WO2000013456A2 - Method and apparatus for load management on a computer network - Google Patents
Method and apparatus for load management on a computer network Download PDFInfo
- Publication number
- WO2000013456A2 WO2000013456A2 PCT/US1999/020056 US9920056W WO0013456A2 WO 2000013456 A2 WO2000013456 A2 WO 2000013456A2 US 9920056 W US9920056 W US 9920056W WO 0013456 A2 WO0013456 A2 WO 0013456A2
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- network
- rebroadcaster
- computer apparatus
- data
- content
- Prior art date
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/10—Protocols in which an application is distributed across nodes in the network
- H04L67/1001—Protocols in which an application is distributed across nodes in the network for accessing one among a plurality of replicated servers
- H04L67/1004—Server selection for load balancing
- H04L67/101—Server selection for load balancing based on network conditions
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/10—Protocols in which an application is distributed across nodes in the network
- H04L67/1001—Protocols in which an application is distributed across nodes in the network for accessing one among a plurality of replicated servers
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/10—Protocols in which an application is distributed across nodes in the network
- H04L67/1001—Protocols in which an application is distributed across nodes in the network for accessing one among a plurality of replicated servers
- H04L67/1004—Server selection for load balancing
- H04L67/1008—Server selection for load balancing based on parameters of servers, e.g. available memory or workload
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/10—Protocols in which an application is distributed across nodes in the network
- H04L67/1001—Protocols in which an application is distributed across nodes in the network for accessing one among a plurality of replicated servers
- H04L67/1004—Server selection for load balancing
- H04L67/1021—Server selection for load balancing based on client or server locations
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/10—Protocols in which an application is distributed across nodes in the network
- H04L67/1001—Protocols in which an application is distributed across nodes in the network for accessing one among a plurality of replicated servers
- H04L67/1004—Server selection for load balancing
- H04L67/1023—Server selection for load balancing based on a hash applied to IP addresses or costs
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/50—Network services
- H04L67/56—Provisioning of proxy services
- H04L67/568—Storing data temporarily at an intermediate stage, e.g. caching
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L69/00—Network arrangements, protocols or services independent of the application payload and not provided for in the other groups of this subclass
- H04L69/30—Definitions, standards or architectural aspects of layered protocol stacks
- H04L69/32—Architecture of open systems interconnection [OSI] 7-layer type protocol stacks, e.g. the interfaces between the data link level and the physical level
- H04L69/322—Intralayer communication protocols among peer entities or protocol data unit [PDU] definitions
- H04L69/329—Intralayer communication protocols among peer entities or protocol data unit [PDU] definitions in the application layer [OSI layer 7]
Definitions
- an audience member or an end user 2 who is a person or a computer system, typically uses a browser or a local Internet Service Provider (ISP) 4 to request and gain access to content from an Internet Content Provider or PubUsher (ICP) 6.
- ISP Internet Service Provider
- DNS local Domain Name System
- the Internet backbone in a particular geographic region includes a collection of ISPs, and Tier 1 ISPs 10 which are large ISPs with necessary resources to peer with a number of other large ISPs. Peering points are points of interconnection between two autonomous LP networks such as ISPs.
- a public Network Access Point (NAP) 12 is a place where ISPs of varying scale and scope interconnect.
- a US ISP aggregator 14 is a
- Performance is defined as the elapsed time between an audience member's request and the successful fulfillment of that request.
- the user's on-line experience is critical in ensuring that they stay on the site, return to the site, and complete transactions.
- the absolute speed of access is less important than the variability because web sites are designed with pages that download in a particular amount of time under normal conditions. It is therefore important to content providers that their site is consistently fast and reliable throughout their market.
- ICPs may duplicate or mirror their own Web sites at overseas locations, but this involves considerable effort.
- a mirror site is effectively a copy of the content and hardware (web servers, etc.) of the original site.
- Few Web hosting companies have a presence in every market. So in each target country, an ICP must find a suitable hosting firm or Internet Hosting Services (IHSs); negotiate a contract; build, install, and run the system; and pay a foreign currency invoice - all the while dealing with language and cultural issues.
- IHSs Internet Hosting Services
- the ICP faces the non-trivial challenge of keeping content replicated and synchronized, and centrally logging traffic.
- ICPs enter new markets by either ignoring the audience performance issues or by taking the extreme step of setting up a mirror site.
- site mirroring is a costly proposition that tends to improve performance in one network-local market.
- ISPs Internet Service Providers
- providers such as AT&T, Sprint, GTE, etc.
- IHSs Internet Hosting Services
- the ISP offerings fall short on market penetration due to their competitive stature with respect to the local service providers, local with respect to a particular market. As a perceived threat, the ISPs find it difficult to establish the requisite peering relationships in many markets.
- the IHS entries fall short due to prohibitive real estate and bandwidth costs as well as the operational complexities associated with maintaining facilities in a large number of localities.
- caching Another solution to this problem is caching, however while caching improves the user's experience it creates very serious problems for the content provider such as stale content and the elimination of information about the user.
- some product companies produce commercial caching devices that can be placed closer to the user and so improve their on-line experience. But they do not have the software needed to control the content stored (i.e. removing unwanted pages) or retrieving the log files (files that record user activity).
