WO2001073573A1

WO2001073573A1 - Load balancing via message source selection

Info

Publication number: WO2001073573A1
Application number: PCT/US2000/007996
Authority: WO
Inventors: Maxon Duncan; Richard Mcewan
Original assignee: Mindarrow Systems, Inc.
Priority date: 2000-03-24
Filing date: 2000-03-24
Publication date: 2001-10-04
Also published as: AU2000240312A1

Abstract

Outgoing rich-media advertising messages in a client-server architecture are load balanced by having an initiator (10) direct multiple servers (42, 44, 52, 54, 56, 62, 64) to create and transmit the messages to recipients (80). The messages include message IDs (16, 66) that may be downloaded from the initiator (10), or generated in a LAN locally to the servers (60). The messages are preferably tracked at least in part using the message ID. Some of the servers may be coupled in local area networks (40, 50, 60), and distribution of recipient addresses may be based at least in part on the number of out-going mail threads being used in servers of the various LANs (40, 50, 60).

Description

LOAD BALANCING VIA MESSAGE SOURCE SELECTION

Field of The Invention

The field of the invention is electronic mail systems and methods.

Background of The Invention

In a typical electronic mail system, when a user sends a mail message to an external address, the following typically occurs: first, the user interacts with a user agent (such as "sendmail" on UNIX systems or Microsoft Outlook® on Microsoft Windows® systems) which accepts text and a request to send the text; second, upon receipt of the request to send, the text is converted into an electronic mail (e-mail) message and transferred to a local mail transfer agent (MTA) (such as "mailhost" on a UNIX systems or "Microsoft Exchange Server®" on Microsoft Windows® systems); finally, the local MTA, either directly or through a relay MTA, transfers the message to a receiving MTA ("mailhost" or "Microsoft Exchange Server®"). Details regarding mail systems as implemented on TCP/IP networks can be found in the series TCP/IP Illustrated published by Addison Wesley Longman, Inc., herein incorporated by reference in its entirety, and particularly Volume 1 of the series by W. Richard Stevens.

The step of transferring the message from the local MTA to the receiving MTA often involves choosing a route from the local MTA to the receiving MTA. Route selection is typically accomplished via low level protocols and "quasi-static" routing tables. Routing tables are "quasi-static" in the sense that they only change when the topography of the network changes, or when manually modified by an operator. Such tables are typically used to implement "cost based" routing in that each link in a table has an assigned cost and route selection is done so as to minimize the cost.

For a system which generates numerous large numbers of e-mail messages, the local

MTA (or the relay MTA if one is used) can become overwhelmed due to the volume of traffic to be handled. For very high volumes of traffic, there may be a significant delay between the time when a message is originally "sent" (given to the user agent) and when it actually gets delivered (transferred by the local or relay MTA to a receiving MTA). One method that might be applied to help prevent overwhelming an MTA is load balancing. Although load balancing mechanisms are known, they are generally limited to handling high volumes of incoming requests for data files such as requests for "web" pages. In such systems, an incoming request is initially directed to a "traffic cop" such as a load balancing router which then directs the request for data file and/or connection to an available computer or other device. Although such systems may be advantageous in the proper circumstances, there existence does not provide a mechanism for balancing the load caused by a high volume of outgoing mail messages.

A modified form of cost based routing replaces the cost figure with a desired relative frequency as is described in U.S.Patent No. 5,872,930 herein incorporated by reference in its entirety. Although this form achieves a type of load balancing, it is still based on quasi-static tables.

Quasi-static tables are not always sufficient because they do not take into account variations in network traffic or the capabilities of individual links. For example, a route over which a message is to travel based on the routing tables may be overburdened and acting as a bottleneck. Such bottlenecks can be particularly troubling when a mail message includes large amounts of data.

Numerous other electronic mail systems and architectures are already known. Suarez (US 5790789) teaches a distributed computing system in which agents control services by manipulating messages to the services. The architecture is particularly designed so that some servers can direct other servers to accomplish particular tasks. The Suarez system is, however, considerably different from the client-server and peer to peer architectures contemplated herein, and is distinguished from such systems throughout the specification.

Otorii (US 5632018) teaches an electronic mailing system in which recipients of a broadcast can readily send their responses not just to the originator of the broadcast, but to all recipients of the broadcast. The system is accomplished by passing along a list of all broadcast recipients to each of the recipients. Otorii's teachings are interesting, but do nothing to reduce bottlenecks by load balancing the outgoing mail. Miloslavsky (US 5765033) teaches routing of e-mails within a LAN according to the content of the e-mails, and availability of the recipient. Under that system, an e-mail requiring technical assistance would be routed to a technician rather than a salesman, and to a relatively available technician over a relatively busy technician. In other words, Miloslavsky is mostly interested in getting e-mails answered by the right person from a content standpoint, not in load balancing the outgoing e-mails.

