Virtual private data network session count limitation

[0002] 1. Field of the Invention

[0003] The present invention relates to the field of data communications networks. More particularly, this invention relates to a method and apparatus for limiting the number of virtual private data network (VPN or VPDN) sessions provided to a group of users locally and network-wide in a data communications network.

[0004] 2. The Background

[0005] ISPs (Internet Service Providers) and Telcos (telephone companies) (collectively referred to as “Wholesale Providers” or “Wholesalers”) typically offer various forms of retail Internet access to their subscribers. In a typical retail model as shown in FIG. 1, a user will contact a network access server (NAS) either using dial-up telephone lines, digital subscriber lines, ISDN (integrated services digital network) lines, or the like. The NAS interfaces the user with the data communications network. The way this works is that a user, for example, [email protected], dials in to a NAS at ISP's Point of Presence (PoP₁) on the Internet. A PPP (point to point protocol) session 12 is raised between NAS₁14, and Joe's terminal 16. An LCP (Link Control Protocol) session is raised between NAS₁and Joe's terminal. At this time the NAS₁generates an AAA authentication request using a protocol such as RADIUS (Remote Authentication Dial-In User Service) to the ISPs AAA (authentication, authorization and accounting) service 20. The AAA service 20 handle Joe's authentication (receipt and verification of password and user name), provisions him with appropriate authorizations, and handles accounting for the time and services used by Joe on the data communications network. The AAA service uses a local database 22 to store and retrieve AAA information. To complete Joe's log-in, an access-accept packet is sent to NAS₁from AAA service 20. Then an IPCP (Internet Protocol Control Protocol) session is raised between NAS₁and Joe's terminal during which an IP address is returned to configure Joe's terminal's PPP stack. This completes the log-in of Joe.

[0006] Wholesale Providers also offer wholesale internet access to subsidiary and specialized service providers, CLECs (Competitive Local Exchange Carriers), corporations, and Community of Interest (COI) providers. Naturally, the processing afforded customers of the wholesale variety differs from the processing afforded customers of the retail variety. Subscriber information for individual wholesale users is usually stored by those who lease data communications network access from the ISP or Telco. Hence, corporations, CLECs and COI providers do not normally share their user information with the wholesale providers. The Wholesaler, however, typically also has its own retail subscribers whose user information is stored in its databases. In some cases, a particular user might have accounts with both a retail and a wholesale provider. Hence, the Wholesaler must distinguish between the user's wholesale and retail accounts and initiate different actions based upon their status or Service Level Agreements (SLAs).

[0007] Traditional wholesale ISPs and Roaming Service Providers offer network access through a technique called “authentication proxying.” Proxying involves the transfer of the authentication responsibility to the “owner” of the subscriber. Thus, if a corporation was to outsource its corporate intranet to a Wholesaler, it would give up the maintenance of its dial-up servers (i.e., the NASes). It would not, however, normally want to give up the control of or information regarding its employees. Hence, when a corporate user connects to such a Wholesaler's network access servers, the user essentially perceives that the user is dialing into a corporate facility when the user is actually dialing into the Wholesaler's domain and then somehow gaining admittance to the corporation's intranet.

[0008] What really happens in that scenario is that the Wholesaler determines that the user belongs to Corporation A (Corp_A) by parsing either the fully qualified domain name (“FQDN”) (e.g., [email protected]) supplied by the user, reading the DNIS ID associated with the call, reading the CLID associated with the call, or by using some other known mechanism. Using a DNIS ID, the Wholesaler looks at the telephone number (or a specific NAS in access networks other than dial-up) through which the user is connecting to the network. So if a user calls in to 123-456-7890 from his number of 123-444-5555, then the Wholesaler can know which number was called, i.e., the completing station. Having determined that the user trying to gain access belongs to Corp_A, the Wholesaler cannot authenticate the user by itself. As noted earlier, the user's record is still located on Corp_A's equipment. Hence, the Wholesaler will “proxy” out the authentication transaction from its AAA proxy service to Corp_A. An AAA service within the corporation domain then identifies the user, verifies the password, and provisions the user with appropriate authorizations. It may also receive accounting information, if desired. Then the AAA service at Corp_Anotifies the Wholesaler's proxy service that the user is acceptable and passes along provisioning details associated with the user (such as an IP (Internet protocol) address to use or a pool identification of an IP address pool from which an IP address needs to be allocated and any other information that may be needed). The Wholesaler then grants the user access to the network based upon the reply it gets back from Corp_A. This technique is called “proxying.” This is shown diagrammatically in FIG. 2.