- Some software companies provide some of this additional functionality.
- a content provider could buy caches from a product company and software from a software company but would still need to obtain server space and bandwidth in all their target markets from either an ISP (Internet Service Provider) in each location or as a package from an International ISP. They would then need to arrange maintenance and support in each country plus the 24 hour a day / 7 day a week ability to monitor and co-ordinate all the servers and the in-country support structure.
- ISP Internet Service Provider
- the present invention relates to global Internet performance and the ability to provide users with consistently fast and reliable access to a content provider's data.
- a further object of the present invention is the ability to provide Internet user monitoring and auditing companies with data on Internet usage.
- Another objective of the present invention is the ability to provide Internet network provision companies with the data they need in order to plan new network installations.
- a global system combines local content caching and mirroring with a traffic management mechanism.
- the system places ICP content in close network proximity to their audience members and routes audience requests to the nearest content server, in terms of network topology.
- the traffic management system of the present invention in conjunction with the local content repositories is a solution to the issue of Internet audience performance.
- the traffic management system is suited to the high-latency, high packet-loss global Internet environment. Further, a geo-traffic manager is capable of scaling to the node count necessary to effectively reach the global Internet audience with acceptable performance. These two functions, local content caching and traffic management, necessarily form a tightly coupled system. One without the other is insufficient to fully address the performance issues of the global Internet. Additionally, the highly distributed nature of the system provides greatly enhanced service availability. In addition to the elimination of single points of failure, the distributed node in this system of the present invention continues to operate correctly in the absence of communications with the central site.
- the system includes a distributed node and central site services.
- the distributed node includes a geo-traffic manager, a rebroadcaster and an audience data collector.
- the distributed node is an extensible platform on which vertical applications can be built.
- the central site services include a network data collector, a customer care website, a billing administration workstation, a network monitor, a network operations console, a customer database, an audience routing database, a configuration version control system and a server content manager.
- the geo-traffic manager provides proximity routing of audience requests to rebroadcaster servers.
- the rebroadcasters provide the platform for local content distribution to global audience members. Additionally, the rebroadcaster supports content caching and mirroring based on configuration files constructed from data in the customer database.
- the rebroadcasters provide passive caching, pro-active caching and content mirroring.
- Fig. 1 is an overview of a computer network environment in which the present invention solves problems of the prior art.
- FIG. 2 is an overview of a computer network environment illustrating the system for load management in accordance with the present invention.
- Fig. 3 is a block diagram illustrating the system interaction in accordance with the system for load management of the present invention.
- Fig. 4 schematically illustrates domain name system traffic management used by the system in accordance with the present invention.
- Figs. 5A and 5B illustrate the geo-traffic manager configuration and interfaces included in the system of the present invention.
- Figs. 6 A and 6B illustrate the rebroadcaster configuration and interfaces included in the system of the present invention.
- Figs. 7A and 7B illustrate the audience data collector configuration and interfaces included in the system of the present invention.
- Fig. 8 is a block diagram of the routing table configuration in accordance with the present invention.
- Fig. 9 is a flow chart illustrating the rebroadcaster routing configuration in accordance with the present invention.
- Fig. 10 is a diagram illustrating an exemplary use of the system in accordance with the present invention where an audience member requests data that is located within the rebroadcaster's cache.
- Fig. 11 is a diagram illustrating another exemplary use of the system in accordance with the present invention where an audience member requests an uncacheable content element.
- Fig. 12 is a diagram illustrating another exemplary use of the system in accordance with the present invention where an audience member requests content not yet in cache.
- Fig. 13 is a diagram illustrating another exemplary use of the system in accordance with the present invention where an audience member requests mirrored content.
- the system in accordance with the present invention is an assemblage of two distinct service classes: distributed services 16 and central site services 18.
- a rebroadcaster, geo-traffic manager and audience data collector make up the distributed services 16 functions.
- a configuration control system 20, server content manager 22, network data collectors 24, network monitor 26, customer care web site 28, customer database 30, audience routing database 32, geographic data table 34, a billing administration workstation 36 and a Network Operations Console (NOC) 38 make up the central site services.
- NOC Network Operations Console
- distributed services nodes 40 are positioned at and serve as the Tier 1 ISP 10, Regional Network 42 and US ISP Aggregator 14.
- the Central Site Services 18 support the server functionality of the distributed services 16 resident in the distributed nodes 40.
- the system in accordance with the present invention allows the start, expansion and end of service to a particular country simply by the flip of a software switch.
- the system can be used to test market demand without the delay, cost, and expense of a mirror site of the prior art.
- the system is cost effective for low volumes of traffic as well as for very high volume but short duration events such as trade shows, major company announcements and international sporting competitions. It can also act as a backup.
- the system allows the content to be placed as close as possible to the end user while allowing the content provider to keep total control.
- the system is effectively a distributed Web server network. It enables an online content provider's web server to be connected via the Internet, to a series of rebroadcast servers located in key countries.
- Web pages consist of a series of elements, such as the menu bar, images, company logo, and text. Each element is transmitted from Web server to user's browser as a separate file using a protocol called Hypertext Transfer Protocol
- HTML HyperText Markup Language
- Dynamic content is normally the information that is provided uniquely for a user.