Despite the existence of these numerous systems for outbound mail load balancing, however, there is still room for improving operating efficiency. This is especially true in the electronic direct mail industry, where many messages are being sent out that are both fairly large (500 kB to 1000 kB) and are substantially similar to one another.

Summary of the Invention

The present invention provides a method for load balancing outgoing advertising messages in a client-server architecture, in which an initiator creates a campaign core that includes a lists of potential recipients and a commercial message, each of a plurality of servers selects a different target from among the list of potential recipients, dynamically composes a rich media electronic message that includes the commercial message, adds a message ID, and causes the message to be sent to the target. The messages are preferably tracked at least in part using the message ID.

Various aspects of preferred embodiments involve combining some of the servers in one or more local area networks (LANs). Thus, it is contemplated that some of the servers can be combined into a first LAN that receives a download of at least a portion of the list of potential recipients, and other servers may be combined into a second LAN that receives another portion of the list of potential recipients. Where multiple LANs are employed, allocation of potential recipients can advantageously depend to at least some extent upon the number of out-going mail threads in use servers forming the LAN.

Other aspects of preferred embodiments involve the message ID. It is contemplated, for example, that at least one of the servers obtains the message ID directly from the initiator. It is also contemplated that message ID can be downloaded to a LAN, and then distributed to servers forming the LAN. Yet another contemplated option is for servers to create their own message IDs at the LAN. Regardless of where they are generated, the message IDs may or may not be unique.

Various objects, features, aspects and advantages of the present invention will become more apparent from the following detailed description of preferred embodiments of the invention, along with the accompanying drawings in which like numerals represent like components.

Brief Description of The Drawings

Figure 1 is a schematic view of a network embodying the claimed subject matter.

Figure 2 is a flowchart of a preferred method according to the present invention.

Detailed Description

In Figure 1 a load balanced mail network 1 generally includes an initiator 10, a potential recipients database 20, a commercial messages database 30, a connection 1 to first, second, and third LANs, 40, 50 and 60, respectively. Outgoing messages are created at servers 42, 44, 52, 54, 56, 62, and 64 and carried into the Internet 70 via connections 42A, 44 A, 52 A, 54A, 56 A, 62 A, and 64 A - the term "Internet", being used herein as a euphemism for any wide area network. Eventually, the messages are carried to recipients 80.

Initiator 10 obtains lists of potential recipients from the potential recipients database 20, and one or more commercial messages from the commercial messages database 30. Either the underlying information, or at least pointers to such information, are then transmitted to the first LAN 40, the second LAN 50, and the third LAN 60 across the connection 11, 11A, 1 IB, and 11C, respectively. First LAN 40 connects with servers 42, 44 through links 41A, 41B, respectively. Second LAN 50 connects with servers 52, 54, 56 through links 51A, 51B, 51C, respectively. Third LAN 60 connects with servers 62, 64 through links 61A, 61B, respectively.

Initiator 10 also produces or accesses a list of message identification codes (message

IDs) 16, and passes that subsets of that list to the first LAN 40 as shown by link 16A, and the second LAN 50 as shown by link 16B. These links 16 A, 16B may overlap with links 11, 11 A, 1 IB, and 1 lC. The lists are preferably, but not necessarily mutually exclusive. The message IDs received at first LAN 40 are distributed between mail server 42 and mail server 46. The distribution can be made on any logical or non-logical basis, but is preferably made as a function of the relative number of out-going mail threads in use by the servers 42, 44. Outgoing messages are generated at the servers 42, 44 from the subsets of potential recipients, commercial messages, and message IDs received by the first LAN 40. It will be appreciated that the initiator 10 may thus only partially and indirectly be controlling which servers 42, 44 send out which outgoing messages.

The operation of second LAN 50 is similar to that of first LAN 40. Message IDs received at second LAN 50 are distributed among mail servers 52, 54, and 56, with the distribution preferably being made as a function of the relative number of out-going mail threads in use by the servers 52, 54, 56. Outgoing messages are generated at the servers 52, 54, 56 from the subsets of potential recipients, commercial messages, and message IDs received by the second LAN 50. Here again the initiator 10 may only partially and indirectly controls which servers 52, 54, 56 send out which outgoing messages.

