sun n-tire

Upload: dj-jam

Post on 01-Jun-2018

215 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/9/2019 Sun N-tire

    1/56

    Sun Microsystems, Inc.4150 Network Circle

    Santa Clara, CA 95045 U.S.A.650 960-1300

    http://www.sun.com/blueprints

    Network Design Patterns:N-Tier Data Centers

    Deepak Kakadia, Staff Engineer, Network Architect,Sun Microsystems

    Richard Croucher, Chief Architect, Professional

    Services, Sun MicrosystemsSun BluePrints™ OnLine—October 2003

    Part No. 817-3997-10

    Revision 1.0, 10/9/03Edition: October 2003

  • 8/9/2019 Sun N-tire

    2/56

    Please

    Recycle

    Copyright2003 SunMicrosystems, Inc. 4150Network Circle, Santa Clara, California 95045 U.S.A.All rightsreserved.

    SunMicrosystems, Inc. hasintellectual propertyrights relatingto technology embodied in theproduct that is describedin this document. Inparticular, andwithout limitation, these intellectual propertyrights mayinclude oneor more of theU.S. patents listedat http://www.sun.com/patents andone or more additionalpatents or pending patentapplicationsin theU.S. andin other countries.

    This product or document is protectedby copyrightand distributed under licenses restricting itsuse, copying, distribution, anddecompilation.No part of this product or document maybe reproduced in anyform by anymeans without prior written authorization of Sunand itslicensors,if any. Third-party software, including font technology, is copyrighted and licensed fromSun suppliers.

    Parts of theproduct maybe derived from Berkeley BSDsystems,licensedfrom theUniversity of California. UNIX is a registered trademark inthe United States and othercountries,exclusivelylicensed through X/OpenCompany, Ltd.

    Sun, SunMicrosystems, theSun logo, SunBluePrints, SunONE, SunStorEdge,Sun Fire, SunDocs, Jump Start,iPlanet,Enterprise Java Beans, Java Database Connectivity, and Solaris are trademarks or registered trademarks of Sun Microsystems,Inc. in the United States and othercountries. AllSPARCtrademarks areused under license andaretrademarks or registered trademarks of SPARCInternational, Inc. in theUSandother countries. Products bearing SPARCtrademarks arebased upon an architecture developedby SunMicrosystems, Inc.

    TheOPEN LOOKand Sun™ GraphicalUser Interface wasdeveloped by SunMicrosystems, Inc. forits users andlicensees. Sunacknowledgesthepioneering efforts of Xerox in researchingand developing theconcept of visualor graphicaluser interfaces forthe computer industry. Sunholds a non-exclusive license from Xerox to theXeroxGraphical User Interface, which license also coversSun’slicensees whoimplement OPENLOOKGUIs andotherwise complywith Sun’s written license agreements.

    U.S. Government Rights—Commercial use.Governmentusersare subject to the Sun Microsystems, Inc. standard license agreementandapplicable provisions of theFar anditssupplements.

    DOCUMENTATION IS PROVIDED "AS IS" AND ALL EXPRESS OR IMPLIED CONDITIONS, REPRESENTATIONS AND WARRANTIES,INCLUDING ANY IMPLIED WARRANTY OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE OR NON-INFRINGEMENT,ARE DISCLAIMED, EXCEPT TO THE EXTENT THAT SUCH DISCLAIMERS ARE HELD TO BE LEGALLY INVALID.

    Copyright 2003 Sun Microsystems, Inc.,4150 Network Circle, SantaClara,Californie 95045 Etats-Unis. Tous droits réservés.

    SunMicrosystems, Inc. a lesdroits de propriété intellectuels relatants à la technologie incorporée dans le produit quiestdécrit dans cedocument. En particulier, et sans la limitation, cesdroits de propriété intellectuelspeuvent inclure un ou plus des brevets américains énumérésà http://www.sun.com/patents et un ou lesbrevets plus supplémentairesou lesapplicationsde brevet en attente dans les Etats-Unis et dansles autrespays.

    Ce produit ou document estprotégé par un copyrightet distribuéavec deslicences quien restreignent l’utilisation, la copie, la distribution, et ladécompilation. Aucunepartiede ce produit ou document ne peut être reproduite sous aucuneforme, parquelque moyen quece soit, sans

    l’autorisation préalable et écrite de Sunet de sesbailleursde licence, s’il y en a. Le logiciel détenupar des tiers,et quicomprend la technologierelative aux polices de caractères, estprotégé par un copyrightet licencié par desfournisseurs de Sun.

    Desparties de ce produit pourrontêtre dérivées dessystèmesBerkeley BSDlicenciés parl’Université de Californie. UNIX estune marqueenregistree auxEtats-Unis et dans d’autres pays et licenciée exclusivement par X/OpenCompany Ltd.

    Sun, SunMicrosystems, le logoSun, SunBluePrints, SunONE, SunStorEdge, SunFire,SunDocs,Jump Start,iPlanet, Enterprise Java Beans, Java Database Connectivity, et Solaris sont des marques de fabrique ou des marques déposées, ou marques de service, de Sun Microsystems,Inc. auxEtats-Unis et dans d’autres pays. Toutes lesmarques SPARCsont utiliséessous licence et sont des marques de fabrique ou des marquesdéposées de SPARCInternational, Inc. auxEtats-Unis et dans d’autres pays. Lesproduitsportant lesmarques SPARCsont basés surunearchitecture développée par Sun Microsystems, Inc.

    L’interface d’utilisation graphiqueOPEN LOOK et Sun™ a étédéveloppée par SunMicrosystems,Inc. pour sesutilisateurset licenciés. Sunreconnaîtles efforts de pionniersde Xerox pour la recherche et le développement du concept des interfaces d’utilisation visuelle ou graphique

    pour l’industrie de l’informatique.Sun détient unelicence nonexclusive de Xerox surl’interface d’utilisation graphique Xerox, cette licencecouvrant égalementles licenciésde Sunqui mettent en place l’interface d’utilisation graphique OPEN LOOKet quien outre se conforment auxlicences écrites de Sun.

    LA DOCUMENTATION EST FOURNIE "EN L’ÉTAT" ET TOUTES AUTRES CONDITIONS, DECLARATIONS ET GARANTIES EXPRESSESOU TACITESSONT FORMELLEMENTEXCLUES, DANS LA MESUREAUTORISEE PAR LA LOIAPPLICABLE, Y COMPRIS NOTAMMENTTOUTE GARANTIE IMPLICITE RELATIVE A LA QUALITE MARCHANDE, A L’APTITUDE A UNE UTILISATION PARTICULIERE OU AL’ABSENCE DE CONTREFAÇON.

  • 8/9/2019 Sun N-tire

    3/56

    1

    Network Design Patterns: N-Tier

    Data Centers

    This article describes recommendations and best practices for designing andimplementing N-Tier architectures for web-based service delivery infrastructures. Itdocuments a set of fully tested configurations, comparing solutions using twodifferent network equipment providers.

    The design and implementation of optimal N-Tier architectures requires a thoroughanalysis of functional and non-functional requirements, including performance,availability, security, scalability and flexibility. The degree to which non-functionalrequirements are met and exceeded often contributes to the quality factor of anarchitecture. In this paper, we present some proven design principles and concepts,which we hope are of value towards the successful delivery of N-Tier architecturesolutions.

    For the purposes of this paper, we distinguish network “architecture” from “design”as follows:

    Architecture  is a  high-level description of how the major components of the systeminterconnect with each other from a logical and physical perspective.

    Design  is more of a  process that specifies , in sufficient detail for implementation,how to construct a network of interconnected nodes that meets or exceeds functionaland non-functional requirements (performance, availability, scalability, and so on).

    There currently exists a sufficient level of available public information describing theconstruction of N-Tier data center architectures from a software standpoint.However, from a systems and network architecture perspective, the best way to goabout constructing these architectures might not be as clear. This article attempts tofill that void, by describing:

      The recommended network design patterns and reasons behind these designs

      Actual implementations that have been tested, tuned and proven in actualdeployments

  • 8/9/2019 Sun N-tire

    4/56

    2 Network Design Patterns: N-Tier Data Centers • October 2003

    This article also describes two popular approaches to network architectures, eachwith its own merits and limitations.

    This article addresses the following topics:

      “Defining Architectures” on page 2   “Implementation Overview” on page 10   “Designing the Network Configuration” on page 11   “Configuring the Physical Network” on page 35   “Network Security” on page 48

    The focus of this article is limited to data flow. Systems and network management isa topic in itself and is not discussed in this paper. Readers should have sufficiently

    detailed information to extend and modify this discussion to their particular needs.

    Defining Architectures

    This section breaks architecture considerations into two sections:   “Logical Architectures” on page 2   “Physical Architectures” on page 6

    Logical Architectures

    A typical high-level functional overview of an N-Tier data center architecture isshown in FIGURE 1. The tiers are logically segregated by functionality as follows:

    1. Web service tier—The set of load-balanced, front-end web servers that handles theinitial ingress client HTTP web requests. Each web server typically includes aservelet engine or JSP engine, and the capability to execute locally hosted cgi/binscripts which execute native code.

    2. Application service tier—The set of application server instances that might span

    one large multi-CPU server or several smaller multi-CPU servers. This tier providesgreater functionality and sophicated processing capabilities and services, such as EJBcontainers, an ORB, or connector modules that integrate with legacy servers. Theapplication server usually interfaces with the web server. In recent years, the webserver has been included in the application server text image for increasedperformance.

    3. Naming service tier—The set of servers, possibly load balanced, that providessecondary infrastructure support services for managing and maintaining naming

    information for hosts addressing data, configuration data, and so on. This caninclude services such as DNS and LDAP.