- the dynamic content files are very small, and normally represent less than 20% of the total amount of data needed to make up the page.
- Most dynamic content is derived from back end systems such as a stock inventory database or an airline reservation system. These are very difficult and expensive items to duplicate not only because of the hardware and software required, but also because of the difficulty of synchronizing the different versions of the same database.
- the system of the present invention positions the content closer to the end user by keeping the static content at the rebroadcast site and providing the dynamic content via an optimized Internet link.
- the optimized link is a way of transmitting data reliably and efficiently over the Internet even when the route from audience member or end user to ICP is convoluted and a significant proportion of the transmitted packets of data are lost.
- the typical Internet infrastructure consists of a series of communication links which are interconnected by routers. Data is sent as packets and each packet is forwarded from one router to the next until the packet reaches its destination or is lost.
- the router decides on the appropriate path by consulting a database called a routing table. This table essentially says "if you receive a packet destined for these end locations then send it to this particular router next".
- the problem with this form of routing is that it is essentially static and often setup to minimize Internet backbone operator's costs rather than optimizing performance.
- Applicants have discovered that by combining information from the Internet routing tables with information on the location of Internet congestion, it is possible to relay the customer's data around the congestion.
- the customer uses the private network included in the system of the present invention which in turn uses the Internet as its network.
- dedicated or reserved communications links are required to bridge the problem area.
- Fig. 2 illustrates the system for load management on a computer network in accordance with the present invention.
- An audience member or user 2 makes a request via a browser (local Internet Service Provider (ISP)) 4 to a local DNS server 8.
- the audience member 2 selects a universal resource locator (URL), and a DNS lookup is initiated with the local DNS server 8. If the local DNS server does not hold a valid record for the target server name (selected URL), the local DNS transmits a datagram request to all authoritative distributed nodes 40 for the name.
- the nearest distributed node 40 having a name server responds to the local DNS 8 first and then determines which rebroadcaster is available.
- the distributed node 40 returns to the local DNS service 8 a response in the form of an IP address identifying the rebroadcaster that can service the request.
- the local DNS (LDNS) caches the response according to the time-to-live (TTL) associated with the identified rebroadcaster name (IP address).
- TTL time-to-live
- IP address IP address
- the TTL field indicates how long the IP address should be cached by the local DNS before being discarded. TTL also gives the DNS server an indication of how recent and therefore how reliable the information they receive about other hosts is.
- the local DNS 8 ignores all subsequent responses from other servers.
- the local DNS forwards the received and cached rebroadcaster's IP address to the audience member 2.
- the audience member 2 then initiates communication directly with the rebroadcaster housed in a distributed node 40.
- Fig. 2 illustrates the scenario of the geo-traffic manager in the distributed node that returns a response in the form of an IP address is the same distributed node housing the available rebroadcaster, a response could have originated from another distributed node.
- a standard hypertext transfer protocol HTTP is used to make a request to the rebroadcaster. If necessary, the rebroadcaster contacts the ICP's origin server to retrieve un-cached content elements. It should be noted that the ICP's/customers cede responsibility for the name subdomain that is hosted on the rebroadcasters.
- the method of the present invention is advantageous to an end user as it allows the user to directly communicate with a rebroadcaster once the IP address of an available rebroadcaster is provided to the end user by a geo-traffic manager.
- the rebroadcaster then communicates with the ICP to provide the desired content to the end user.
- a portion of the content, such as the static content, is moved into the rebroadcaster which improves the speed and reliability for the user.
- the method of the present invention is advantageous to the ICP as the ICPs do not have to build expensive mirror sites.
- Fig. 3 illustrates the system interaction between the distributed services 16 or nodes 40 and the central site services 18.
- Each distributed node includes a respective rebroadcaster 50, geo-traffic manager 52 and audience data collector 54 which are each discussed later.
- the node 40 support all three functions, the system is capable of operating with the functions physically separated from one another. Any grouping of the functions within a distributed node is, in fact, supportable by the system due to the network transparent nature of the system.
- the functions of the Internet Content Publisher 6, Local DNS 8 and audience member 2 are outside the scope of control of the distributed nodes 40 of the present invention.
- the Configuration Version Control System 20 serves as the central point of control and audit for server configuration files throughout the system in accordance with the present invention.
- the Configuration Version Control System 20 facilitates full roll-forward and rollback functionality.
- Rollforward functionality is the configuration version control needed for releasing new versions of the configuration to production systems.
- Rollback functionality is the configuration control needed for referencing a previous version released to production systems.
- the version control system simplifies change tracking, auditability and rollback in the event of a misconfiguration.
- the configuration version control system receives data from the Customer Database 30 and the audience routing database 32 (discussed later).
- the Network Data Collectors 24 (Figs. 1 and 3) serve as an independent set of eyes into the global Internet. These collectors gather data on network connectivity, latency and routing, DNS response times, as well as HTTP performance.
- the Configuration Version Control system 20 oversees the configuration files for the collectors.
- the configurations are the result of queries against the Customer Database 30 indicating: tests to be performed, regions to be tested, networks to be tested, and URLs to be probed.
- the Customer Database 30 acts as the repository for the test results. Test results contribute to customer reporting, and geo-traffic manager routing tables.