[0009] To be able to perform basic proxying, the Wholesaler maintains minimal information on its proxy service 24 at its PoP. Information such as supported domain names, the IP address to which the transaction is to be sent, the port number (typically an OSI Layer 4 port number) to which the transaction is to be addressed, a shared secret between the proxy service and the remote AAA service, etc., are typically stored on proxy service 24's local configuration database 30.

[0010] For example, user [email protected] dials in to NAS₁. A PPP (point to point protocol) session 26 is typically raised between Joe's terminal and NAS₁as is a LCP (Link Control Protocol) session. At this time the NAS₁generates an authentication request using a protocol such as RADIUS (Remote Authentication Dial-In User Service) to proxy service 24. Proxy service 24 then consults its local configuration database 30. Proxy service 24 then makes a determination about where to send the access-request packet. Here it decides to send it to the AAA service 32 maintained in the Corp_Adomain 34. The Corp_AAAA 32 then consults its local database 36 and authenticates [email protected]. Corp_AAAA 32 then returns an access-accept packet to proxy service 24 which, in turn, sends an access-accept packet to NAS₁. Then an IPCP (Internet Protocol Control Protocol) session is raised between NAS₁and Joe's terminal during which an IP address is returned to configure Joe's terminal's PPP stack thus completing the log-in of [email protected].

[0011] Frequently a large corporation or similar entity will have a need to provide PoPs at a number of locations to service its clients, customers and/or employees in a number of different cities. For example, a corporation “Corp_A” located in Los Angeles, Calif. might have some employees using dial-up lines from San Francisco, Calif. and New York City, N.Y. Particularly in this situation, and in other situations, it is frequently desirable to establish a secure virtual private network (VPN) session which uses a tunneling protocol such as L2TP (layer two tunneling protocol) to effectively direct the communications from one node on the Internet to another node on the Internet. Communications using these techniques are particularly secure because the entire packet is encapsulated by the protocol and within the encapsulation the data is encrypted. A particularly good way to establish the VPN session is between the NAS that the user is connected to and a gateway device at the location with which the user is to communicate. This might be, for example, a corporate intranet.

[0012] The overhead of a VPN session exceeds that of a normal session of a retail user logged in to a NAS because the NAS must handle the encryption duties and the additional overhead of the larger encapsulated packets. As a result, it would be desirable to provide a mechanism for accounting for such sessions, charging for them and limiting their use to a quantity contracted for. Doing so is, however, not particularly easy given the current limited accounting capabilities on the Internet.

[0013] For example, if Corp_Ahas a great number of employees in San Francisco, they could overwhelm the PoP with VPN sessions back to the home office and possibly prevent regular retail or other wholesale customers of the Wholesaler from enjoying the quality and level of service that they have paid for. Similarly, a large number of employees using uncontrolled VPN sessions spread over many regions could potentially overwhelm the network maintained by the Wholesaler. Accordingly, the Wholesaler would like to enter into an arrangement with Corp_Awhereby Corp_Apays a fee for a more or less specific maximum number of VPN sessions to occur at any one time. When Corp_Aexceeds this contracted number it is either cut off or charged an extra fee. In this way, the Wholesaler is able to plan for its expansion and receive realistic information on the number of these sessions that it must be able to support.