  • 8/9/2019 Sun N-tire

    5/56

    Defining Architectures   3

    4. Data service tier—The set of highly available servers that provides persistent datastorage services for critical corporate data. Due to the the sensitive nature of thedata, this tier is usually highly protected in terms of security and availability. We

     briefly describe new virtual tiered firewalling techniques that provide increasedsecurity without sacrificing performance, using hardware-based firewalls.

    FIGURE 1   Logical High-Level Overview of N-Tier Architecture

    Note – Client access represents all the client systems that can access the referencearchitecture (client systems, web browsers, and so forth). This access is not

    considered part of the N-Tier architecture

    The actual reference implementation described leveraged Sun™ ONE technology.The Sun ONE software stack provides the software needed to deploy applicationsusing this reference implementation. The tiers of the N-Tier data center architecture,shown in FIGURE 1, are mapped to the Sun ONE software as follows:

      Web service tier—Supported by Sun ONE Web Server, Enterprise Edition software

      Naming service tier—Supported by Sun ONE Directory Server software   Application service tier—Supported by Sun ONE Application Server software

    Applica

    tion ser

    vice tier

    Naming ser

    vice tier

    Web se

    rvice tie

    r

    Client acc

    ess

    Refer

    ence 

    archit

    ecture

     servi

    ce tiers

    LDAP

    DNS

    Data se

    rvice tie

    r

    SAP

    Oracle

  • 8/9/2019 Sun N-tire

    6/56

    4 Network Design Patterns: N-Tier Data Centers • October 2003

      Data service tier—Provided by Oracle 9i database software, which is not part of the Sun ONE software stack

    FIGURE 2 illustrates in greater detail how the logical N-Tier architecture is mapped tothe software systems architecture, and how Sun ONE and other softwarecomponents fit to create a complete system.

  • 8/9/2019 Sun N-tire

    7/56

    Defining Architectures   5

    FIGURE 2   Logical N-Tier Software Systems Architecture

    F i   r  ew al   l   

    F i   r  ew al   l   

    F i   r  ew al   l   

    JDBC

    Connec

    tions

    Naming

    Messag

    ing

    Mail

    Security

    JPDA

    Administratio

    n

    and mo

    nitoring 

    Servlet

    contain

    er

    EJB

    contain

    er

    HTTP

    server

    ORB

    JDBC

    Connec

    tions

    Naming

    Messag

    ing

    Mail

    Security

    JPDA

    Adminis

    tration

    and mo

    nitoring

     

    Servl

    et

    container

    EJB

    contain

    er

    HTTP

    server

    ORB

    Web

    server

    Router

    Internetor

    Intranet

    Oracle

    Legacy

    Sun ONE app

    lication

     server 

    instanc

    es

    Data

    service

    tier

    Applica

    tion

    servi

    ce

    tier

    Web

    service

    tier

    LDAP

    LDAP

    DNS

    DNS

    Naming

    service

    tier

    Identity 

    and policy

    Platform

    Service

     contain

    er

    Web servi

    cesApp

    lication

    s core

    web ser

    vices

    Servic

    e

    delivery

    Service

    integrat

    ionWe

    b

    server

    Router Ser

    vice cre

    ation, a

    ssembl

    y and de

    ployment 

  • 8/9/2019 Sun N-tire

    8/56

    6 Network Design Patterns: N-Tier Data Centers • October 2003

    Physical ArchitecturesThis logical architecture is then mapped to a physical realization as shown in

    FIGURE 3, which describes the physical architecture. This mapping is influencedheavily at this point by non-functional requirements, such as performance andavailability. For example, redundant servers and redundant network connectivity area result of availability requirements and possibly performance requirements.Wirespeed multilayer switches might satisfy performance requirements, providedthat the switch performs packet processing within timing constraints. Most Layer 2and Layer 3 switches are implemented using ASICs, which are commonly available.Higher layer packet service implementations differentiate vendors from each other.

    Some vendors, such as Foundry in the ServerIron product line, implement Layer 7using general purpose CPUs. This is much slower that an ASIC or an FPGAimplementation.

  • 8/9/2019 Sun N-tire

    9/56

    Defining Architectures   7

    FIGURE 3   Physical N-Tier Architecture Showing Hardware Components

    Client 1 Client 2

    L2-L3

    Edge switch

    Web

    service

    tier

    Edge

    network

    Sun Fire

    6800

    Sun Fire

    6800

    Sun Fire

    6800

    Sun Fire

    6800

    Clientaccess

    Applicationservice

    tier

    Data

    service

    tier

    Naming

    servicetier

    Core switches

    Sun Fire 280R Sun Fire 280R Sun Fire 280R Sun Fire 280R

    Sun Fire 280R Sun Fire 280R Sun Fire 280R Sun Fire 280R

    T3

    T3

  • 8/9/2019 Sun N-tire

    10/56

    8 Network Design Patterns: N-Tier Data Centers • October 2003

    TABLE 1 through TABLE 3 describe the actual tested hardware components andconfigurations. This architecture is tested and proven. However, it is easilyextensible for adapting to specific customer requirements while keeping the

    fundamental pattern intact. For example, increasing or decreasing the number of redundant servers of a particular tier directly impacts the degree of availability andperformance (if configured as Active Active), but does not violate the fundamentalarchitecture.

    TABLE 1   Tested Systems Configuration: Sun Servers and Storage

    Service Tier Function Equipment Features

    Memory

    (RAM)

    Storage

    Capacity

    Operating

    Enviro nmen t Software

    Web   Web servers (4) Sun Fire™ 280Rservers

    Two 900-MHz CPUs

    2 GB 36 GB Solaris 9 10/02with patches

    Sun ONEWeb Server6.0, ServicePack 5,EnterpriseEdition

    Naming   DNS servers (2) Sun Fire 280R

    servers

    Two 900-

    MHz CPUs

    2 GB 36 GB Solaris 9 10/02

    with patches

    n/a

    Directoryserver

    (2) Sun Fire 280Rservers

    Two 900-MHz CPUs

    4 GB 36 GB Solaris 9 10/02with patches

    Sun ONEDirectoryServer 5.1

    Application Applicationservers

    (2) Sun Fire 6800servers

    24 1-GHzCPUs

    64 GB 200 GB Solaris 9 10/02with patches

    Sun ONEApplicationServer 7.0

    Data   Data servers (2) Sun Fire 6800servers

    24 1-GHzCPUs

    32 GB 36 GB Solaris 9 10/02with patches

    VERITASVolumeManager 3.5software,Oracle 9iRelease 2software

    Databasestorage

    (2) Sun StorEdge™T3 Arrays

    n/a n/a 200 GB n/a n/a

  • 8/9/2019 Sun N-tire

    11/56

    Defining Architectures   9

    TABLE 2   Tested Network Configuration 1 - Extreme Networks Equipment

    Function Equipment Features Operating Environment

    Edge switch (1) Extreme NetworksSummit7i switch

    24 1-Gb ports Extreme OS 6.2.2 build 18

    Core switch (2) Extreme NetworksBlackDiamond 6800switches

    Chassis switch with 7 blades. Each blade has 81-Gb ports (56 portstotal)

    Extreme OS 6.2.2 build 18

    TABLE 3   Tested Network Configuration 2 - Foundry Networks Equipment

    Function Equipment Features Operating Environment

    Edge switch (1) Extreme NetworksSummit7i switch

    24 1-Gb ports Extreme OS 6.2.2 build 18

    Core switch (2) Foundry BigIron4000 Layer 2/3

    Server load- balancingswitch

    (2) FoundryServerIron XL load-

     balancing switches

  • 8/9/2019 Sun N-tire

    12/56

    10 Network Design Patterns: N-Tier Data Centers • October 2003

    Implementation OverviewOnce the design and component selection is complete, follow these implementationsteps to set up your service tiers and their software:

     To Implement Systems as Service Tiers

    1. Install computer systems and storage hardware.

    2. Establish network connectivity.

    3. Configure partitions and domains on the Sun Fire 6800 servers in the applicationand data service tiers.

    4. Install and configure the Solaris™ Operating System on all systems.

    5. Configure the Sun StorEdge T3 arrays on the Sun Fire 6800 servers in the dataservice tier.

    6. Install Oracle 9i  Release 2 database software on the Sun Fire 6800 servers in thedata service tier.

    7. Install the Sun ONE software on the following systems:

    a. Web service tier—Install the Sun ONE Web Server software on the four SunFire 280R servers in this service tier.

    b. Naming service tier—Configure DNS on two of the Sun Fire 280R servers.

    c. Naming service tier—Install and configure the Sun ONE Directory Server 5.1software on the other two Sun Fire 280R servers.

    Refer to the Sun ONE Directory Server 5.1 documentation collection (see “RelatedResources” on page 54 for more information).

    d. Application service tier—Install and configure the Sun ONE Application Serversoftware on two Sun Fire 6800 servers.

    Note – See http://docs.sun.com for installation and hardware guides.

  • 8/9/2019 Sun N-tire

    13/56

    Designing the Network Configuration   11

    Designing the Network ConfigurationOne can argue that the heart of any architecture composed of multiple nodes is thenetwork architecture. The network architecture is vital to assembling components or

     building blocks in such a way that the system delivery is high performance, flexible,highly available, scalable, and secure. We describe a recommended methodology forcreating these architectures. We also describe actual implementations which showthe contrasts between two different network vendors, and how to optimally leverageeach of them. This section discusses the following topics:

      “Logical Network Architecture” on page 11   “IP Services” on page 11   “Designing Networks for Availability” on page 19   “Networks and VLANs” on page 23   “Inter-Tier Traffic Patterns” on page 32

    Logical Network ArchitectureThe logical network design is composed of segregated networks, implementedphysically using virtual local area networks (VLANs) defined by the networkswitches. The internal network uses private IP address spaces (for example,10.0.0.0) as specified in RFC 1918 for security and portability advantages. Lookahead to FIGURE 10 for a high-level overview of the logical network architecture.