- the Network Monitor 26 is the point of concentration for all system monitoring logging with the exception of the Network Data Collector 24.
- the operating system platform of the Network Data Collectors is incompatible with the server agents that forward data to the monitor. In another particular embodiment, the operating system platform can be compatible with the server agents that forward data to the monitor.
- the Network Operations Console 38 provides a high-level summary- view of the present invention system. It also gives the operators of the invention system an interface to control system components through the server agents and the Customer Database 30.
- the Network Operations Console 38 makes changes to the Customer Database 30 to reflect availability of the rebroadcaster 50. The changes are reflected in the geo-traffic manager routing tables 64.
- the Customer Care Web Site 28 allows the ICP 6 to alter a number of parameters that influence the provisioning of the present invention service.
- the customer may make changes in near real-time to a number of service parameters, including:
- the Customer Database 30 houses customer contact, account, service configuration, and usage data with respect to the end users 2 using the customer or ICPs 6 content. Usage by the end user 2 can be tracked by measures such as pages served from local cache and requests routed to the ICP's origin server. The value of a transaction between an ICP 6 and a user 2 can be monitored or calculated and the Customer/ICP can be charged accordingly.
- the Customer Database 30 is the source of the rebroadcaster configuration data as well as a source of data for the geo-traffic manager's content routing tables. This database contains information on customer URL caching and mirroring on a per rebroadcaster basis. It also acts as the repository for each geo-traffic manager 52, rebroadcaster 50, and Network Data Collector 24 logs. In a preferred embodiment, this log data arrives at the Customer Database 30 over a database link from a WebSpective Manager database serving as the Network Monitor 26.
- the geo-traffic manager 52, rebroadcaster 50 and audience data collector 54 that are built from the Customer Database 30 are checked into a Configuration Version Control system 20 prior to distribution to the target servers by WebSpective's binary coded decimal (BCD) service.
- BCD binary coded decimal
- the version control system simplifies change tracking, auditability and rollback in the event of a misconfiguration. Additionally, all changes to the Customer Database are logged in an audit log.
- the customer database server is based on the Sun PCI SPARC platform from the Concorde Group (2x366mHz, 1 GB RAM, 2x9GB internal and 6x18GB hot-swappable disks). It makes use of secured version of Solaris 2.6. Disk partition management is overseen by Sun Solstice Disk Suite (SDS). SDS provides striping and mirroring capabilities. WebSpective's v3.0 agent collects system statistics to the central WebSpective Manager database (Oracle 81). WebSpective's BCD service transports configuration files for the services on the rebroadcaster. Netscape FastTrack 3.0.1 acts as the web server for administrative access.
- Sun PCI SPARC platform from the Concorde Group (2x366mHz, 1 GB RAM, 2x9GB internal and 6x18GB hot-swappable disks). It makes use of secured version of Solaris 2.6. Disk partition management is overseen by Sun Solstice Disk Suite (SDS). SDS provides striping and mirroring capabilities. WebSpective's v
- the audience routing database 32 serves as the repository for all network topology data.
- a number of routing feeds contribute to the database including the Internet Routing Data Feed 60, the Geographic Data 34 and the Network Exception Data 62.
- Several classification parameters help in analyzing the relative weight of the particular datum in defining the topology portion of the geo-traffic manager routing table.
- the relative weight of a data point is set as a function of volatility and age or time. With respect to time weighting an exponentially decaying weight to data points is applied.
- volatility weighting can be achieved by at least two different approaches to obtain the proper smoothing. The first approach is time domain averaging which is used to avoid skewing the calculation. The n
- the second approach used is a frequency domain averaging in which an exponentially decaying weight is applied to data points below a specific threshold.
- the Server Content Manager 22 acts as the central control point for data distribution throughout the invention system with exception of the Network Data Collector 24.
- the operating system platform of the Network Data Collectors is incompatible with the server agents that accept data from the Server Content Manager 22.
- the Server Content Manager 22 reports on distribution status to the Network Monitor 26 and in turn to the Network Operations Console 38.
- the Billing Administration Workstation 36 does not play an active role in the load management, data distribution and trafficking operation of the present invention but instead provides business automation between the owner/operators of the system and the Customers/ICPs.
- the Billing Workstation takes summary usage data from the Customer Database 30 to generate summary and detailed statements.
- the Billing Workstation is based on NT and Visual Basic, and communicates with the Customer Database 30 through ODBC over Transmission Control Protocol/Internet Protocol (TCP/IP).
- TCP/IP Transmission Control Protocol/Internet Protocol
- Illustrated in Fig. 4 is the Domain Name System (DNS) based traffic management in accordance with the system of the present invention between an audience member 2 and a subject distributed node 40.
- DNS Domain Name System
- the geo-traffic manager 52 provides intelligent proximity routing of audience 2 requests to rebroadcasters 50.
- the geo-traffic manager 52 performs the audience routing function by answering domain name system (DNS) requests for name resolution from the audience member's local DNS server 8.
- DNS domain name system
- UDP User Datagram Protocol
- the IP address of the optimum rebroadcaster is sent to the audience member by the geo- traffic manager 52.
- the audience member 2 then communicates directly with the rebroadcaster 50 to access the desired content.