[0014] A data communications network with a plurality of PoPs maintains a local database associated with each PoP and a central database somewhere on the data communications network. The local database contains a group identification such as a domain identification corresponding to a group of users, a maximum number of VPN sessions to provide the group of users at the PoP and a dynamic VPN session count corresponding to active VPN sessions currently provided to the group of users at the PoP. The central database contains a maximum number of VPN sessions to provide the group of users over the entire data communications network and a dynamic network-wide VPN session count corresponding to active VPN sessions currently provided to the group of users on the entire data communications network. Actions are taken when the group attempts to exceed either the local maximum number of sessions or the network-wide maximum number of sessions by more than a predetermined number. The actions may include assessing extra charges, denying access, and sending warning messages to appropriate recipients.

[0015] FIG. 1 is a system block diagram of a simple ISP PoP using a conventional retail-only paradigm.

[0016] FIG. 2 is a system block diagram of wholesale ISP PoP using a conventional wholesale-only paradigm.

[0017] FIG. 3 is a diagram illustrating the information stored in a central database in accordance with a presently preferred embodiment of the present invention.

[0018] FIG. 4 is a diagram illustrating the information stored in a local database in accordance with a presently preferred embodiment of the present invention.

[0019] FIG. 5 is a system block diagram of a data communications network in accordance with a presently preferred embodiment of the present invention.

[0020] FIG. 6 is a flow diagram of a process for verifying slot availability for a VPN user on a data communications network in accordance with a presently preferred embodiment of the present invention.

[0021] FIG. 7 is a flow diagram of a process for updating slot availability information in a data communications network in accordance with a presently preferred embodiment of the present invention.

[0022] FIG. 8 is a system diagram of a data communications network in accordance with a presently preferred embodiment of the present invention.

[0023] FIG. 9 is a block diagram of a broker-publisher system used in a presently preferred embodiment of the present invention.

[0024] Those of ordinary skill in the art will realize that the following description of the present invention is illustrative only and not in any way limiting. Other embodiments of the invention will readily suggest themselves to such skilled persons after a perusal of the within disclosure.

[0025] In accordance with a presently preferred embodiment of the present invention, the components, processes and/or data structures may be implemented using C++ programs running on high performance computers (such as an Enterprise 2000™ server running Sun Solaris™ as its operating system. The Enterprise 2000™ server and Sun Solaris™ operating system are products available from Sun Microsystems, Inc. of Mountain View, Calif.). Different implementations may be used and may include other types of operating systems, computing platforms, computer programs, firmware and/or general purpose machines. In addition, those of ordinary skill in the art will readily recognize that devices of a less general purpose nature, such as hardwired devices, devices relying on FPGA (field programmable gate array) or ASIC (Application Specific Integrated Circuit) technology, or the like, may also be used without departing from the scope and spirit of the inventive concepts disclosed herein.

[0026] In accordance with one embodiment of the present invention the AAA proxy service may be implemented within a protocol gateway (PGW). PGWs are devices which couple users via a network access server (NAS) to the data communications network by dynamically converting protocols. The term gateway is not meant to be limited to a single type of device, as any device, hardware or software, that may act as a bridge between the user and the network may be considered a gateway for the purposes of this application. In accordance with one presently preferred embodiment of the present invention, the PGW may be a software service operating on a general purpose computer running the User Control Point (UCP) software package available from Cisco Systems, Inc. of San Jose, Calif.

[0027] The authentication, authorization and accounting (AAA) service performs user authentication, user authorization and user accounting functions. It may be a Cisco ACS™ product such as Cisco Secure™, available from Cisco Systems, Inc. of San Jose, Calif., or an equivalent product. In accordance with a presently preferred embodiment of the present invention, the Remote Authentication Dial-In User Service (RADIUS) protocol is used as the communication protocol for carrying AAA information. RADIUS is an Internet standard track protocol for carrying authentication, authorization, accounting and configuration information between devices that desire to authenticate their links and a shared AAA or AAA proxy service. Those of ordinary skill in the art will realize that other authentication protocols such as TACACS+ or DIAMETER can be used as acceptable authentication communications links between the various communications devices that encompass the data communications network and still be within the inventive concepts disclosed herein.