    The management network provides centralized data collection and management of all devices. Each device has a separate interface to the management network toavoid contaminating the production network in terms of security and performance.The management network is also used for automating the installation of thesoftware using Solaris JumpStart™ technology.

    Although several networks physically reside on a single active core switch, networktraffic is segregated and secured using static routes, access control lists, and VLANs.From a practical perspective, this is as secure as separate individual switches.

    IP ServicesThe following subsections provide a description of some of the more importantemerging IP services that are often important components of a complete networkdesign for a Sun ONE deployment. These IP services are divided into two categories:

  • 8/9/2019 Sun N-tire

    14/56

  • 8/9/2019 Sun N-tire

    15/56

    Designing the Network Configuration   13

    FIGURE 4   IP Services Switch Functions on Incoming Packets

    FIGURE 4 shows that a typical client request is destined for an external VIP, with IPaddress a.b.c.d and port 123. The SLB algorithm eventually intercepts the packet,and rewrites the destination IP address corresponding to the real server, which waschosen by a particular algorithm. The packet is then returned as indicated by thesource IP address.

    Application Redirection - Class 2 Stateless

    The Application Redirection function intercepts a client’s HTTP request andredirects the request to another destination, which is usually a group of cacheservers. Application Redirection rewrites the IP destination field. This is differentfrom proxy switching, where the socket connection is terminated and a newconnection to the server is created to fetch the requested web page.

    Application Redirection serves the following purposes:

      Reduces the load on one set of web servers and redirects it to another set, usuallycache servers for specific content

      Intercepts client requests and redirects them to another destination for control of certain types of traffic based on filtered criteria

    FIGURE 5 illustrates the functional model of Application Redirection, which onlyrewrites the IP header.

    VIP1 = a.b.c.d:123 VIP2 = e.f.g.h:456

    Real Dest. IP[1,2,3,...n]

    srcIP:e.f.g.h srcPort: 456 dstIP:RealIP http://www.a.com/index.html

    URLStringMatch

    HTTPHeader Cookie Cache

    SSLSession

    ID

    srcIP:a.b.c.d srcPort: 123 dstIP:VIP1 http://www.a.com/index.htmlPacket A

    Packet A1 Packet A2

    SLB function, finds the best server, and rewrites thedstIP of the target real server - Real IP

    First Function rewrote srcIP and port, so that the real serverwill reply to this switch, which is at srcIP e.f.g.h and port 456.Dest is set to the SLB function

    Client

    srcIP:e.f.g.h srcPort: 456 dstIP:VIP2 http://www.a.com/index.html

    SLB

    CustomAlgorithm

    RoundRobin

    LeastConnections

  • 8/9/2019 Sun N-tire

    16/56

    14 Network Design Patterns: N-Tier Data Centers • October 2003

    FIGURE 5   Application Redirection Functional Model

    Content Switching - Class 1 Stateful

    Content Switching is also known as Layer 7 processing, proxy switching, or URLswitching. This function accepts a client’s incoming HTTP request, terminates thesocket connection, and creates another socket connection to the target web server,which is chosen based on a user-defined rule. The Content Switching function thenfetches the requested web page and returns it to the client.

    FIGURE 6 shows an overview of the functional Content Switching mode:

    Application Redirection

    Filter has a defined destination IP

     Client request meets filter criteria,request is intercepted, IP dest isrewritten to new destination IP addrDEST = B

    servergroup 1DEST = A

    srcIP:a.b.c.d srcPort: 123 dstIP:VIP1 http://www.a.com/index.html

    client http requestDEST = A

    Client

    srcIP:a.b.c.d srcPort: 123 dstIP:VIPA http://www.a.com/index.html

    servergroup 2

    DEST = BsrcIP:a.b.c.d srcPort: 123 dstIP:VIPB http://www.a.com/index.html

  • 8/9/2019 Sun N-tire

    17/56

    Designing the Network Configuration   15

    FIGURE 6   Content Switching Functional Model

    Content Switching with full Network Address Translation (NAT) serves thefollowing purposes:

      Isolates internal IP addresses from being exposed to the public Internet.

      Allows reuse of a single IP address. For example, clients can send their webrequests to www.a.com or www.b.com, where DNS maps both a.com and b.comdomains to a single IP address. The proxy switch receives this request, with thepacket containing an HTTP header in the payload that contains the target domain(a.com or b.com), and redirects this request to the appropriate group of servers.

      Allows parallel fetching of different parts of web pages from servers optimizedand tuned for that type of data. For example, a complex web page may need GIFs,dynamic content, and cached content. With Content Switching, one set of web

    servers can hold the GIFs, and another can hold the dynamic content. The proxyswitch can make parallel fetches and retrieve the entire page at a faster rate thanwould otherwise be possible.

      Ensures requests with cookies or Secure Sockets Layer (SSL) session IDs areredirected to the same server to take advantage of persistence.

    FIGURE 6 shows that the client’s socket connection is terminated by the proxyfunction. The proxy retrieves as much of the URL as is needed to make a decision

     based on the retrieved URL. In this example, various URLs map to various server

    client httprequest

    Layer 7 Switching Function

    VIP

    http://www.a.com/SMA/stata/index.html servergroup1

    http://www.a.com/SMA/dnsa/index.html servergroup2

    http://www.a.com/SMB/statb/index.html servergroup3

    http://www.a.com/SMB/CACHEB/index.html servergroup4

    http://www.a.com/SMB/DYNA/index.html servergroup1

    servergroup 1

    stata

    servergroup 2dnsa

    servergroup 3statb

    servergroup 4cacheb

    servergroup 5dynab

    - terminate socket connection

    - get URL

    - check against rules

    - forward to servergroup/slb function- or get valid cookie, with server ID  and forward it to the same server

  • 8/9/2019 Sun N-tire

    18/56

    16 Network Design Patterns: N-Tier Data Centers • October 2003

    groups, which are virtual IP addresses. At this stage, the request may be forwardedto the server, or passed off to the SLB function that is waiting for traffic destined forthe server group.

    The proxy is configured with a virtual IP, so the switch forwards all client requestsdestined for this virtual IP to the proxy function. The proxy function rewrites the IPheader, particularly the source IP and port, so that the server sends back therequested data to the proxy instead of directly to the client.

    Network Address Translation - Class 1 Stateful

    Network Address Translation (NAT) is a critical component for security and propertraffic direction. There are two basic types of NAT: half and full. Half NAT rewritesthe destination IP address and MAC address to a redirected location such as a webcache, which returns the packet directly to the client because the source IP address isunchanged. Full NAT is where the socket connection is terminated by a proxy, so thesource IP and MAC are changed to that of the proxy server.

    NAT serves the following purposes:

      Security—Prevents exposing internal private IP addresses to the public.   IP Address Conservation—Requires only one valid exposed IP address to fetch

    Internet traffic from internal networks with invalid IP addresses.

      Redirection—Intercepts traffic destined for one set of servers and redirects it toanother by rewriting the destination IP and MAC addresses. The redirectedservers are able to send back the request directly to the clients with half NATtranslated traffic because the original source IP address has not been rewritten.

    NAT is configured with a set of filters, usually a quintuple Layer 3 rule. If theincoming traffic matches a certain filter rule, the packet IP header is rewritten oranother socket connection is initiated to the target server, which itself can bechanged, depending on the rule.

    Secure Sockets Layer (SSL) Session ID Persistence - Class 1Stateful

    SSL can be implemented in a variety of ways: in software, hardware, or both. SSLcan be terminated at the target server, an intermediate server, an SSL networkappliance, or at an SSL-capable network switch. An SSL appliance, such as Netscaleror Array Networks, tends to be implemented with a PC board and have a PCI-basedcard containing the SSL accelerator ASIC. Hence the SSL acceleration isimplemented in libraries, which offload only the mathematical computations. Therest of the SSL processing is implemented in software, directing selective functions

    to the hardware accelerator. Clearly, one immediate limitation is the PCI bus. Other

  • 8/9/2019 Sun N-tire

    19/56

    Designing the Network Configuration   17

    newer SSL devices have an SSL accelerator integrated into the datapath of thenetwork switch. These are very advanced products, just starting to emerge fromstartups such as Wincom Systems.

    FIGURE 7 shows an example of switch activity. Once a client has made initial contactwith a particular server, which may have been selected based on SLB, the switchensures that subsequent requests are forwarded to the same SSL server based on theSSL ID that the switch has stored during the initial SSL handshake. The switch keepsstate information based on the client’s initial request for HTTPS on port 443, whichcontains a hello message. This first request is then forwarded to the server selected

     by the SLB algorithm or by another function. The server responds to the client’shello message with an SSL session ID. The switch then intercepts this SSL session

    and stores it in a table. The switch forwards all of the client’s subsequent requests tothe same server, as long as each request contains the SSL session ID in the HTTPheader. There may be several different TCP socket connections that span the sameSSL session. State is maintained by the SSL Session ID in each HTTP request sent bythe same client.

    FIGURE 7   Network Switch with Persistence Based on SSL Session ID

    An appliance can be added for increased performance in terms of SSL handshakesand bulk encryption throughput.  FIGURE 8 shows how an SSL appliance could bedeployed. Client requests first come in for a particular URL using the HTTPSprotocol on port 443. The switch recognizes that this needs to be directed to theappliance, which is configured to provide that SSL service. A typical appliance, suchas Netscaler, can also be configured to provide Content Switching and LoadBalancing in addition to SSL acceleration. The appliance then reads or insertscookies and resubmits the HTTP request to an appropriate server, which maintainsstate based on the cookie in the HTTP header.

