US 20070208877 A1
An embodiment of the invention comprises a method and apparatus for resolving textual references into numerical addresses. One exemplary method comprises obtaining a multi-part textual reference associated with a name resolution server. The multipart textual reference comprises at least a first host domain reference and a first resource location reference. The method further comprises parsing said multi-part textual reference to dynamically generate a multi-part canonical name. The name comprises a first part comprising a second resource location reference, and a second part comprising a second host domain name. The method further comprises resolving the canonical name into a numerical address.
1. A method comprising:
obtaining a multi-part textual reference associated with a name resolution server, wherein said multipart textual reference comprises at least a first host domain reference and a first resource location reference;
parsing said multi-part textual reference to dynamically generate a multi-part canonical name comprising a first part comprising a second resource location reference, and a second part comprising a second host domain name; and,
resolving said canonical name into a numerical address.
2. The method of
3. The method of
4. The method of
5. The method of
6. The method of
7. The method of
8. The method of
9. The method of
10. A method for registering a name resolution server reference comprising:
obtaining a textual reference comprising at least two parts, wherein said at least two parts comprises a first part indicating a network location, and a second part indicating a host domain name, said second part being a sub-domain part reference in said textual reference;
registering said textual reference with a central authority; and
registering a numerical address with said central authority in association with said textual reference.
11. The method of
12. The method of
13. A method comprising:
receiving a request for a resource located on a resource server, wherein said request comprises access to said resource on a domain name;
obtaining a dynamically generated multi-part canonical name from a name resolution server; and
obtaining an access location to said resource by parsing said dynamically generated multi-part canonical name.
14. The method of
15. The method of
16. The method of
17. The method of
18. The method of
19. The method of
20. The method of
This application is a continuation of application Ser. No. 10/210,205, filed Jul. 31, 2002, which is entirely incorporated herein by reference.
This invention relates to the field of computer software. More specifically, the invention relates to a method and apparatus for providing name services, allowing for enhanced manageability and configurability of name services on a network of computers.
In order for a computer to be part of a network, it must have a unique identity so that messages can be properly routed to the computer or network node. On the Internet network, for example, the address is represented in accordance with the Internet Protocol (IP) address. In the Internet Protocol each network node maybe represented with a numerical address. In version four of the Internet Protocol (Ipv4), the numerical address is a thirty two (32) bit address used in combination with a 32 bit mask number to define sub-networks accessible from and by the address in question. By convention, the 32 bit address and the mask is represented as “x.y.z.a”, where x, y, z and a may take any number between 0 and 255.
For humans it is much more convenient to utilize textual references to identify network locations than a numerical address. Therefore, to be able to utilize textual names alongside numbers, an infrastructure is required to provide name resolution whenever necessary. For example, a name resolution system has been implemented to provide name resolution in order to access the Internet. The system is commonly referred to as the Domain Name Service (DNS).
The Domain Name Service is a distributed database that includes the capability of translating Internet addresses to facilitate the location of computers connected to the Internet. A DNS name server translates a numerical address, assigned to a particular computer, into a sequence of words and vice versa. The word-based address is far more commonly used by people than the numerical address in order to make the address easier to remember and enter into a computer. The word-based address, for example, “president.whitehouse.gov”, is translated into a numerical address that consists of four sets of numbers separated by periods, for example, “126.96.36.199”, by the DNS system.
DNS names, by convention, are words written in lowercase letters with the words separated by periods, where the words refer to “domain” and “sub-domain” names. Each domain has a unique name and can be broken into a number of subdomains, with the responsibility for the subdomains residing the organization corresponding to the particular domain. The subdomain is the name of a computer host or an Internet provider and it may consist of several parts, or what amount to additional levels of subdomains. The DNS names take the form of “subdomain.domainname” (for example: “whitehouse.gov” or “cs.ucla.edu”), where the “domainname” represents a domain of the database, also referred to as a directory, partition or zone. Typically, the domain name indicates the type of organization. Common domain names include “corn” (for commercial organization), “net” (for networking organization), “edu” (for educational institutions) and “gov” (for the U.S. Federal government, the original sponsor of the Internet). Each domain and subdomain combination is referred to as a different zone. For example, in “cs.ucla.edu”, there is a edu zone, a ucla.edu zone and a cs.ucla.edu zone.