[0028] In accordance with the present invention wholesale users such as Corp_Aand Corp_Bwill subscribe with a Wholesaler for a certain level of VPN coverage both at particular PoPs on the network and network-wide. For example, the wholesale user could subscribe to a maximum number of VPN sessions of 50 network wide and 15 at a particular PoP. If the maximum is exceeded either at the PoP or network-wide then the next attempted log in can be handled in a number of different ways. First, it can simply be refused. Second, it can be accepted up to an “absolute maximum” of, say, 100 network-wide and 25 at the PoP. This absolute maximum provides some warning room before denying service to a user. Optionally, wholesale customers can be warned when they are exceeding the maximum without suffering a denial of service. They may also be charged an extra fee for the extra service, if desired. The extra fee would be charged by initiating a special RADIUS accounting packet in response to the log in over the maximum and sending it to the proper AAA service. Ideally the wholesale customer would adjust its service contract to the level of use it is actually making of the network and the PoPs.

[0029] To implement the invention, it is necessary to keep track of data not previously tracked in a data communications network. Central database 38 shown in FIG. 3 maintains a list identifying domains (or otherwise identifying customer “groups”) that have subscribed to wholesale VPN access. Each domain or other group identification has associated therewith a value for “network maximum sessions” which indicates the number of VPN sessions that the group is ordinarily entitled to on a network-wide basis. In the example of FIG. 3, Corp_Ais entitled to 50 and Corp_Bis entitled to 75. Normally this means that if a user from Corp_Adials into a PoP, the Wholesaler is required to service the user as long as the user's log in won't exceed the maximum network-wide or at the PoP. The Wholesaler may provide this access in a conventional manner.

[0030] Note that the “group” may be based on network domain name such as “Corpa.com”, or on some other determinable characteristic. Since the DNIS ID and/or the CLID may be used, it is possible to group users by the telephone numbers that they call from or dial into. Hence a group may have any number of diverse users in it. The term “group” will be used herein for simplicity, but it is meant in this broader sense to be inclusive of any such determinable group of users.

[0031] Optionally, a “Network Absolute Maximum Sessions” value may be kept to indicate how many sessions the group may actually get before denial of service. In the example of FIG. 3 Corp_Ais entitled to 100 and Corp_Bis entitled to 150. As pointed out above, there may be an extra charge levied for exceeding the network maximum sessions value.

[0032] The Central Database 38 also keeps a count of the current VPN sessions in effect network wide. This count is maintained by the broker-publisher system described in more detail below.

[0033] A Wholesaler may wish to take action in a conservative manner, as by adding a fixed number of “grace” sessions to avoid penalizing a client before the client has clearly exceeded the terms of its service contract by more than a reasonable level. On the other hand, the “grace” number may also be set to zero.

[0034] The Central Database 38 may also contain additional information as desired including a mirror of same or all of the data contained in the local databases 40 if desired.

[0035] Local Databases 40 as shown in FIG. 4 are associated with each PoP. They may be stored in local cache memory at each PoP or in any other convenient fashion. There is no requirement that they be stored at the PoP, it just may be convenient to do so.

[0036] Local Database 40 operates much like Central Database 38. Local Database 40 relates to a single PoP and contains a list of domain or customer identifications coupled with a local maximum sessions value, local absolute maximum sessions value and a local session count value indicative of the actual current number of VPN sessions belonging to the group and occurring at the particular PoP.

[0037] As before, attempts to log in more than the permitted number of local VPN sessions can be handled in a number of different ways. Local VPN session log ins increment the appropriate domain's local session count and log outs decrement the count.