    SSL Server1Client

    SSL tunnel - SSL session IDHTTP Session  1

    HTTP Session  2

    HTTP Session  3

    HTTP Session  4

    Network

    Switch

  • 8/9/2019 Sun N-tire

    20/56

    18 Network Design Patterns: N-Tier Data Centers • October 2003

    FIGURE 8   Tested SSL Accelerator Configuration

    Cookie Persistence - Class 1 StatefulThe HTTP/1.0 protocol was originally designed to provide static pages in onetransaction. As more complex web sites evolved, which required multiple HTTP getson the same server, it was later discovered that performance was severely limited bythe tearing down and opening of TCP socket connections. This problem was solved

     by HTTP 1.1, which allowed persistent connections. Immediately after a socketconnection, the client could pipeline multiple requests. However, as more complex

    web sites evolved for applications such as the shopping cart, persistence acrossmultiple HTTP 1.1 requests was required. The problem was further complicated byproxies and load balancers that interfered with the traffic being redirected to thesame web server. Another mechanism was required to maintain state across multipleHTTP 1.1 requests. The solution was the introduction of two new headers in theHTTP request, Set-Cookie and Cookies, as defined in RFC 2109. These headerscarried the state information between the client and server. Typically, most load-

     balancing switches have enough intelligence to ensure that a particular client’ssession with a particular server is maintained, based on the cookie that is inserted bythe server and maintained by the client.

    Client

    Internet

    SSL http

    http

    MultilayerSwitch

    Session PersistenceBased on SSL Session ID

    Session PersistenceBased on Cookie

    Sun ONEWeb Servers

    SSL Accelerator ApplianceKey Exchange and Bulk Encryption

    https

  • 8/9/2019 Sun N-tire

    21/56

    Designing the Network Configuration   19

    FIGURE 9   Network Availability Strategies

    Designing Networks for Availability

    FIGURE 9 shows a cross-sectional view of the tier types and the functions performedat each tier. It also shows the availability strategies for the network and web tier.External tier availability strategies are outside the scope of this article. Ourdiscussion is limited to the services tiers: Web, Application Services, Naming, and soon.

    Designing network architectures for optimal availability requires maximizing twoorthogonal components:

      Intra Availability—Refers to maximizing the function that considers theestimated failure probability of the components themselves. The components thatcause the failure are only considered by the following equation:

    F Availability = MTBF   ÷ (MTBF + MTTR)

    Where MTBF is Mean Time Between Failures and MTTR is Mean Time toRecovery.

    Client Tier External NetworkConnectivity

    Network Tier Services Tier

    Layer 3 redundancy

    session based services

    requires session sharing.

    Other stateless servicesfailover with no problem.

    Sun Servers with IPMP-

    dual NICs

  • 8/9/2019 Sun N-tire

    22/56

    20 Network Design Patterns: N-Tier Data Centers • October 2003

      Inter Availability—Refers to minimizing the impact of failures caused by factorsexternal to the system by the surrounding environment, such as single points of failure (SPOFs), power outages, or a technician accidently pulling out a cable.

    It is not sufficient to simply maximize the F Availabili ty function. The SPOF andenvironmental factors also need to be considered. The networks designed in thisarticle describe a highly available architecture which conforms to these designprinciples.

    FIGURE 10 provides an overview of the logical network architecture. This shows howthe tiers map to the different networks, which are also mapped to segregate VLANs.This segregation allows inter-tier traffic to be controlled by filters on the switch, orpossibly a firewall (which is the only bridge point between VLANs).

  • 8/9/2019 Sun N-tire

    23/56

    Designing the Network Configuration   21

    FIGURE 10   Logical Network Architecture Overview

    The following list describes each subnetwork shown in  FIGURE 10:

      External network—The external-facing network that directly connects to theinternet. All IP addresses must be registered, and it is advisable to secure themwith a firewall.

     

    Externalnetwork

    192.168.10.0/24

    Web servicenetwork

    10.10.0.0/24

    Namingservices network

    10.20.0.0/24

    Loadbalancer

    Loadbalancer

    Managementnetwork

    10.100.0.0/24

    Accessto all

    networksDatabase

    service network10.100.0.0/24

    Backupnetwork

    10.110.0.0/24

    Devicenetwork

    10.0.0.0/24

    Applicationservices network

    10.30.0.0/24

    Clientnetwork

    172.16.0.0/24

    Productionnetwork

    Externalnetwork

    Managementnetwork

  • 8/9/2019 Sun N-tire

    24/56

    22 Network Design Patterns: N-Tier Data Centers • October 2003

    The following networks are assigned non-routable IP addresses, as specified byRFC 1918, which can also be based on the following:

      10.0.0.0 - 10.255.255.255 (10/8 prefix)

      172.16.0.0 - 172.31.255.255 (172.16/12 prefix)   192.168.0.0 - 192.168.255.255 (192.168/16 prefix)

      Web Services network—A dedicated network containing web servers. Typicalconfigurations include a load-balancing switch. This switch can be configured toallow the web server to return the client’s HTTP request directly, or require theload-balancing device to return the request on behalf of the provider’s web server.

      Naming Services network—A dedicated network consisting of servers thatprovide LDAP, DNS, NIS, and other naming services. The services are for internal

    use only, and should be highly secure. Internal infrastructure support servicesmust be sure that requests originate and are destined for internal servers. Mostrequests tend to be read intensive, hence the potential use of caching strategies forincreased performance.

      Management network—A dedicated service network that provides managementand configuration for all servers, including the JumpStart installation option fornew systems.

      Backup network—A dedicated service network that provides backup and restoreoperations. This network is pivotal to minimizing disturbances to otherproduction service networks during backup and other network bandwidth-intensive operations.

      Device network—A dedicated network that attaches IP devices for storage, aswell as other devices.

      Application Services network—A dedicated network, typically consisting of larger multi-CPU servers that host multiple instances of the Sun ONE Application

    Server software image. These requests tend to only require low network bandwidth, but can span multiple protocols, including HTTP, CORBA,proprietary TCP, and UDP-based protocols. The network traffic can also besignificant when Sun ONE application server clustering is enabled. Every updateto a stateful session bean triggers a multicast update to all servers on thisdedicated network so that participating cluster nodes update the appropriatestateful session bean. Network utilization increases directly proportional to theintensity of session bean updates.

      Database network—A dedicated network, typically consisting of one or twomulti-CPU database servers. Network traffic typically consists of jdbc traffic

     between the application server and the web server.

  • 8/9/2019 Sun N-tire

    25/56

    Designing the Network Configuration   23

    Networks and VLANsEach service is deployed in a dedicated Class C network where the first three octets

    represent the network number. The design represents an innovative approach whereseparate Layer 2 devices are not required because the functionality was collapsedinto the core switch. Decreasing the management and configuration of separatedevices while maintaining the same functionality is a major step toward cuttingcosts and increasing reliability.

    FIGURE 11 shows how a traditional configuration requires two Layer 2 switches. Aparticular VLAN spans the six segments, which gives each interface access to theVLAN on failover.

    FIGURE 11   Traditional Availability Network Design Using Separate Layer 2 Switches

    The design shown in  FIGURE 12 results in the same network functionality, buteliminates the need for two Layer 2 devices. This is accomplished using a taggedVLAN interconnect between the two core switches. Collapsing the Layer 2functionality reduces the number of network devices. This provides a reduced riskof unit failure, lower cost, and reduced manageability issues.

    Edge switch

    Master

    switch

    (layer 3)

    Standby

    switch

    (layer 3)

    Layer 2

    switch

    Layer 2

    switch

    Client network

    Network

    interface 0

    Network

    interface 1

    Sun server

  • 8/9/2019 Sun N-tire

    26/56

    24 Network Design Patterns: N-Tier Data Centers • October 2003

    FIGURE 12   Availability Network Design Using Large Chassis-Based Switches

    Solaris IPMP Host Network Interface RedundancyIntroduction

    Solaris 8 introduced a novel technology that added network load balancing and path

    failover with IP Multi Path (IPMP) built into the base operating system. IPMP isused to provide failover between multiple physical network interfaces. Itsadvantages over NAFO and other network interface redundancy approaches aresupport for outbound load balancing, and a single mechanism to support bothclustered and non-clustered nodes. Failover is typically fast enough to preserveapplication session connectivity. The default failover time is ten seconds. NormalTCP/IP recovery ensures no loss of data. Interfaces are collected into groups. Theminimum number of interfaces assigned to a group is two. Where larger domains

    are built, the interface group can scale to the total number of installed networkinterface cards (NICs) that are attached to the same VLAN.

    Client network

    Edgeswitch

    Standby core switch 2Master core switch 1

    Networkinterface 0

    Network

    interface 1

    Sun server

  • 8/9/2019 Sun N-tire

    27/56

    Designing the Network Configuration   25

    Note – Currently, configuration of Quad Fast Ethernet cards is limited to eightcards, and configuration of Gigabit Ethernet cards is limited to six cards. AdditionalI/0 cards can be added by blacklisting and adding as needed. Please consult a Sun

    Systems Engineer for the latest configuration rules.

    Solaris IPMP Host Network Interface Redundancy Internals

    FIGURE 13 describes the internal components of Solaris IPMP. The key is the GroupIdentifier (GID) associated with each physical MAC. IP forwards packets destined

    for a particular GID, as opposed to a physical interface.

  • 8/9/2019 Sun N-tire

    28/56

    26 Network Design Patterns: N-Tier Data Centers • October 2003

    FIGURE 13   IPMP Internals

    Further, IP has the ability to loadbalance across all physical interfaces that have acommon GID. As can be seen, each physical interface has a unique MAC and a

    unique test IP address (TIP). The Virtual IP address associated with each physicalinterface that belongs to the same GID has two modes of operation:

    VIP

    TIP

    GID

    MAC

    PHY

    10.6.97.

    1

    TCP

    Stream head

    10.6.