- the geo-traffic manager 52 includes the function blocks of content mirroring and configuration files 66; a geo- traffic manager server 68, an HTTP server 70, a server agent 72 for data distribution, collection and management, an operating system 74, Redundant Array of Independent Disks (RAID) 76 which is a method of storing same data in different places on multiple disks to improve input/output operating, and server hardware 78.
- the Server Content Manager 22 controls transportation of the configuration files and mirrored content to the Server Agents on the geo-traffic manager system.
- the geo- traffic manager server interfaces with the audience LDNS and the audience data collector 54.
- the public BIND server 77 interfaces with the audience/user LDNS.
- the public BIND server 79 accepts the LDNS request from the audience and performs a look-up for the ICP origin server having the target content and the available rebroadcaster.
- the public BIND server returns the IP address of the optimum rebroadcaster to the LDNS.
- the rebroadcaster Bind Server provides a centrally administered means of resolving origin server IP address for rebroadcaster's. This service is important for the retrieval of cache content by the rebroadcasters and for routing requests for uncacheable content.
- the DNS server uses BIND 8.2 with modifications to query a proprietary key-value data structure. This data structure incorporates a number of factors: • Internet Routing Data (in the form of Classless Internet Domain
- the out-of-band audience data (from the customer database and audience data collector).
- the out-of-band mechanism provides audience specific network data to fine tune the geo-traffic manager routing tables. This mechanism has several advantages over existing, polled-agent products as it provides fast local lookup, the local lookup provided is insensitive to link latency and loss, and the local lookup provided is highly scaleable.
- Data from the Network Data Collectors 24, the Audience Data Collector 54, the Network Monitor 26, and the Network Operations Console 38 reside in the Customer Database 30 with the customer and server configuration data.
- the distillation process combines the data from the Customer Database 30 with data from the Internet Routing Data Feed 60, Geographic Data 34, and a Network Exception Table 62.
- the distillation process checks the resultant table structures into a version control system prior to distribution to the distributed nodes 40 by the Server Content Manager 22.
- This mechanism is relatively insensitive to the LDNS honoring time-to-live (TTL) on the DNS lookups for proximity routing. Since subsequent audience requests, to the same LDNS, are highly likely to yield the same target rebroadcaster 50, local DNS caching of the response rarely leads to stale data caching. Local DNS caching becomes an issue when a rebroadcaster 50 becomes unavailable due to an equipment or network fault. Here a DNS that does not honor a small TTL will continue to send audience members 2 to a rebroadcaster 50 that has been marked as down within the geo-redirector's tables. As all modern versions of DNS servers support TTL functionality, only servers specifically configured to ignore the parameter will exhibit the aberrant behavior. It is possible to detect the signature of poorly configured LDNS server through geo-traffic manager log analysis.
- each distributed node acts as a DNS server.
- Using first response in this configuration provides first tier routing for the LDNS.
- the table lookups can place less weight on network proximity, relying on network latency during the look up to determine proximity.
- routing table proximity plays a more active role in determining the best rebroadcaster 50 to service the request.
- the relative weighting of the factors in a routing decision is fully configurable.
- the geo-traffic manager 52 is based on the Sun PCI SPARC platform from the Concorde Group (1x366MHz, 512MB RAM, 2x9GB internal disks). It makes use of secured version of Solaris 2.6. Disk partition management, striping, and mirroring is overseen by Sun SDS. WebSpective's v3.0 agent collects system statistics and DNS server logs to the central WebSpective
- WebSpective's BCD service transports configuration files for the DNS services on the geo-traffic manager.
- Netscape FastTrack 3.0.1. acts as the web server for administrative access.
- the system in accordance with the present invention uses modified bind answers DNS request from local DNS servers and from rebroadcasters.
- the geo-traffic manager 52 illustrated for carrying out proximity routing of audience requests to rebroadcasters 50 of the present invention is purely exemplary. Other traffic managers can be utilized in light of the teachings herein.
- Fig. 6A Illustrated in Fig. 6A is the configuration of the rebroadcaster 50.
- the rebroadcaster provides data collection, local load management, content distribution, content cache, content mirroring, and web servers.
- the rebroadcasters 50 are distributed throughout the network/Internet to provide localized content services globally.
- the rebroadcaster 50 provides the platform for local content distribution to global audience members 2.
- the rebroadcaster includes the function blocks of content mirroring and configuration files 80; an HTTP cache server 82, an HTTP server 84, a server agent 86 for data distribution, collection and management, an operating system 88, RAID 90 and server hardware 92.
- the rebroadcaster 50 supports content caching and mirroring based on configuration files constructed from data in the Customer Database 30.
- the Configuration Version Control server 20 acts as a point of control and audit for the rebroadcaster configuration files.
- the Server Content Manager 22 controls transportation of the configuration files and mirrored content to the Server Agents on the rebroadcaster system.
- a commercially available HTTP caching technology forms the basis for the caching services on the rebroadcaster 50.
- the Server Agent 86 collects HTTP Cache Server 82 log information and transports the data to the Network Monitor 26 database (and subsequently to the Customer Database 30).
- the agent 86 also offers an administrative interface for cache invalidation and pre- population.
- cache pre-population is a three-step process: 1. Spider the ICP's site to build a complete content tree listing.