The DNS distributed database is indexed by the domain names and these provide paths in a large inverted tree structure, called the domain name space. At the top of the inverted tree, there is a single root with a limit of 127 levels from the root to the lowest level domain name.
The DNS system is really a DNS client-server system, where the name servers are computer programs that constitute the server side of the DNS client-server system. Information about some portions of the database is made available by the name servers to clients, called “resolvers” that often create queries and send them across a network to a name server. Programs running on a host computer, that require information from the domain name space, obtain the information from the resolvers, which access the name servers at different levels in the inverted tree. In addition to querying a name server, the resolver interprets the response and returns the information requested to the program that made the request. If the query is not answered, the resolver may resend the request and it may include a cache to store name related data retrieved previously from the name servers.
Root name servers are name servers at the root of a domain name, near the top of the inverted tree, and they are able to locate the “authoritative” name servers for the top level zones in the inverted tree. A name server is “authoritative” for servers under it on the tree because it knows the names of the servers at least one level under it on the tree. Thus, the top level root servers can provide the list of authoritative second level zone name servers. The second level zone name servers can provide the list of authoritative lower level name servers. Requests for addresses thus travel from level to level progressively through the tree structure until they are resolved and the necessary information is then returned to the requester along the same path.
A host computer (or a cluster of computers) has one or more unique IP addresses, and one or more domain names, that identify the computer to other computers and users on the network. At the end of name resolution by the name services server, a host name is resolved into a numerical address.
The actual Internet address is used in the form of a URL (Uniform Resource Locator). URLs may use the IP address in either the numerical format or the word based address discussed earlier. An example of this convention is “http://www.whitehouse.gov”. The “http” indicates the use of the hypertext transfer protocol for access to the particular location on the Internet. Name servers map the IP numbers to domain names (gov) and subdomain names (whitehouse) in the above URL on the world wide web (www) and guarantee that the correct IP number of the source and the destination are provided for all transmitted data packets.
The present DNS system for the Internet can become very complex for firms providing host computers, such as Internet service providers, because of the use of multiple servers that must be synchronized. Frequent address changes must be constantly updated in hosting firms' very large local DNS databases and the servers must be synchronized so that hosting firms' customers' web sites can be located. What is needed is a way to locate web and user domains on a web server and for emails without a local database. There is a need for a way to be a mass host of domain names for users of hosting services without having a large local database of domain information.
The invention is a method and apparatus for providing name services for servers and clients in a network of computing devices, without requiring the use of a database for mapping each name on the network to a corresponding numerical address. Embodiments of the invention utilize a multi-part registration name that is associated with the name resolution service server. When the name resolution service server receives a request to resolve a domain name, it fetches the reference of the name resolution service server associated with the domain name in question. The name resolution server parses the reference and dynamically generates a canonical name that may be analyzed and interpreted by other network computers to perform network tasks. The canonical name may be looked up on a name server to obtain a numerical address. Embodiments of the invention comprise application modules specialized in requesting the canonical names from name resolution servers, and mapping the information contained in the canonical name to resource properties.
In existing name resolution systems a root server provides a reference to a name resolution server. The name resolution server contains a database where each host and domain has an association with one or more numerical addresses. In embodiments of the invention, the name resolution server is enabled to dynamically generate a canonical name based on the information stored in the reference to the name resolution server. A second request to a name resolution server to resolve the canonical name returns a numerical address associated with the domain name addressed in the canonical name. The canonical name maybe resolved by the same name resolution server, or by a different name resolution server. Therefore, embodiments of the invention allow for resolving domain names without requiring a database for mapping between each registered domain name and a numerical address.
The invention is a method and apparatus for providing name services for servers and clients in a network of computing devices. In the following description, numerous specific details are set forth to provide a more thorough description of embodiments of the invention. It will be apparent, however, to one skilled in the art, that the invention may be practiced without these specific details. In other instances, well known features have not been described in detail so as not to obscure the invention.