    97.53

    ha

    8:0:20:f

    f:9b:b1

    ce0

    VIP

    TIP

    GID

    MAC

    PHY

    10.6.97.

    1

    10.6.97.

    54

    ha

    8:0:20:f

    f:9b:b2

    ce1

    Appli

    cation

    in.mpath

    d

       P o  r   t   1

       P o  r   t   2

       P o  r   t   3

       P o  r   t   2  3

       P o

      r   t   1

       P o  r   t   2

       P o  r   t   3

       P o  r   t   2  3

       L  a  y  e  r   3

        F   D   B

       N  e   t  w

     o  r   k

       P o  r   t

       L  a  y  e  r   2

        F   D   B

       M  A  C

       A   D   D

       R   P o

      r   t

       P o  r   t   1

       P o  r   t   2

       P o  r   t   3

       P o  r   t   2  3

       P o  r   t   1

       P o  r   t   2

       P o  r   t   3

       P o  r   t   2  3

       L  a  y  e  r   3

        F   D   B

       N  e

       t  w o  r   k

       P o  r   t

       L  a  y  e  r   2

        F   D   B

       M  A  C

       A   D   D

       R   P o

      r   t

    Sun server

    Network switch

    Network switch

  • 8/9/2019 Sun N-tire

    29/56

    Designing the Network Configuration   27

    1. Active Active —The VIP must be specified using the addif command for eachactive interface. Only active interfaces are load balanced for outgoing packets.

    2. Active Standby—The VIP is not specified on a particular interface, and the

    standby command is issued for the standby interface. In this case, the standbyinterface does not participate in outbound load balancing. It associates with the VIPonly in the event that an active interface fails and remains down.

    It is important to configure the network switch carefully when using IPMP. Asshown in FIGURE 13, there are two types of forwarding tables: Layer 3 for IP-basedforwarding, and Layer 2 forwarding. It is important that the MAC addresses aredifferent: the switch must have a unique port associated with each MAC address.Otherwise the switch may get confused, depending on the network forwardingimplementation. Also, if a standby interface has just failed over, the VIP will beassociated with the old, failed MAC and the switch may incorrectly forward to thefailed NIC. It is important to use tested network switches that know, when a linkfails, to remove the appropriate entries from the forwarding table are re-ARPed foran updated entry.

    IPMP detects failures using two methods:

      Link level—The NIC knows when the link is physically down. The driver issues a

    message to IP to perform failure detection operations based on the configurations.   IP level—The daemon in.mpathd detects Layer 3 failures with the default router.

    Periodically in.mpathd tests the connectivity between the TIP and the defaultrouter, ensuring correct Layer 3 connectivity. In the event of a failure, a message issent to IP to to perform failure detection operations, based on the configurations.

    Solaris IPMP Host Network Interface RedundancyAvailability

    A typical highly available configuration includes a Sun server that has dual NICcards, which increases the availability of these components by several orders of magnitude. For example, the Gigabit Ethernet card, part number X2069A, by itself has an MTBF of 199,156 hours, and assuming approximately 2 hours MTTR, has anavailability of 0.999989958. With two cards, the MTBF increases so that availability

     becomes 9 9’s at .9999999996 availability. This small incremental cost has a bigimpact on the overall availability computation.

    FIGURE 14 shows the Sun server redundant NIC model using IPMP in Active Standbymode. The server has two NICs, ge0 and ge1, with fixed IP addresses a.b.c.d ande.f.g.h. The virtual IP address w.x.y.z is the IP address of the service. Clientrequests use this IP address as the destination. This IP address floats between thetwo interfaces: ge0 or ge1. Only one interface can be associated with the virtual IPaddress at any one instant. If the ge0 interface owns the virtual IP address, then data

  • 8/9/2019 Sun N-tire

    30/56

    28 Network Design Patterns: N-Tier Data Centers • October 2003

    traffic follows the P1 path. If the ge0 interface fails, the ge1 interface takes over andassociates the virtual IP address, and then data traffic follows the P2 path. Failurescan be detected within two seconds, depending on the configuration.

    .

    FIGURE 14   High-Availability Network Interface Cards on Sun Servers in Active Standby

    Layer 3—Integrated Virtual Router Redundancy Protocol(VRRP) RFC 2338 and IPMP

    By combining the availability technologies of routers and server NICs, we can createa reusable cell that can be reused in any deployment where servers are connected to

    routers. This reusable cell is highly available and scalable. In most LANenvironments, there are many servers or client nodes, with one default router. If thatdefault router fails, or the link that connects the hosts to the default router fails, thenconnectivity to the outside network is down. To solve this problem, Virtual RouterRedundancy Protocol (VRRP), as defined in RFC 2338, was introduced. VRRP is

     basically a multicast protocol used by switches to monitor one another ’s health.When a backup switch detects a failure of the master switch, it assumes the IPaddress of the master. This is the same IP address used by all servers of a particular

    VLAN as the default route. To integrate IPMP and VRRP so that both redundancyprotocols interoperate properly, care must be taken to ensure correct operating andcorrective operations in the event of a failure.  FIGURE 15 describes a workingconfiguration. Lines 1 and 2 show the VRRP protocol used by the routers to monitorone another. If one router detects that the other has failed, the surviving routerassumes the role of master and inherits the IP address and MAC address of themaster.

    Lines 3 and 5 in  FIGURE 15 show how a switch can verify that a particular connection

    is up and running. This connection can be port-based, link-based, or based on Layers3, 4, and 7. The router can make synthetic requests to the server and verify that a

    HCS HCS

    VLAN

    ge0:a.b.c.d

    w.x.y.z

    ge1:e.f.g.h

    P1

    P2

    IPMP-dual NIC

  • 8/9/2019 Sun N-tire

    31/56

    Designing the Network Configuration   29

    particular service is up and running. If the router detects that the service has failed,the VRRP can be configured on some switches to consider this failure as part of theelection algorithm and tie the failure to the priority of the VRRP router.Simultaneously, the server is also monitoring links. Currently, IPMP consists of a

    daemon, in.mpathd, that constantly pings the default router. As long as the defaultrouter can be pinged, the master interface (ge0) assumes ownership of the IPaddress. If the in.mpathd daemon detects that the default router is not reachable,automatic failover occurs, which brings down the link and floats over the IP addressof the server to the surviving interface (ge1).

    In the lab, we were able to tune IPMP and VRRP to achieve failure detection andrecovery within one second. There is a trade-off, however. Because the control

    packets are on the same network as the production network, and because VRRP is aCPU-intensive task, false failures are possible if the switches, networks, or servers become overloaded. This is because the device might take longer than the stricttimeout to respond to the peer’s heartbeat.

    FIGURE 15   Design Pattern for IPMP and VRRP Integrated Availability Solution

    N-Tier Data Center Logical Network Design

    The logical network design for the N-Tier data center (FIGURE 16) incorporates serverredundant network interfaces and integrated VRRP and IPMP.

    1

    4

    3 65

    2VRRP

    VRRP

    ge0

    ge1

    in.mpathd

  • 8/9/2019 Sun N-tire

    32/56

    30 Network Design Patterns: N-Tier Data Centers • October 2003

    FIGURE 16   Logical Network Architecture With Virtual Routers, VLANs, and Networks

    172.16.0.1

    Clients

    Edge switch

    192.168.0.1

    Master

    Servers

    Slave

    192.168.0.2

    10.10.0.110.50.0.1

    10.20.0.110.40.0.1

    10.30.0.1

    192.168.0.2

    10.10.0.1 10.50.0.1

    10.20.0.1 10.40.0.1

    10.30.0.1

  • 8/9/2019 Sun N-tire

    33/56

    Designing the Network Configuration   31

    TABLE 4 summarizes the eight separate networks and associated VLANs.

    The edge network connects redundantly to the internal reference architecturenetwork. One of the core switches has ownership of the 192.16.0.2 IP address,which means that switch is the master and the other is in slave mode. When theswitch is in slave mode, it does not respond to any traffic, including ARPs. Themaster also assumes ownership of the MAC that floats along with the virtual IPaddress of 192.16.0.2.

    Note – If you have multiple NICs, make sure each NIC uses its unique MACaddress.

    Each switch is configured with the identical networks and associated VLANS thatare shown in TABLE 4. An interconnect between the switches extends each VLAN, butis tagged to allow multiple VLANs to share a physical link. (This requires a networkinterface, such as cards that use the ce driver, that supports tagged VLANS.) TheSun servers connect to both switches in the appropriate slots, where only one of the

    two interfaces will be active.Most switches support Routing Information Protocol (RIP and RIPv2), Open ShortestPath First (OSPF), and Border Gateway Protocol v4 (BGP4). However, static routesprovide a more secure environment. A redundancy protocol based on Virtual RouterRedundancy Protocol (VRRP, RFC 2338) runs between the virtual routers. The MACaddress of the virtual routers floats among the active virtual routers, so that the ARPcaches on the servers do not need any updates when a failover occurs.

    TABLE 4   Network and VLAN Design

    Name Network Default Router VLAN Purpose

    client 172.16.0.0 172.16.0.1 client Client load generation

    edge 192.16.0.0 192.16.0.1 edge Connects client network to internalreference architecture network

    web 10.10.0.0 10.10.0.1 web Web services

    ds 10.20.0.0 10.20.0.1 ds Directory services

    db 10.30.0.0 10.30.0.1 db Database services

    app 10.40.0.0 10.40.0.1 app Application services

    dns 10.50.0.0 10.50.0.1 dns DNS services

    mgt 10.100.0.0 10.100.0.1 mgt Management and administration

    san 10.0.0.0 10.0.0.1 san Devices network

  • 8/9/2019 Sun N-tire

    34/56

  • 8/9/2019 Sun N-tire

    35/56

    Designing the Network Configuration   33

    FIGURE 17   Logical Network Describing Application Data Flow

    1 12

    2 11

    3 5

    4

    8

    9

    10

    7

    6

    Clients

    Switching

    services

    Load

    balancer

    Static contentNFS server

    Web services

    Directory

    services

    Application

    services

    Database

    services

  • 8/9/2019 Sun N-tire

    36/56

    34 Network Design Patterns: N-Tier Data Centers • October 2003

    TABLE 5   Sequence of Events for FIGURE 17

    Item Interface1 Interface2 Protocol Description

    1 Client Switch HTTP/HTTPS

    Client initiates web request. Clientcommunication can be HTTP or HTTPS(HTTP with secure socket layer). HTTPScan be terminated at the switch or at theweb server.

