State and Resource +--------------------+ ValidationFigure 2. Policy Architecture Functional Definition
A Framework for Policy-Based Admission Control • Internet Draft: Yavatkar et al., April 1999. • IP’s best effort service QoS service • Integrated Service (int-serv) • Certain data flows receive preferential treatment over other flows • Requires Explicit Signaling (Example: RFC 2205 - RSVP) • Differentiated Service (diff-serv) • Current Admission Control Schemes do not include “policy” criteria such as identity of user and application, ingress point, traffic/bandwidth requirements, security, time of day/week, etc.
________________ Policy server | | ______ | Network Node | | |-------------> | _____ | | | May use LDAP,SNMP,.. for accessing | | | | | | policy database, authentication,etc. | | PEP |<-----|------->| PDP |-------------> | |_____| | |_____| | | |________________| Figure 1: A simple configuration with the primary policy control architecture components. PDP may use additional mechanisms and protocols for the purpose of accounting, authentication, policy storage, etc.
________________ ____________________ | | | | | Network Node | Policy Server | Network Node | | _____ | _____ | _____ _____ | | | | | | | | | | | | | | | PEP |<-----|---->| PDP | | | PEP |<-->| PDP | | | |_____| | |_____| | |_____| |_____| | | ^ | | | | | _____ | |____________________| | \-->| | | | | LPDP| | | |_____| | | | |________________|Figure 2: Two other possible configurations of policy control architecture components. The configuration on left shows a local decision point at a network node and the configuration on the right shows PEP and PDP co-located at the same node. Note: LPDP = Local Policy Decision Point
______________________________ | | | RSVP Router | | ________ _____ | _____ | | | | | | | | | | RSVP |<------->| PEP |<--|---------->| PDP | | |________| |_____| | |_____| | ^ | | | Traffic control | | | _____________ | | \---->| _________ | | | | |capacity | | | | | | ADM CTL | | | | | |_________| | | --|----------->| ____ ____ | | | Data | | PC | PS | | | | | |____|____| | | | |_____________| | | | |______________________________|Figure 3: Relationship between PEP and other int-serv components within an RSVP router. Note: PC = Packet Classifier, PS = Packet Scheduler
AD-1 AD-2 AD-3 ________________/\_______________ __/\___ __/\___ { } { } { } A B C D E +-------+ +-----+ +-------+ +-------+ +-------+ | RSVP | | RSVP| | RSVP | | RSVP | | RSVP | +----+ |-------| |-----| |-------| |-------| |-------| | S1 |--| P | L |---| |----| P | L |----| P | P |----| P | +----+ +----+ | E | D | +-----+ | E | D | | E | D | | E |----| R1 | | P | P | | P | P | | P | P | | P | +----+ +-------+ +-------+ +------+ +-------+ ^ ^ ^ | | | | | | | | +-------+ | | | PDP | | +------+ | |-------| +-------->| PDP |<-------+ | | |------| +-------+ | | PS-2 +------+ PS-1Figure 4: Placement of Policy Elements in an internet
RSVP Extensions for Policy Control • Internet Draft: Herzog, April 1999. • These extensions include the standard format of POLICY_DATA objects, and a description of RSVP's handling of policy events. • POLICY_DATA objects are carried by RSVP messages and contain policy information. • All policy-capable nodes can generate, modify, or remove policy objects. • Such policies can be successfully deployed across multiple administrative domains when border nodesmanipulate and translate POLICY_DATA objects according to established sets of bilateral agreements.