[0038] FIG. 5 is a simplified block diagram of a data communications network 50 in accordance with a presently preferred embodiment of the present invention. The Network includes PoP₁, POP₂and PoP₃. Each of PoP₁, PoP₂and PoP₃has an associated local database 52, 54, 56 as shown. A network operations center (NOC) includes a primary mother cache (PMC) and a redundant secondary mother cache (SMC) as known to those of ordinary skill in the art. The PMC and SMC are preferably used to store the Central Database 38. A VPN session 58 has been raised between a NAS at PoP₁and a gateway device at Corp_Aas shown. While the VPN session 58 is routed in a conventional way, the VPN encapsulation and encryption of the packets isolates it from any likely interception, hence it is shown as a separate, isolated link.

[0039] FIG. 6 is a flow diagram illustrating the log in process for a VPN user in accordance with a presently preferred embodiment of the present invention. At block 60 the user attempts the log in at a local PoP and raise a VPN session. At block 62 the group of the user is determined. This may be done in one of several ways as discussed above, but most commonly involves parsing the FQDN of the user to determine his domain. At block 64 the local database 40 is searched for the group entry corresponding to the user. If VPN session slots are available (i.e., absolute maximum sessions is not exceeded if this session is to be allowed) then control transfers to block 70. Otherwise, the VPN session is rejected at block 66 and optionally the group administrator and/or user is notified in a conventional manner as set forth in the process at block 68.

[0040] Optionally, some sort of “grace” level may be applied at block 64 so that rejection of the VPN session does not occur precisely at the level specified in the service agreement with the domain.

[0041] At block 70 the central database 38 is queried to see if VPN slots are available. If they are, control transfers to block 72, otherwise the VPN session attempt may be rejected. “Grace” slots may also be considered for application here.

[0042] At block 72, the session count in the local database 40 is incremented. At block 74 the session count in the central database 38 is incremented. At block 76 the user is permitted to raise the VPN session 58 to access the desired domain. The user must still be authenticated and authorized, presumably by an AAA service at Corp_A.

[0043] When a user drops the VPN session, e.g., at log out, it is important to decrement the session counts so that an accurate session count is always kept locally as well as centrally. FIG. 7 shows how this is done. At block 78 the user initiates the log out process. At block 80 the local session count in local database 40 is decremented. At block 82 the central session count in central database 38 is decremented.

[0044] FIG. 8 is a system block diagram of a portion of data communication network 50 in accordance with a presently preferred embodiment of the present invention. Network Operations Control Center (NOC) 100 is located centrally on the network 50. There may be more than one NOC on a particular data communications network. Typically each Wholesaler will have at least one NOC. The Primary Mother Cache (PMC 102) and Secondary Mother Cache (SMC 104) are memory storage components associated with NOC 100. An access database adapter 106 is in communication with PMC 102 and SMC 104 as well as an information bus 108. This adapter, and other adapters in the invention, provide bi-directional mapping of information between components and other services comprising the data communications network 50. Adapters, such as the access database adapter 106 subscribe to and publish events. An event is an independent entity which contains an unspecified amount of non-time critical information. For example, the access database adapter 106 receives commands to publish an event. The information contained in the event may be found in the request or the access database adapter 106 may communicate with the PMC 102 to find the required information. The event is then published to other services and components within the data communications network across an information bus 108 which may be the data communications network itself. Of course this is only one way to achieve data synchronization—once can easily implement a centralized model or a fully meshed communication model to achieve the same end.

[0045] The information bus 108 that serves as the transportation medium for the presently preferred embodiment of the present invention can be Common Object Request Broker Architecture (CORBA)-based. The CORBA-based information bus is capable of handling the communication of events to and from objects in a distributed, multi-platform environment. The concept of a CORBA-based information bus is well known by those of ordinary skill in the art. Other acceptable communication languages can be used as are also known by those of ordinary skill in the art. Other transportation models are available as well, such as the Distributed Component Object Model (DCOM), which will readily achieve the same end.