    2 Switch Web server HTTP/

    HTTPS

    Switch redirects client request to

    appropriate web server.3 Web Server Application

    serverApplicationserver webconnectorover TCP

    The web server redirects the request tothe application server for processing.Communication passes through a webserver plug-in over a proprietary TCP-

     based protocol.

    4 Applicationserver

    Directoryserver

    LDAP The Java 2 Enterprise Edition (J2EE)application hosted by the application

    server identifies the requested process asrequiring specific authorization. It sendsa request to the directory server to verifythat the user has valid authorization.

    5 Directoryserver

    Applicationserver

    LDAP The directory server successfully verifiesthe authorization through the user’sLDAP role. The validated response isreturned to the application server.Application server then processes

     business logic represented in J2EEapplication.

    6 Applicationserver

    Databaseserver

     JDBC The business logic requests data from adatabase as input for processing. Therequests may come from servlets, JavaData Access Objects, or Enterprise

     JavaBeans™ (EJBs) that in turn use JavaDataBase Connectivity (JDBC™) to access

    the database.7 Database

    serverApplicationserver

     JDBC The JDBC request can contain any validSQL statement. The database processesthe request natively and returns theappropriate result through JDBC to theapplication server.

    TABLE 5 Sequence of Events for FIGURE 17 (Continued)

  • 8/9/2019 Sun N-tire

    37/56

    Configuring the Physical Network   35

    Configuring the Physical NetworkThe next step involves implementing a real network based on the logical network

    design. This process involves:   “Physical Network Connectivity” on page 38   “Configuring Switches” on page 40

    There are several approaches to realize a network that functionally satisfies thelogical architectural requirements.

    The Sun ONE N-Tier Data Center is vendor independent, so you can use the

    network equipment that best suits your environment. In the lab, we built twodifferent network configurations. One configuration uses Extreme Networks BlackDiamond 6800 Core Switches, (FIGURE 18), and the other uses Foundry Networks BigIron Core Switch and Server Iron XL Load Balancer (FIGURE 19). The ExtremeNetworks switch that we used has built-in load balancing, so there is no need for anexternal load-balancing device. The Foundry Networks products required a separateload-balancing switch.

    Similar implementations could be realized using products from Cisco or Nortel.

    8 Application

    server

    Web server Application

    server Webconnectorover TCP

    The J2EE application completes the

     business logic processing, packages thedata for display, usually through a JSPthat renders HTML, and returns theresponse to the web server.

    9 Web server Switch HTTP/HTTPS

    Switch receives reply from web server.

    10 Switch Client HTTP/

    HTTPS

    Switch rewrites IP header and returnsrequest to client.

    TABLE 5   Sequence of Events for FIGURE 17 (Continued)

    Item Interface1 Interface2 Protocol Description

  • 8/9/2019 Sun N-tire

    38/56

    36 Network Design Patterns: N-Tier Data Centers • October 2003

    FIGURE 18   Sun ONE N-Tier Network Configuration With Extreme Networks Equipment

    Client 1 Client 2

    L2-L3edge switch

    Webservice

    tier

    Sun Fire6800

    Sun Fire680010.30.0.101

    Sun Fire6800

    Sun Fire6800

    10.30.0.100

    Client

    access

    Extreme

    switches

    Application

    service

    tier

    Data

    service

    tier

    Naming

    servicetier

    Extreme switch192.168.10.2

    Sun Fire 280R Sun Fire 280R Sun Fire 280R Sun Fire 280R

    Sun Fire 280R Sun Fire 280R Sun Fire 280R Sun Fire 280R

    T3

    T3

    192.168.10.1

    10.10.0.1

    10.20.0.110.40.0.110.30.0.1

    10.50.0.1

    10.10.0.100 10.10.0.101 10.10.0.102 10.10.0.103

    10.20.0.100 10.20.0.101 10.20.0.102 10.20.0.103

    10.40.0.100 10.40.0.101

    Extreme switch192.168.10.3Core Core

  • 8/9/2019 Sun N-tire

    39/56

    Configuring the Physical Network   37

    FIGURE 19   Sun ONE N-Tier Network Configuration With Foundry Networks Equipment

    Client 1 Client 2

    Level 2-3 edge switch

    Web

    service

    tier

    Sun Fire6800

    Sun Fire680010.30.0.101

    Sun Fire6800

    Sun Fire6800

    10.30.0.100

    Client

    access

    Applicationservice

    tier

    Dataservice

    tier

    Namingservice

    tier

    Master core192.168.10.2

    Sun Fire 280R Sun Fire 280R Sun Fire 280R Sun Fire 280R

    Sun Fire 280R Sun Fire 280R Sun Fire 280R Sun Fire 280R

    T3

    T3

    192.168.10.1

    10.10.0.1

    10.20.0.110.40.0.110.30.0.1

    10.50.0.1

    10.10.0.100 10.10.0.101 10.10.0.102 10.10.0.103

    10.20.0.100 10.20.0.101 10.20.0.102 10.20.0.103

    10.40.0.100 10.40.0.101

    Standby core192.168.10.3

    Server load-balancer switches

    Foundry

    switches

    Physical Network Connectivity

  • 8/9/2019 Sun N-tire

    40/56

    38 Network Design Patterns: N-Tier Data Centers • October 2003

    Physical Network ConnectivityFIGURE 20 is a standardized diagram showing the IP addresses for both the ExtremeNetworks and Foundry Networks configurations.

    The Extreme Networks configuration would be mapped directly to mls1 and mls2, because these switches incorporate SLB functionality. Foundry, however, requiresadditional load-balancing appliances.

    FIGURE 20 elements are described further in  TABLE 6.

  • 8/9/2019 Sun N-tire

    41/56

    Configuring the Physical Network   39

    FIGURE 20   Physical Network Connections and Addressing

    Client1Client2

    Edge

    mls1   mls2

    ge0:172.16.0.101/24ge0:172.16.0.102/24

    172.16.0.1/24

    192.168.0.1/24

    192.168.0.2/24   192.168.0.2/24

    192.168.0.101/24   192.168.0.102/24

    1 2 3   4

    5   67   hme0:10.100.16.101hme0:10.100.16.10210.100.16.1

    10.100.168.2

    hme010.100.10.101

    10.100.168.2

    hme0:10.100.10.105

    hme0:10.100.20.101   hme0:10.100.20.103

    192.168.0.2/24 192.168.0.2/24

    hme0:10.100.30.101   hme0:10.100.30.103

    ge0:10.10.0.101/24

    ge1:10.10.0.102/24

    ge0:10.10.0.103/24

    ge1: 10.10.0.104/24

    ge0:10.10.0.105/24

    ge1:10.10.0.106/26

    ge0:10.10.0.107/24

    ge1:10.10.0.108/26

    app1

    app2

    app1

    app2

    10.10.0.1/24   10.10.0.1/24

    ge0:10.40.0.101/24 ge1:10.40.0.102/24

    ge0:10.40.0.103/24 ge1:10.40.0.104/24

    ge0:10.40.0.105/24   ge1:10.40.0.106/24

    ge0:10.10.0.107/24   ge1:10.40.0.108/24

    web1 web1 web1 web1

    ds1   ds2

    10.20.0.1/24   10.20.0.1/24

    ge0:10.20.0.101/24

    ge1:10.20.0.102/24   ge1:10.20.0.104/24

    ge0:10.20.0.103/24

    10.30.0.1/24   10.30.0.1/24

    ge1:10.30.0.104/24

    db1 db2

    ge0:10.30.0.101/24

    ge1:10.30.0.102/24

    ge0:10.30.0.103/24

  • 8/9/2019 Sun N-tire

    42/56

    40 Network Design Patterns: N-Tier Data Centers • October 2003

    Configuring SwitchesA high-level overview of the switch configuration is shown in  FIGURE 21.

    TABLE 6   Physical Network Connections and Addressing

    Switch Description Port Speed (Mb/s)

    Base

    Address Netmask

    edge Client network to externalnetwork router

    1,2,3,4 1000 172.16.0.1 255.255.255.0

    edge External network - mls1 5,6 1000 192.168.10.1 255.255.255.0

    mls1 External network 1 1000 192.168.10.2 255.255.255.0mls1 Web/application service

    router3,4,5,6 1000 10.10.0.1 255.255.255.0

    mls1 Directory service router 7,8 1000 10.20.0.1 255.255.255.0

    mls1 Database service router 9,10 1000 10.30.0.1 255.255.255.0

    mls2 External network 1 1000 192.168.10.2 255.255.255.0

    mls2 Web/application servicerouter

    3,4,5,6 1000 10.10.0.1 255.255.255.0

    mls2 Directory service router 7,8 1000 10.20.0.1 255.255.255.0

    mls2 Database services router 9,10 1000 10.30.0.1 255.255.255.0

  • 8/9/2019 Sun N-tire

    43/56

    Configuring the Physical Network   41

    FIGURE 21   Collapsed Design Without Layer 2 Switches

    Extre

    me Ne

    tworks Su

    mmit 7

    i

    Extre

    me Ne

    tworks

     - Blac

    k Diamond

     6808

    Slot 1

    client

    172.16

    .0.1

    web

    10.10.