PDP1 PDP2 | | | | +---+ +---+ +---+ | A +---------+ B +---------+ C | +---+ +---+ +---+ PEP2 PIN PEP2Figure 1: Autonomous Domain scenario Note: PIN = Policy Ignorant Node
IANA Considerations:RSVP Policy Elements (P-Types) • Numbers 0-4915: • standard policy elements by IETF Consensus action • Numbers 49152-53247: • vendor specific (one per vendor) by First Come First Serve basis • Numbers 53248-65535: • reserved for private use and are not assigned by IANA
A Policy Based Networking Architecture for Enterprise Networks • Nomura et al. – ICC’99 • Policy-based Systems Management (PSM) • To reduce the total cost of ownership (TCO) while providing highly functional enterprise systems:conventional data, voice data, and video, etc. • Tendencies of enterprise networks: • Scalability • QoS guarantee for multimedia data: QoS/CoS control • Reliability over IP networks • Problems of conventional enterprise systems • Increasing cost of system operation and management • Complication of network device control • Not sharing management information
★ • PSM Componets: • System • Application • Network ★
Policy-Based Networking Architecture for QoS Interworking in IP Management • Blight & Hamada – IFIP/IEEE INM’99 • Scalable architecture for large-scale enterprise public interoperation • Two approaches: • Policy abstraction • Hierarchy organization with precedence rules
Three approaches to meet QoS requirements in IP networks: • Integrated Services (IntServ) • Differential Services (DiffServ) • Traffic Shaping and Policing • To differentiate traffic: • In the traditional IP management – • IP Address, Protocol, Port, TOS(IPv4) and Priority(IPv6) • In the new enhanced IP management – • Users, Service, User Priority
Netscape LDAP servers,Novell NDS,MS Active Directories LDAP COPS,DIAMETER
★ ★ ★Edge switches are primarily responsible for supporting end-to-end QoS: ◆ Management view translation ◆ Policy interoperation ◆ SLA management
Scalability • Physical Scalability • A primary concern • X.500 series • Distributed Directory System Agents (DSA) • Control Path Scalability • Not a critical factor (∵control traffic is low in volume) • Management Scalability • The main theme of this paper
Two mechanisms that interact between the directory server and network elements: • Active Networks: Network elements maypull policies from the directory server • Passive Networks: The directory server must push policies to network elements • Two mechanisms may coexist in a domain. • The difference between the two mechanisms does not affect physical scalability, but the balance of load and intelligence.
To manage complexity, two orthogonal manners to organize policies: • (1) Policy Abstraction • Specific policies for QoS routing • Identity of traffic streams • Minimum and maximum QoS • Priority • Identity of policy source • Enabling conditions
Corporate Policy (A) Tokyo D < AA < C B < A D San Jose Department Policy (B) London Department Policy (C) • (2) Aggregation Hierarchy with Precedence Rules • Policy Divergence: • It becomes increasingly difficult to detect and predict policy conflicts within large policy sets. • A solution to override policy conflicts locally:Precedence Rules
The COPS Protocol(Common Open Policy Service) • Internet Draft: Boyle, August 1999. • A client/server model to exchange policy information between a PDP and its clients (PEP). • Assume that at least one policy server exists in each controlled administrative domain.
+----------------+ | | | Network Node | Policy Server | | | +-----+ | COPS +-----+ | | PEP |<-----|-------------->| PDP | | +-----+ | +-----+ | ^ | | | | | \-->+-----+ | | | LDP | | | +-----+ | | | +----------------+ Figure 1: A COPS illustration.
The PEP is responsible for initiating a persistent TCP connection to a PDP. • When the PEP sends a configuration request, it expects the PDP to continuously send named units of configuration data to the PEP via decision messages. • When a unit of named configuration data is successfully installed on the PEP, the PEP should send a report message to the PDP confirming the installation. • The PDP may then update or remove the named configuration information via a new decision message. • The PEP will delete the specified configuration and send a report message to the PDP as confirmation. • The PEP is responsible for notifyingthe PDP when a request state has changed on the PEP. • The PEP may make a local policy decision via its LDP. But the PEP must relay the local decision to the PDP.
Signaled Preemption Priority Policy • Internet Draft: Herzog, February 1999. • Traditional Capacity-based Admission Control (CAC) vs.Policy-based Admission Control (PAC) • Preemptive Priority • To preempt some previous admitted low-priority flows in order to make room for a newer, higher-priority flow • Assumptions: • They are created by PDPs • They can be processed by LDPs • The are enforced by PEPs • Stateless Policy
Policy Element Format: +-------------+-------------+-------------+-------------+ | Length (12) | P-Type = PREEMPTION_PRI | +------+------+-------------+-------------+-------------+ | Flags (0) | M. Strategy | Error Code | Reserved(0) | +------+------+-------------+-------------+-------------+ | Preemption Priority | Defending Priority | +------+------+-------------+-------------+-------------+ • P-type = PREEMPTIN_PRI • Preemption Priority: • Used to compete when admitting • Defending Priority: • Used to defense once admitted • PP ≦ DP • If DP – PP is large → more stable
Priority Merging • Problem: • F1: QoS = High, Priority = LowF2: QoS = Low, Priority = HighF1 + F2 = F3: QoS = High, Priority = ? • If Priority of F3 = High, F1 becomes a Free-RiderIf Priority of F3 = Low, F2 may gets a Denial of Service • Different Merging Strategies(specified in the policy element: “M. Strategy”) • Take priority of highest QoS → Low • Take highest priority → High • Force error on heterogeneous merge (different QoS) → Error
Issues on Policy Based Networking/Management • Policy Representation • Policy Conflicts • Inter-operability • Inter-Network and Inter-Domain Communication • Scalability • Multicasting • AAA (Authorization, Authentication, and Accounting) • Reliability • Security