- SPARC platform from the Concorde Group (2x366MHz,512MB RAM, 2x9GB internal and 3x18GB hot-swappable disks). It makes use of secured version of Solaris 2.6. Disk partition management is overseen by Sun SDS. SDS provides striping and mirroring capabilities. WebSpective's v3.0 agent collects system statistics to the central WebSpective Manager database (Oracle ⁇ l). WebSpective's binary coded decimal (BCD) service transports configuration files for the services on the rebroadcaster. Netscape FastTrack 3.0.1 acts as the web server for mirrored content and for administrative access. Inktomi's Traffic Server 3.0 supports content caching.
- BCD binary coded decimal
- caching services on the rebroadcaster 50 are based on Inktomi's Traffic Server 3.0.
- WebSpective's BCD supplies configuration files from the central CVS production configuration repository.
- WebSpective's agent collects Traffic Server log information and transports the data to the central database.
- WebSpective's agent also offers an administrative interface for cache invalidation and pre-population.
- the Inktomi traffic server 3.0 illustrated for carrying out the caching function of the present invention is purely exemplary. Other forward caching hardware or software can be utilized in light of the teachings herein.
- WebSpective's BCD transports client content to the re-broadcasters from a central BCD server located, for example, at HarvardNet. This content is stored locally in the UNIX file systems and is served by a Netscape FastTrack 3.0.1. webserver.
- the WebSpective server agent monitors and manages the Netscape Server. It also collects the web server logs to the central WebSpective Manager database.
- ICP content delivery to the server content manager is performed by a number of mechanisms: File Transfer Protocol (FTP) sent to an authenticated server, SCP and HTTP.
- FTP File Transfer Protocol
- the audience data collector 54 provides a data collection service along with the Network Data Collector 24.
- the audience data collector includes the function blocks of content mirroring and configuration files 100, an audience data collector server 102, an HTTP server 104, a server agent 106 for data distribution, collection and management, an operating system 108, RAID 110 and server hardware 112.
- the Server Content Manager 22 controls transportation of the configuration files and mirrored content to the Server Agents 106 on the audience data collector system.
- the audience data collectors 54 collect data on network connectivity to actual CustomerTCP networks and ultimately Customer/ICP IP addresses, if desired.
- the audience data collector 54 can operate as a stand- alone service on independent hardware as an alternative to the illustrated embodiment with the audience data collector integrated in the distributed node 40.
- Th audience data collectors receive configuration data both from the Customer Database 30 and directly from the geo-traffic manager 52. The latter feeds near real-time routing and response time data for actual audience members 2.
- Data collected by the audience data collectors 54 feed into the central Customer Database 30 and is used for both routing table calculations and reporting services.
- the audience data collector 54 is managed by the server agent 106 and provides a web administrative interface for network operations.
- Fig. 8 illustrates the routing table configuration in accordance with the system of the present invention.
- the calculation of the geo-traffic manager routing tables 64 aggregates a number of data sources into a format searched efficiently by the BIND process.
- Data from the Network Operations Console 38, the audience data collector 54, the Network Monitor 26, and the Network Data Collectors 24 resides in the Customer Database 30.
- the Router Table Distillation Process 120 combines the data from the Customer Database 30 with data from the Audience Routing Database 32 which includes data from the Internet Routing Data Feed 60, Geographic Data 34 and the Network Exception Data 62. This data then forms an input into the geo-traffic manager routing tables 64 (shown also in Fig. 3).
- the resultant table structures form an input into the Configuration Version Control System 20 prior to distribution to the geo-traffic managers 52 in the distributed nodes 40 by the Server Content Manager 22.
- Factors contributing to the geo-traffic manager routing tables include:
- the router table distillation process condenses this data into a key-value structure that is suitable for use by the BIND function call.
- Fig. 9 illustrates the rebroadcaster configuration in accordance with the system of the present invention.
- the calculation of the rebroadcaster configuration is simpler than the calculation of the geo-traffic manager routing tables 64.
- Rebroadcaster configurations 63 are the formatted results of queries against the Customer Database 30. Data relevant to this process arrives primarily from the
- Fig. 10 illustrates a preferred embodiment showing a characteristic interaction between audience members 2 and the system of the present invention as well as interactions between the system of the present invention and the ICPs 6 web site when required.
- the figure illustrates a use case where an audience member 2 requests data that is located within the rebroadcaster's 50 cache.
- the interaction is similar to that between an audience member and a local web server.
- the audience member 2 makes a DNS request for a target content using a local DNS 8 which is forwarded to the geo-traffic manager 52.
- the geo-traffic manager 52 after performing look-up for the ICP 6 having the target content and an optimal rebroadcaster 50, returns a DNS response to the LDNS in the form of an LP address of the available rebroadcaster which is forwarded to the audience member 2.
- the audience member directly communicates via HTTP with the available rebroadcaster identified.
- the rebroadcaster returns the target content located within the rebroadcaster cache to satisfy the audience member request.
- the audience member 2 requests an un-cacheable content element by making a DNS request similar to the process described in Fig. 10.
- Personalized pages and stock quotes are examples of un-cacheable content.
- the rebroadcaster 50 transparently (to the audience member) forwards the request to the ICP/origin server 6, accepts the response, and forwards the requested data to the audience member 2.