    0.1

    Slot 2

    ds

    10.20.

    0.1

    Slot 3

    db

    10.30.0.1

    Slot 4

    app

    10.40.0.1

    Slot 5

    DNS

    10.50.

    0.1

    Slot 6

    Slot 7

    Slot 8

    mgt

    10.100.0.

    1

    Extre

    me Ne

    tworks

     - Black D

    iamond

     6808

    Slot 1

    web

    10.10.

    0.1

    Slot 2

    ds

    10.20.

    0.1

    Slot 3

    db

    10.30.

    0.1

    Slot 4

    app

    10.40

    .0.1

    Slot 5

    DNS

    10.50.

    0.1

    Slot 6

    Slot 7

    Slot 8

    mgt

    10.10

    0.0.1

    edge

    192.168.0

    .2

    esrp in

    tercon

    nect - 

    web, d

    s, db, a

    pp, DN

    S

    edge

    192.16

    8.0.2

    esrp in

    tercon

    nect - 

    web, d

    s, db, app, 

    DNS

    edge

    192.

    168.0.2

    Configuring the Extreme Networks Switches

  • 8/9/2019 Sun N-tire

    44/56

    42 Network Design Patterns: N-Tier Data Centers • October 2003

    For the Sun ONE N-Tier, two Extreme Networks BlackDiamond switches were usedfor the core switches and one Summit7i switch was used for the edge switch.

    Note – Network equipment from Foundry Networks can be used instead. Refer to“Configuring the Foundry Networks Switches” on page 43.

     To Configure the Extreme Networks Switches

    1. Configure the core switches.

    The following example shows an excerpt of the switch configuration file.

    #

    # MSM64 Configuration generated Thu Dec 6 20:19:20 2001

    # Software Version 6.1.9 (Build 11) By Release_Master on 08/30/01 11:34:27

    configure slot 1 module g8x

    configure slot 2 module g8x

    configure slot 3 module g8x

    configure slot 4 module g8x

    configure slot 5 module g8x

    configure slot 6 module g8x

    configure slot 7 module f48t

    configure slot 8 module f48t

    .....................................................

    configure dot1q ethertype 8100

    configure dot1p type dot1p_priority 0 qosprofile QP1

    configure dot1p type dot1p_priority 1 qosprofile QP2

    configure dot1p type dot1p_priority 2 qosprofile QP3

    configure dot1p type dot1p_priority 3 qosprofile QP4

    .....................................................

    enable sys-health-check

    configure sys-health-check alarm-level log

    enable system-watchdog

    config qosprofile QP1 minbw 0% maxbw 100% priority Low minbuf 0% maxbuf 0 K

    config qosprofile QP2 minbw 0% maxbw 100% priority LowHi minbuf 0% maxbuf 0 K

    2. Configure the edge switch.

    Th f ll i l h t f th it h fi ti fil

  • 8/9/2019 Sun N-tire

    45/56

    Configuring the Physical Network   43

    The following example shows an excerpt of the switch configuration file.

    Configuring the Foundry Networks Switches

    This section describes the network architecture implementation using FoundryNetworks equipment instead of Extreme Networks equipment. The overall setup isshown in FIGURE 22.

    #

    # Summit7i Configuration generated Mon Dec 10 14:39:46 2001

    # Software Version 6.1.9 (Build 11) By Release_Master on 08/30/01 11:34:27

    configure dot1q ethertype 8100

    configure dot1p type dot1p_priority 0 qosprofile QP1

    ....................................................

    enable system-watchdog

    config qosprofile QP1 minbw 0% maxbw 100% priority Low minbuf 0% maxbuf 0 K

    ....................................................

    delete protocol ip

    delete protocol ipx

    delete protocol netbios

    delete protocol decnetdelete protocol appletalk

    ....................................................

    # Config information for VLAN Default.

    config vlan “Default” tag 1 # VLAN-ID=0x1 Global Tag 1

    config vlan “Default” protocol “ANY”

    config vlan “Default” qosprofile “QP1”

    enable bootp vlan “Default”

    ....................................................

  • 8/9/2019 Sun N-tire

    46/56

    44 Network Design Patterns: N-Tier Data Centers • October 2003

    FIGURE 22   Foundry Networks Implementation

    Client Client Client Client

    Servers Servers

    Servers Servers

    Servers Servers

    Servers Servers

    Servers Servers

    Webservice

    tier

    Directory

    service

    tier

    Application

    service

    tier

    Database

    service

    tier

    Netscreen

    firewall

    NS5200

    Netscreen

    firewall

    NS5200

    Server loadbalancer

    SLB1

    BigIronlayer 2/3 switch

    MLS0

    BigIronlayer 2/3 switch

    MLS1

    Server loadbalancer

    SLB0

    Extreme Networks

    Summit 7i

    S7i

    Foundry Networks Master Core Switch Configuration

  • 8/9/2019 Sun N-tire

    47/56

    Configuring the Physical Network   45

    The following code box shows part of the configuration file for the BigIron mastercore switch (called MLS0 in the lab).

    module 1 bi-jc-8-port-gig-m4-management-modulemodule 3 bi-jc-48e-port-100-module!global-protocol-vlan!vlan 1 name DEFAULT-VLAN by portvlan 10 name refarch by port untagged ethe 1/1 ethe 3/1 to 3/16 router-interface ve 10!vlan 99 name mgmt by port untagged ethe 3/47 to 3/48 router-interface ve 99!hostname MLS0ip default-network 129.146.138.0/16ip route 192.168.0.0 255.255.255.0 172.0.0.1ip route 129.148.181.0 255.255.255.0 129.146.138.1ip route 0.0.0.0 0.0.0.0 129.146.138.1!router vrrp-extendedinterface ve 10 ip address 20.20.0.102 255.255.255.0 ip address 172.0.0.70 255.255.255.0 ip vrrp-extended vrid 1

      backup priority 100 track-priority 20  advertise backup  ip-address 172.0.0.10  dead-interval 1  track-port e 3/1  enable ip vrrp-extended vrid 2  backup priority 100 track-priority 20  advertise backup  ip-address 20.20.0.100  dead-interval 1  track-port e 3/13  enable!interface ve 99 ip address 129.146.138.10 255.255.255.0!end

    Foundry Networks Standby Core Switch Configuration

    f f f f f

  • 8/9/2019 Sun N-tire

    48/56

    46 Network Design Patterns: N-Tier Data Centers • October 2003

    The following code box shows a partial listing of the configuration file for theBigIron standby core switch (called MLS1 in the lab).

    ver 07.5.05cT53!module 1 bi-jc-8-port-gig-m4-management-modulemodule 3 bi-jc-48e-port-100-module!global-protocol-vlan!!vlan 1 name DEFAULT-VLAN by port!vlan 99 name swan by port untagged ethe 1/6 to 1/8 router-interface ve 99!vlan 10 name refarch by port untagged ethe 3/1 to 3/16 router-interface ve 10!!hostname MLS1ip default-network 129.146.138.0/1ip route 192.168.0.0 255.255.255.0 172.0.0.1ip route 0.0.0.0 0.0.0.0 129.146.138.1!router vrrp-extended

    interface ve 10 ip address 20.20.0.102 255.255.255.0 ip address 172.0.0.71 255.255.255.0 ip vrrp-extended vrid 1  backup priority 100 track-priority 20  advertise backup  ip-address 172.0.0.10  dead-interval 1  track-port e 3/1  enable ip vrrp-extended vrid 2  backup priority 100 track-priority 20  advertise backup  ip-address 20.20.0.100  dead-interval 1  track-port e 3/13enableinterface ve 99ip address 129.146.138.11 255.255.255.0

    !

    !

    !

    !

    !

    sflow sample 512

    sflow source ethernet 3/1

    sflow enable!

    !

    !

    end

    Foundry Networks Master Server Load Balancer Configuration

    Th f ll i d b h ti l li ti f th fi ti fil d f th

  • 8/9/2019 Sun N-tire

    49/56

    Configuring the Physical Network   47

    The following code box shows a partial listing of the configuration file used for themaster Server XL server load balancer (called SLB0 in the lab).

    ver 07.3.05T12

    global-protocol-vlan

    !

    !

    server source-ip 20.20.0.50 255.255.255.0 172.0.0.10

    !

    !!

    !

    server real web1 10.20.0.1

     port http

     port http url "HEAD /"!

    server real web2 10.20.0.2

     port http

     port http url "HEAD /"

    !

    !

    server virtual WebVip1 192.168.0.100

     port http

     port http dsr

     bind http web1 http web2 http

    !

    !

    vlan 1 name DEFAULT-VLAN by port

     no spanning-tree

    !

    hostname SLB0

    ip address 192.168.0.111 255.255.255.0

    ip default-gateway 192.168.0.10

    web-management allow-no-password

    banner motd ^C

    Reference Architecture -- Enterprise Engineering^C

    Server Load Balancer-- SLB0 129.146.138.12/24^C

    !!end

    Foundry Networks Standby Server Load Balancer Configuration

    The following code box shows a partial listing of the configuration file used for the

  • 8/9/2019 Sun N-tire

    50/56

    48 Network Design Patterns: N-Tier Data Centers • October 2003

    The following code box shows a partial listing of the configuration file used for thestandby Server XL server load balancer (called SLB1 in the lab).

    Network SecurityFor the N-Tier network design, firewalls were configured between each service tier toprovide network security.  FIGURE 23 shows the relationship between the firewallsand the service tiers.

    ver 07.3.05T12

    global-protocol-vlan

    !

    !

    server source-ip 20.20.0.51 255.255.255.0 172.0.0.10

    !

    !!

    !

    server real s1 20.20.0.1

     port http

     port http url "HEAD /"!

    server real s2 20.20.0.2

     port http

     port http url "HEAD /"

    !