The operation of third LAN 60 is different in that the message IDs 66 are generated at the LAN 64, either by one or more of servers 62, 64, or by another machine (not shown). In any event, the locally generated message IDs 66 are distributed among mail servers 62, 64, along links 66A and 66B with the distribution again preferably being made as a function of the relative number of out-going mail threads in use by the servers 62, 64. Outgoing messages are generated at the servers 62, 64 from the subsets of potential recipients and commercial messages received by the third LAN 60, as well as from the locally generated message IDs 66. Thus, in this instance initiator 10 has even less control over the content of the outgoing messages.

The message IDs can be virtually type of coding. For example, message IDs can be a recipient's name, or an acronym generated from the name, such as RFISH002. Message IDs can also be quite obscure in derivation, such as xxlyz2. It is preferred that message IDs are unique, but they are not necessarily so.

Tracking is preferably accomplished by capturing information regarding the recipient's interaction with his or her computer during execution of the file or other message being sent. Information tracked may include file opening time, video start and stop times, cursor positioning, and forwarding of the commercial to others. Such information may advantageously be stored in the registry of the recipient's computer, and transmitted to a tracking server along with the message ID.

As used herein, the term "mail network" means the initiator, the servers, and the communication links among them. A load balanced mail network is a mail network in which at least some of the various components cooperate to balance the load on the network in an attempt to maximize throughput on at least one parameter. Any computer, routers, bridges, or other devices that are not part of the mail network are referred to as being "external" to the mail network. Thus, an "external computer" is a computer which is not an initiator, a server, or part of the communication linkway between the initiator and various servers. External devices may be interconnected so as to form one or more external networks such as the Internet. As an example, on an office LAN which has a single computer to connect the LAN to the Internet and to host an MTA for accepting e-mail messages from user agents on individual machines, the LAN is the mail network and the Internet is an external network comprising many external devices.

Servers 42, 44, 52, 54, 56, 62, 64 act as gateways to provide clean points of separation between the mail network and the external network, and may therefore be referred to as gateways. In most embodiments, the separation between the mail network and the external network is physical. Thus, in the office LAN example previously discussed, the server and any device or communication link on the LAN side of the server is part of the mail network and any device or communication link which is on the Internet side of the server is an external device. However, not all embodiments need to physically have separation between the mail network and the external network, such as where one or more communication links in the mail network are shared with the external network. In such embodiments, the separation is based more on separation of device functions and flow of data than on physical separation with any data/packets destined for the external network flowing through the server as though the mail network and external network were physically distinct.

It is contemplated that the servers 42, 44, 52, 54, 56, 62, 64 may each have numerous multimedia files stored on them for inclusion in outgoing mail messages. The stored messages may have merely been automatically cached from previous use, or may be intentionally stored for future use. In that manner it is not always necessary for the initiator 10 to send all of the commercial messages 30 to a server, but instead need only send a request specifying the recipient of the message and what to include in the message. The various servers 42, 44, 52, 54, 56, 62, 64 may also be able to access the potential recipients database 20 and the commercial messages database 30 without going through the initiator 10.

One advantage of having the initiator 10 send requests to "assemble and mail" rather than complete messages is that it helps prevent the network connecting the initiator 10 to the servers from carrying excess traffic. Thus, even though the system as a whole might require gigabytes per second of bandwidth, internal links do not correspond in size as the requests to "assemble and mail" take relatively little bandwidth.

Each of the LANs 40, 50, 60, and possibly even each of the servers, 42, 44, 52, 54, 56,

62, 64 may advantageously be connected to the Internet at different points. In such a case, any outgoing mail or mail requests could be directed through any of the servers. It is especially contemplated that a particular server could be chosen over another server to minimize the time needed to transfer a particular message to its recipient, or based upon some other parameter such as round trip time, server response time, and files locally available.

The step of having the initiator 10 choose the LANs and/or servers to use from among several possibilities can be viewed as providing a second layer of routing to the network. . Such application layer routing allows the initial choice of routes to be based on information generally not available to lower level protocols. The ability to perform application layer routing is at least partially supported by maintaining a plurality of servers having similar capabilities. It is contemplated that the concept of application can have many embodiments. The concept might best be described as, in a network comprising at least two layers or levels of protocols where one layer is a routing layer, using at least one layer other than the routing layer to route packets from one machine to another.