- the system then includes the loading of local display elements that make up the remainder of the page.
- the audience member requests a previously un-cached content element by making a DNS request similar to the process described in Fig. 10.
- the rebroadcaster 50 once again, transparently forwards the request to the ICP/origin server 6, accepts the response, and forwards the requested data to the audience member 2.
- Fig. 13 illustrates an exemplary case where the content requested by the audience member 2 by making a DNS request similar to the process described in Fig. 10, is served from a local site mirror on the rebroadcaster 50.
- the dotted arrow 130 indicates the batch, bulk, content replication mechanism used to populate the mirror.
- the audience member 2 receives an error condition. With properly constructed content, this should never be the case since the mirror should behave identically to the ICP/origin server 6.
- a particular embodiment supports mixing of mirrored, cached, and uncacheable content through a technique of URL rewrite. While this invention has been particularly shown and described with references to preferred embodiments thereof, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the scope of the invention encompassed by the appended claims.
Abstract
Description
Claims
Priority Applications (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CA002342186A CA2342186A1 (en) | 1998-08-31 | 1999-08-31 | Method and apparatus for load management on a computer network |
EP99945394A EP1110363A2 (en) | 1998-08-31 | 1999-08-31 | Method and apparatus for load management on a computer network |
AU57997/99A AU5799799A (en) | 1998-08-31 | 1999-08-31 | Method and apparatus for load management on a computer network |
JP2000568286A JP2002524945A (en) | 1998-08-31 | 1999-08-31 | Method and apparatus for load management in computer networks |
Applications Claiming Priority (6)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US9848898P | 1998-08-31 | 1998-08-31 | |
US60/098,488 | 1998-08-31 | ||
US11887499P | 1999-02-05 | 1999-02-05 | |
US60/118,874 | 1999-02-05 | ||
US11978999P | 1999-02-11 | 1999-02-11 | |
US60/119,789 | 1999-02-11 |
Publications (2)
Publication Number | Publication Date |
---|---|
WO2000013456A2 true WO2000013456A2 (en) | 2000-03-09 |
WO2000013456A3 WO2000013456A3 (en) | 2000-07-06 |
Family
ID=27378611
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/US1999/020056 WO2000013456A2 (en) | 1998-08-31 | 1999-08-31 | Method and apparatus for load management on a computer network |
Country Status (4)
Country | Link |
---|---|
EP (1) | EP1110363A2 (en) |
JP (1) | JP2002524945A (en) |
CA (1) | CA2342186A1 (en) |
WO (1) | WO2000013456A2 (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2001093534A2 (en) * | 2000-06-01 | 2001-12-06 | Aerocast.Com, Inc. | Selective routing |
US6668275B1 (en) * | 1999-12-17 | 2003-12-23 | Honeywell International Inc. | System and method for multiprocessor management |
US6836806B1 (en) | 2000-06-01 | 2004-12-28 | Aerocast, Inc. | System for network addressing |
US6904460B1 (en) | 2000-06-01 | 2005-06-07 | Aerocast.Com, Inc. | Reverse content harvester |
US7213062B1 (en) | 2000-06-01 | 2007-05-01 | General Instrument Corporation | Self-publishing network directory |
US7747772B2 (en) | 2000-06-01 | 2010-06-29 | Aerocast.Com, Inc. | Viewer object proxy |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US9473743B2 (en) | 2007-12-11 | 2016-10-18 | Thomson Licensing | Device and method for optimizing access to contents by users |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO1998018076A1 (en) * | 1996-10-18 | 1998-04-30 | Intervu, Inc. | System and method for optimized storage and retrieval of data on a distributed computer network |
US5764906A (en) * | 1995-11-07 | 1998-06-09 | Netword Llc | Universal electronic resource denotation, request and delivery system |
-
1999
- 1999-08-31 CA CA002342186A patent/CA2342186A1/en not_active Abandoned
- 1999-08-31 JP JP2000568286A patent/JP2002524945A/en active Pending
- 1999-08-31 WO PCT/US1999/020056 patent/WO2000013456A2/en not_active Application Discontinuation
- 1999-08-31 EP EP99945394A patent/EP1110363A2/en not_active Withdrawn
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5764906A (en) * | 1995-11-07 | 1998-06-09 | Netword Llc | Universal electronic resource denotation, request and delivery system |
WO1998018076A1 (en) * | 1996-10-18 | 1998-04-30 | Intervu, Inc. | System and method for optimized storage and retrieval of data on a distributed computer network |
Non-Patent Citations (1)
Title |
---|
FEI Z -M ET AL: "A novel server selection technique for improving the response time of a replicated service" PROCEEDINGS IEEE INFOCOM. THE CONFERENCE ON COMPUTER COMMUNICATIONS,US,NEW YORK, NY: IEEE, page 783-791-791vol2 XP002109463 ISBN: 0-7803-4384-0 * |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6668275B1 (en) * | 1999-12-17 | 2003-12-23 | Honeywell International Inc. | System and method for multiprocessor management |