    !

    server virtual vip1 172.0.0.11

     port http

     port http dsr

     bind http s1 http s2 http

    !

    !

    vlan 1 name DEFAULT-VLAN by port

    !

    hostname SLB1

    ip address 172.0.0.112 255.255.255.0

    ip default-gateway 172.0.0.10

    web-management allow-no-password

    banner motd ^C

    Reference Architecture - Enterprise Engineering^C

    Server Load Balancer - SLB1 - 129.146.138.13/24^C

    !

    Cli Cli

  • 8/9/2019 Sun N-tire

    51/56

    Network Security   49

    FIGURE 23   Firewalls Between Service Tiers

    In the lab, one physical firewall device was used to create multiple virtual firewalls.Network traffic was directed to pass through the firewalls between the service tiers,as shown in  FIGURE 24.

    The core switch is only configured for Layer 2 with separate port-based VLANs. Theconnection between the Netscreen device and the core switch uses tagged VLANS.Trust zones are created on the Netscreen device, and they map directly to the taggedVLANs. The Netscreen firewall device performs the Layer 3 routing. This

    configuration directs all traffic through the firewall, resulting in firewall protection between each service tier.

    Firewall

    Firewall

    Firewall

    Web service

    tier

    Application service

    tier

    Database service

    tier

    Edge

    switch

    Intranet/Internet

    Client Client

    Client Client

  • 8/9/2019 Sun N-tire

    52/56

    50 Network Design Patterns: N-Tier Data Centers • October 2003

    FIGURE 24   Virtual Firewall Architecture Using Netscreen and Foundry Networks

    Products

    Edge

    switch

    Intranet/Internet

    Client Client

    Core

    switch

    Core

    switchVLAN*

    Database VLAN

    Netscreendevice Netscreen

    device

    VLAN*

    Application VLAN

    Web VLAN

    Database VLAN

    Application VLAN

    Web VLAN

    Web service

    tier

    Application service

    tier

    Database service

    tier

    Traffic

    *Web, application and database traffic multiplexed on one VLAN

    Netscreen Firewall

    The following code box shows a partial example of a configuration file used to

  • 8/9/2019 Sun N-tire

    53/56

    Network Security   51

    e o o g co e o s o s p e p e o co g o e se oconfigure the Netscreen device.

    set auth timeout 10

    set clock "timezone" 0

    set admin format dos

    set admin name "netscreen"

    set admin password nKVUM2rwMUzPcrkG5sWIHdCtqkAibn

    set admin sys-ip 0.0.0.0

    set admin auth timeout 0

    set admin auth type Local

    set zone id 1000 "DMZ1"

    set zone id 1001 "web"

    set zone id 1002 "appsrvr"

    set zone "Untrust" block

    set zone "DMZ" vrouter untrust-vr

    set zone "MGT" block

    set zone "DMZ1" vrouter trust-vr

    set zone "web" vrouter trust-vr

    set zone "appsrvr" vrouter trust-vr

    set ip tftp retry 10set ip tftp timeout 2

    set interface ethernet1 zone DMZ1

    set interface ethernet2 zone web

    set interface ethernet3 zone appsrvr

    set interface ethernet1 ip 192.168.0.253/24

    set interface ethernet1 route

    set interface ethernet2 ip 10.10.0.253/24

    set interface ethernet2 route

    set interface ethernet3 ip 20.20.0.253/24

    set interface ethernet3 route

    unset interface vlan1 bypass-others-ipsec

    unset interface vlan1 bypass-non-ip

    set interface ethernet1 manage ping

    unset interface ethernet1 manage scs

    unset interface ethernet1 manage telnet

    unset interface ethernet1 manage snmp

    unset interface ethernet1 manage global

    unset interface ethernet1 manage global-pro

    unset interface ethernet1 manage sslset interface ethernet1 manage web

    unset interface ethernet1 ident-reset

    set interface vlan1 manage ping

    set interface vlan1 manage scs

    set interface vlan1 manage telnet

    set interface vlan1 manage snmp

    set interface vlan1 manage global

    set interface vlan1 manage global-pro

    set interface vlan1 manage ssl

    set interface vlan1 manage web

    set interface v1-trust manage ping

    set interface v1-trust manage scs

    set interface v1-trust manage telnet

  • 8/9/2019 Sun N-tire

    54/56

    52 Network Design Patterns: N-Tier Data Centers • October 2003

    set interface v1-trust manage snmp

    set interface v1-trust manage global

    set interface v1-trust manage global-pro

    set interface v1-trust manage ssl

    set interface v1-trust manage web

    unset interface v1-trust ident-reset

    unset interface v1-untrust manage ping

    unset interface v1-untrust manage scs

    unset interface v1-untrust manage telnet

    unset interface v1-untrust manage snmp

    unset interface v1-untrust manage global

    unset interface v1-untrust manage global-pro

    unset interface v1-untrust manage ssl

    unset interface v1-untrust manage web

    unset interface v1-untrust ident-reset

    set interface v1-dmz manage ping

    unset interface v1-dmz manage scs

    unset interface v1-dmz manage telnet

    unset interface v1-dmz manage snmp

    unset interface v1-dmz manage global

    unset interface v1-dmz manage global-pro

    unset interface v1-dmz manage ssl

    unset interface v1-dmz manage web

    unset interface v1-dmz ident-reset

    set interface ethernet2 manage ping

    unset interface ethernet2 manage scs

    unset interface ethernet2 manage telnet

    unset interface ethernet2 manage snmp

    unset interface ethernet2 manage global

    unset interface ethernet2 manage global-pro

    unset interface ethernet2 manage sslunset interface ethernet2 manage web

    unset interface ethernet2 ident-reset

    set interface ethernet3 manage ping

    unset interface ethernet3 manage scs

    unset interface ethernet3 manage telnet

    unset interface ethernet3 manage snmp

    unset interface ethernet3 manage global

    unset interface ethernet3 manage global-pro

    unset interface ethernet3 manage ssl

    unset interface ethernet3 manage web

    unset interface ethernet3 ident-reset

    set interface v1-untrust screen tear-drop

    set interface v1-untrust screen syn-flood

    set interface v1-untrust screen ping-death

    set interface v1-untrust screen ip-filter-src

    set interface v1-untrust screen land

    set flow mac-flooding

    set flow check-session

    set address DMZ1 "dmznet" 192.168.0.0 255.255.255.0set address web "webnet" 10.10.0.0 255.255.255.0

    set address appsrvr "appnet" 20.20.0.0 255.255.255.0

    set snmp name "ns208"

    set traffic-shaping ip_precedence 7 6 5 4 3 2 1 0

    t ik li h ki

  • 8/9/2019 Sun N-tire

    55/56

    About the Authors   53

    About the AuthorsDeepak Kakadia is a staff engineer and network architect in the ReferenceArchitecture Engineering Group for Sun Microsystems in Menlo Park, California.Deepak has been with Sun since 1994. He has previously worked with CoronaNetworks, Digital Equipment Corp., and Nortel Networks. He has a Bachelor’sDegree in Computer Systems Engineering, a Master’s Degree in Computer Science,has completed Ph.D qualifying exams, and has completed his SCPD certificate inNetworking through the Department of Electrical Engineering at StanfordUniversity. He has filed four patents in the area of Networking and SystemsManagement.

    Richard Croucher is Chief Architect of Sun’s Professional Services in EMEA. He isalso an SMI Technical Director. Richard joined Sun in 1995. He has nearly 20 years of experience with UNIX and nearly 30 years of experience in the IT and electronicsindustries. He holds a Higher National Certificate in Applied Physics, a Higher

    National Diploma in Electronics, and has received a post-graduate Certificate of Advanced Study in Non-Destructive Testing and Materials Science from BrunelUniversity. He was elected to membership of the British Computer Society in 1993.

    set ike policy-checking

    set ike respond-bad-spi 1

    set ike id-mode subnet

    set l2tp default auth local

    set l2tp default ppp-auth any

    set l2tp default radius-port 1645

    set policy id 0 from DMZ1 to web "dmznet" "webnet" "ANY" Permit

    set policy id 1 from web to DMZ1 "webnet" "dmznet" "ANY" Permit

    set policy id 2 from DMZ1 to appsrvr "dmznet" "appnet" "ANY" Permit

    set policy id 3 from appsrvr to DMZ1 "appnet" "dmznet" "ANY" Permit

    set ha interface ethernet8

    set ha track threshold 255

    set pki authority default scep mode "auto"

    set pki x509 default cert-path partial

     _____________________ 

    Acknowledgements

  • 8/9/2019 Sun N-tire

    56/56

    54 Network Design Patterns: N-Tier Data Centers • October 2003

    AcknowledgementsSincere thanks to Bill Sprouse, Kemer Thomson, Brad Carlile, and many others whoprovided tremendous support, guidance, and direction to this article.

    Related ResourcesSun ONE Directory Server 5.1 documentation collection

    Note: This product was formerly known as the iPlanet™ Directory Serversoftware. You can locate the documentation collection at:

    http://docs.sun.com

    Search for the phrase directory server 5.1. Start with the  iPlanet Directory

    Server 5.1 Deployment Guide.

    Ordering Sun Documents

    The SunDocsSM

    program provides more than 250 manuals from Sun Microsystems,Inc. If you live in the United States, Canada, Europe, or Japan, you can purchasedocumentation sets or individual manuals through this program.

    Accessing Sun Documentation OnlineThe docs.sun.com web site enables you to access Sun technical documentationonline. You can browse the docs.sun.com archive or search for a specific book titleor subject. The URL is http://docs.sun.com/

    To reference Sun BluePrints™ OnLine articles, visit the Sun BluePrints OnLine web site at:

    http://www.sun.com/blueprints/online.html