Computers in a network (e.g., the Internet) communicate with one another using numerical addresses to identify the network location of a particular node. The addressing follows one or more protocols, such as Internet Protocol Version 4 (IPv4) and Internet Protocol Version 6 (IPv6). However, humans generally find textual addresses (e.g., names) more suitable than numerical addresses to address objects and locations. In order to allow humans to use textual references to network locations and computers to use numerical addresses, the network infrastructure provides services for mapping textual references into numerical addresses and provides the mapping to clients that automatically handle the name resolution process. For example, modern web browsers, such as Netscape Navigator and Microsoft Internet Explorer, allow users to access world wide web sites by entering a web site domain name (e.g., www.somedomain.ext). The browser then resolves the domain name to a numerical address utilizing the name resolution services supported by a network of computers.
Network location textual references may be represented in accordance to a set of rules that are part of a specific protocol. For example, dot-separated multi-string names are used in accordance to the Internet Protocol. Internet location names are represented at the right end of the name by a string that indicates a root name such as “com”, “org”, the two-character country codes representation etc. The next strings added to the left end represent a hierarchy of sub-domains. Usually, the string at the left most end represents a host name.
Throughout the disclosure, the term “server” is utilized to refer to hardware, including a number of physical machines virtually linked together to act as a single functional entity hosting one or more services. The term server farm is also used to refer to a cluster of machines including servers, routers firewalls and every equipment necessary to build a cluster of servers.
The term “server” is commonly loosely utilized in the art of computers to refer to the hardware (the physical machines) running one or more computer programs that serve data and to the computer programs serving data. In the following, the term “service” refers to the function (e.g., name resolution service) provided by a server and the term “server” will be used to refer to the machines providing the service. However, as it is common in the art of network services, the term “server” may be permutably utilized in the disclosure to refer to the service or to the machines providing the service when the distinction is not critical to the description of the subject matter. For example, a client making a connection to a name resolution server should mean that a client machine makes a network connection to a machine running a name resolution service; and in particular the network connection is made with the name resolution service.
Throughout the disclosure, examples of domain names will end with a “.ext” which refers to root domains such as “.com”, “.net”, “.org”, the two letter country codes and other existing and future root domains.
In embodiments of the invention, each step of the computation and data serving may involve the process of caching data or a result of a computation. Systems implementing the invention are also configurably enabled to serve data from an internal or an external cache, or generate a result de novo.
In a typical setting, such as the one illustrated in
When a user enters a reference in a network application to access a resource through the network, the first function the application performs is to retrieve the numerical address that corresponds to the textual reference. To allow users to access network resource using textual references, a server with the appropriate mapping is required in existing systems.
The infrastructure that supports name resolution involves a central authority, or root domain name server, 140 that is accessible to any client of server connected to the Internet. The central authority provides numerical addresses corresponding to domain names, and, most frequently, a reference to a name resolution server instead. To this end, each existing domain name intended to be accessed on the Internet has to be registered with a central authority. Furthermore, an address mapping has to be registered with the root servers, where the term address mapping refers to one or more textual references and the corresponding numerical addresses of one or more servers. A proximal name resolution server 180, whish is typically provided as a service along with the access to a network, obtains a reference to a remote (or distal) name resolution server 130 that is typically part of the resource hosting network. The proximal name resolution server submits a request to the remote name resolution server to resolve the domain name obtained from the user. The remote domain name resolution system, or domain name service server farm, 130 returns a numerical address to the resource server, or virtual domains server farm 120. The client 100 then accesses the resource server 120 using the numerical address to make one or more network connections and submits the URL to request the resource from the server. The resource server then returns the data.
When a user attempts to access a resource through the Internet (e.g., from client 100 to access a resource on server 120), the user typically enters a textual reference such a Uniform Resource Locator (URL) in a network application. Entering a textual reference may be performed by literally typing text, clicking on a hyper-link, issuing a voice command or performing any other action that results in submitting a textual reference to access a resource. The network application parses the URL to extract the part that refers to a domain name and attempts to resolve the textual reference (address) into a numerical address. The network application typically uses a resolver. The resolver is a service provided by the client's operating system or a helper application, running on the client machine, that mediates name resolution for applications running on the same client. The resolver requests a name resolution for the textual reference from a proximal domain name service host 180, also referred to as the user's ISP domain name server. The proximal domain name service host connects the central authority root domain name servers 140 to request a resolution of the textual reference. The central authority typically returns a reference to the domain name service host that stores and serves the information about the domain name in question. For example, a user tries to contact a resource on a remote server with the URL text: http://www.somedomain.ext/resource_name.html. The network application parses the URL text and determines that the resource is located on a server referenced by the domain name www.somedomain.ext, and the resource is accessible using the Hyper-Text Transfer Protocol (HTTP). The local resolver accesses a proximal name resolution system 180 that takes on the task of requesting a name resolution from a root server 140. The root server returns the address of a remote name resolution system 130 that hosts the mapping for the domain name. The proximal name resolution system 180 then issues a request to the remote name resolution system 130 to resolve the domain name into a numerical address. The proximal name resolution system then returns the result to the local resolver that in turn returns the results to the network application.