[0046] CORBA provides a standard way of executing program modules in a distributed environment. A broker 110, therefore, may be incorporated into an Object Request Broker (ORB) within a CORBA compliant network. To make a request of an ORB, a client may use a dynamic invocation interface (which is a standard interface which is independent of the target object's interface) or an Object Management Group Interface Definition Language (OMG IDL) stub (the specific stub depending on the interface of the target object). For some functions, the client may also directly interact with the ORB. The object is then invoked. When an invocation occurs, the ORB core arranges so a call is made to the appropriate method of the implementation. A parameter to that method specifies the object being invoked, which the method can use to locate the data for the object. When the method is complete, it returns, causing output parameters or exception results to be transmitted back to the client.

[0047] In accordance with a presently preferred embodiment of the present invention an Enterprise Application Integration (EAI) system is used to broker the flow of information between the various services and adapters comprising the data network management system of the present invention. An example of an EAI system that can be incorporated in the presently preferred invention is the Active Works Integration System, available from Active Software of Santa Clara, Calif. As shown in FIG. 9, such an EAI system 112 uses an information broker 110 as the hub of the system. The information broker 110 acts as the central control and storage point for the system. The information broker 110 can reside on a server and serves to mediate requests to and from networked clients; automatically queuing, filtering and routing events while guaranteeing delivery. The information broker 110 is capable of storing subscription information and using such subscription information to determine where published information is to be sent. Referring back to FIG. 8, the information broker 110 is shown as being located at a point along the information bus 108. In most instances, the broker will be located within the same NOC 100 as the host that runs the access database adapter 106 application. Another key feature to the EAI system 112 of FIG. 9 is the use of service adapters 112 and 114 that allow users of the EAI system 112 to integrate diverse applications and other information when using the integration system. Service adapters 112 and 114 provide bidirectional mapping of information between an application's native format and integration system events, enabling all custom and packaged applications, databases, and Internet and other network applications to exchange information. As shown in FIG. 9 the adapters 112 and 114 run in association with the various services 116 and 118 from which information is published and subscribed on to an information bus 108 that has its hub at the broker 110.

[0048] Referring back to FIG. 8 the information bus 108 is in communication with a Point of Presence (POP) (POP₁) within the data communications network 50. POP₁is one of many PoPs with which the information bus 108 is in communication. Located within PoP₁is a host or node which may comprise one or more computing devices on which some or all of the services shown in FIG. 8 may be running. The node is in communication with the information bus 108 through a control adapter 116 which provides control communications with the various services such a PGW 118 through their respective service adapters 114, via service adapter 112 of control adapter 116. Those of ordinary skill in the art will appreciate that the services shown are not intended to be limiting and that other services and other service configurations can be used without departing from the inventive concepts herein disclosed. The system services may also be distributed over two or more servers to provide improved performance and redundancy.

[0049] The protocol gateway service 118 is used to couple the network user to the data communication network. The protocol gateway service 118 functions as an interface to the NASes that allows access requests received from a user to be serviced using components that may communicate using different protocols. A typical protocol gateway service 118 may be able to support different user access methodologies, such as dial-up, frame relay, leased lines, cable, DSL (Digital Subscriber Line), and the like.

[0050] Thus, in one embodiment of the invention, the broker-publisher system described above is used to publish VPN start events from the PGW 118 or other access port of the data communications network 50 to the central database at the PMC 102 (and SMC 104). Similarly, the local databases may subscribe to receive the central database information so that the decision to accept or reject may be made locally at the PoP, or they may simply query the PMC 102 to determine if an incremental user may raise a VPN. In an alternative embodiment, the central database may be omitted and each local database will then subscribe to each other local PoP's VPN start events and VPN stop events so as to locally maintain a current network-wide VPN session count as well as a local VPN session count.

[0051] Alternative Embodiments

[0052] While embodiments and applications of the invention have been shown and described, it would be apparent to those of ordinary skill in the art, after a perusal of the within disclosure, that many more modifications than mentioned above are possible without departing from the inventive concepts herein. The invention, therefore, is not to be restricted except in the spirit of the appended claims.