As used herein, "commercial messages" includes any material designed to cause someone to think or act in a particular way, such as to vote for a particularly political candidate, or purchase a particular product. Commercial messages may advantageously include pictures, graphics, sound, and video either individually or in combination, particularly electronically stored versions of such works. The term "multimedia file" as used herein includes computer files containing one or more such works. Related methods of routing, load balancing, and additional information on possible relationships between the initiator and servers can be found in U.S. Patent Applications titled "Message content based routing", no. 60/158,993; "Dynamic Routing via Shortest Delivery Time", no. 60/158,926; "Historical Delivery Time Based Routing Tables", no. 60/158,925; and "Load Balanced Mail Network Using Source Selection", no. 60/158,994, all filed October 12, 1999, and all of which are hereby incorporated by reference in their entireties.

In Figure 2, a preferred method 100 of load balancing outgoing advertising in a client-server architect comprises: providing a mail network having an initiator and a plurality of alternative servers distinct from the initiator 110; the initiator creating a campaign core that includes a list of potential recipients selected at least in part from a prospect database, and a commercial message selected at least in part from a commercial messages database 120; each of the plurality of alternative servers selecting a different target from among the list of potential recipients 130, dynamically composing a rich media electronic message that includes the commercial message 140, adding a message ID 150, and causing the message to be sent to the target 160; and tracking at least one of the messages at least in part using the message's message ID 170.

Thus, specific embodiments and applications of methods and devices for load balancing via server selection have been disclosed. It should be apparent, however, to those skilled in the art that many more modifications besides those already described are possible without departing from the inventive concepts herein. The inventive subject matter, therefore, is not to be restricted except in the spirit of the appended claims. Moreover, in interpreting both the specification and the claims, all terms should be interpreted in the broadest possible manner consistent with the context. In particular, the terms "comprises" and "comprising" should be interpreted as referring to elements, components, or steps in a non-exclusive manner, indicating that the referenced elements, components, or steps may be present, utilized, or combined with other elements, components, or steps that are not expressly referenced.

Claims

CLAIMSWhat is claimed is:

1. In a client-server architecture, a method for load balancing outgoing advertising comprising: providing a mail network having an initiator and a plurality of alternative servers distinct from the initiator; the initiator creating a campaign core that includes a list of potential recipients selected at least in part from a prospect database, and a commercial message selected at least in part from a commercial messages database; each of the plurality of alternative servers selecting a different target from among the list of potential recipients, dynamically composing a rich media electronic message that includes the commercial message, adding a message ID, and causing the message to be sent to the target; and tracking at least one of the messages at least in part using the message's message ID.

2. The method of claim 1 wherein the step of selecting a target comprises logically grouping at least two of the plurality of alternative servers in a LAN, and downloading at least a portion of the list of potential recipients to the LAN.

3. The method of claim 1 wherein the step of selecting a target comprises logically grouping at least four of the plurality of alternative servers in first and second distinct LANs, and downloading a first portion of the list of potential recipients to the first

LAN, and downloading a second portion of the list of potential recipients to the second LAN.

4. The method of claim 3 wherein the initiator allocates the first and second portions of the list of potential recipients at least in part according to a number of out-going mail threads in use by at least some of the servers the first LAN and at least some of the servers in the second LAN.

5. The method of claim 1 wherein the step of adding the message ID comprises at least one of the servers obtaining the message ID directly from the initiator.

6. The method of claim 1 wherein the step of adding the message ID comprises logically grouping at least two of the plurality of alternative servers in a LAN, and downloading a plurality of message IDs to the LAN, and at least one of the servers in the LAN selecting the message ID from the plurality of message IDs.

7. The method of claim 1 wherein the step of adding the message ID comprises logically grouping at least two of the plurality of alternative servers in a LAN, and creating the message IDs at the LAN.

8. The method of claim 1 wherein the message ID is unique.

AMENDED CLAIMS

[received by the International Bureau on 30 August 2000 (30.08.00) original claim 1 amended, remaining claim unchanged (1 page)]

1. In a client-server architecture, a method for load balancing outgoing advertising comprising: providing a mail network having an initiator and a plurality of alternative servers distinct from the initiator; the initiator creating a campaign core that includes a list of potential recipients selected at least in part from a prospect database, and a commercial message selected at least in part from a commercial messages database; each of the plurality of alternative servers selecting a different target from among the list of potential recipients, dynamically composing content of a rich media electronic message that includes the commercial message, adding a message ID, and causing the message to be sent to the target; and tracking at least one of the messages at least in part using the message's message ID.

3. The method of claim 1 wherein the step of selecting a target comprises logically grouping at least four of the plurality of alternative servers in first and second distinct LANs, and downloading a first portion of the list of potential recipients to the first LAN, and downloading a second portion of the list of potential recipients to the second LAN.

5. The method of claim 1 wherein the step of adding the message ID comprises at least one of the servers obtaining the message ID directly from the initiator