WO2001093534A2 (en) * | 2000-06-01 | 2001-12-06 | Aerocast.Com, Inc. | Selective routing |
WO2001093534A3 (en) * | 2000-06-01 | 2002-08-29 | Aerocast Com Inc | Selective routing |
US6658000B1 (en) | 2000-06-01 | 2003-12-02 | Aerocast.Com, Inc. | Selective routing |
US6836806B1 (en) | 2000-06-01 | 2004-12-28 | Aerocast, Inc. | System for network addressing |
US6904460B1 (en) | 2000-06-01 | 2005-06-07 | Aerocast.Com, Inc. | Reverse content harvester |
US7213062B1 (en) | 2000-06-01 | 2007-05-01 | General Instrument Corporation | Self-publishing network directory |
US7747772B2 (en) | 2000-06-01 | 2010-06-29 | Aerocast.Com, Inc. | Viewer object proxy |
Also Published As
Publication number | Publication date |
---|---|
EP1110363A2 (en) | 2001-06-27 |
WO2000013456A3 (en) | 2000-07-06 |
JP2002524945A (en) | 2002-08-06 |
CA2342186A1 (en) | 2000-03-09 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Dilley et al. | Globally distributed content delivery | |
US9300560B2 (en) | Network performance monitoring in a content delivery system | |
US7086061B1 (en) | Statistical tracking of global server load balancing for selecting the best network address from ordered list of network addresses based on a set of performance metrics | |
US7284055B1 (en) | Method and system for network redirecting | |
US8291046B2 (en) | Shared content delivery infrastructure with rendezvous based on load balancing and network conditions | |
US8725861B2 (en) | Content delivery network service provider (CDNSP)-managed content delivery network (CDN) for network service provider (NSP) | |
US20030149581A1 (en) | Method and system for providing intelligent network content delivery | |
JP5264961B2 (en) | Global document hosting system using embedded content distribution ghost server | |
US7949779B2 (en) | Controlling subscriber information rates in a content delivery network | |
US6484143B1 (en) | User device and system for traffic management and content distribution over a world wide area network | |
US7676576B1 (en) | Method and system to clear counters used for statistical tracking for global server load balancing | |
US9825903B2 (en) | Provisioning tool for a content delivery network (CDN) | |
WO2001065402A2 (en) | Method and system for providing intelligent network content delivery | |
US20020143798A1 (en) | Highly available distributed storage system for internet content with storage site redirection | |
WO2001014990A1 (en) | Method for content delivery over the internet | |
CN102439913A (en) | System and method for network traffic management and load balancing | |
Gayek et al. | A web content serving utility | |
EP1110363A2 (en) | Method and apparatus for load management on a computer network | |
AU5799799A (en) | Method and apparatus for load management on a computer network |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
AK | Designated states |
Kind code of ref document: A2 Designated state(s): AE AL AM AT AU AZ BA BB BG BR BY CA CH CN CR CU CZ DE DK DM EE ES FI GB GD GE GH GM HR HU ID IL IN IS JP KE KG KP KR KZ LC LK LR LS LT LU LV MD MG MK MN MW MX NO NZ PL PT RO RU SD SE SG SI SK SL TJ TM TR TT UA UG US UZ VN YU ZA ZW |
|
AL | Designated countries for regional patents |
Kind code of ref document: A2 Designated state(s): GH GM KE LS MW SD SL SZ UG ZW AM AZ BY KG KZ MD RU TJ TM AT BE CH CY DE DK ES FI FR GB GR IE IT LU MC NL PT SE BF BJ CF CG CI CM GA GN GW ML MR NE SN TD TG |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application | ||
DFPE | Request for preliminary examination filed prior to expiration of 19th month from priority date (pct application filed before 20040101) | ||
AK | Designated states |
Kind code of ref document: A3 Designated state(s): AE AL AM AT AU AZ BA BB BG BR BY CA CH CN CR CU CZ DE DK DM EE ES FI GB GD GE GH GM HR HU ID IL IN IS JP KE KG KP KR KZ LC LK LR LS LT LU LV MD MG MK MN MW MX NO NZ PL PT RO RU SD SE SG SI SK SL TJ TM TR TT UA UG US UZ VN YU ZA ZW |
|
AL | Designated countries for regional patents |
Kind code of ref document: A3 Designated state(s): GH GM KE LS MW SD SL SZ UG ZW AM AZ BY KG KZ MD RU TJ TM AT BE CH CY DE DK ES FI FR GB GR IE IT LU MC NL PT SE BF BJ CF CG CI CM GA GN GW ML MR NE SN TD TG |
|
WWE | Wipo information: entry into national phase |
Ref document number: 57997/99 Country of ref document: AU |
|
ENP | Entry into the national phase |
Ref document number: 2342186 Country of ref document: CA Ref country code: CA Ref document number: 2342186 Kind code of ref document: A Format of ref document f/p: F |
|
ENP | Entry into the national phase |
Ref country code: JP Ref document number: 2000 568286 Kind code of ref document: A Format of ref document f/p: F |
|
WWE | Wipo information: entry into national phase |
Ref document number: 1999945394 Country of ref document: EP |
|
WWP | Wipo information: published in national office |
Ref document number: 1999945394 Country of ref document: EP |
|
REG | Reference to national code |
Ref country code: DE Ref legal event code: 8642 |
|
WWW | Wipo information: withdrawn in national office |
Ref document number: 1999945394 Country of ref document: EP |