It is important to note that in existing name resolution systems, a database, configuration file or a simple algorithm provides the mapping between the domain name in the request and the resource on the resource server. For example, the request “http://www.joe.ext/resource_name.html” might be mapped by the server via a table lookup to the file resource_name.html in a storage location “/www/users/jsmith” on the resource server, while a request such as “http://www.bill.ext/resource_na- me.html” might be mapped by the server to a different file by the same name in the storage location “/www/users/bjones”. This approach is commonly referred to as “virtual hosting”. In embodiments of the invention, the request “http://jsmith.pages.hostingsite.ext/resource_file- .html” could be mapped algorithmically to “resource_name.html” In storage location “/www/users/jsmith”, and “http://bjones.pages.hostingsite.ext/re- source_file.html” could mapped to a file by the same name located in storage location “/www/users/bjones”. Existing implementations have used these or similar techniques based on resources available within the hosting network 110. In the following, a detailed description is provided for a system and method for performing resource mappings without relying on a local stored database, or simple substitution algorithm.
At step 220, the user obtains at least one domain name service reference. In embodiments of the invention, the user combines the name obtained from the reference to a storage location with the domain name of the ISP and the intended domain name to registered. For example, with a hosting service named “isp.ext” and a registered domain name “joe.ext” combined with the previously obtained name, a user may compose a textual reference to a first name server written as “isp.ext.jsmith.pages1.ns.joe- .ext” and to a second name server written as “isp.ext.jsmith.pages2.ns.joe- .ext”. At step 230, a user registers the domain name “joe.ext” using the domain name service reference with the central authority providing the name servers and, for each name server, obtains a numerical address of servers that provide name resolution services for the domain in question.
The proximal name resolution server uses the reference information obtained from the root server to query the host name resolution server at step 450. The host name resolution server may have the result already cached and retrieve it at step 455. If the result is found in the cache, the result is returned to the proximal name resolution server at step 460. If the result is not stored in the cache, the host name resolution server dynamically generates the response. To resolve a domain name, the host name resolution server queries the root server, at step 465, to obtain the reference to the host name resolution server (to itself). At step 470, the host name resolution server parses the reference and computes a canonical name, to which it is commonly referred as a name. Computing the name may involve any number of rules for parsing and interpreting information stored in the reference to the name resolution server. For example, “isp.ext.jsmith.pages1.ns.joe.ext” and “isp.ext.jsmith.pages2.ns.joe.ext” may both be associated with the domain to be resolved, and may be returned along with their respective network addresses by a root server. The reference may include portions that have an indicator function, rather than a part of the domain name or the name. In the latter example, “ns” may indicate a separator between the hosting network information and the specific domain name information. The host name resolution server is able to determine, based on the numerical address, which literal address was used to reach it and determines that the name should take into account directing traffic to a specific location on the network. In embodiments of the invention, name resolution servers are placed in separate networks. If one name resolution server is reached, it may signify that the predecessor could not handle a request properly, and logically directs succeeding requests to backup hosting servers. For example, pages1 and pages2 could indicate separate zones in a host network and may even indicate two completely separate networks. Thus, embodiments of the invention provide an inherent failover solution i.e. when one host system is not functioning properly, a second system is called into action automatically. Similar inherent properties may also support load balancing between separate network zones or between different networks.
Embodiments of the invention are enabled to query a name resolution server to resolve the domain name that is submitted with the query for the resource. For example, a computer program module may be implemented in any embodiment of the invention to perform the name resolution on behalf of any service network application. In one embodiment a module is incorporated with a hypertext transfer protocol based server to take in charge the name resolution of the queried domain name (e.g., www.joe.ext). At step 520, the resource server queries a name resolution server, or host DNS server, (e.g. a host name resolution server such as 130). Embodiments of the invention may follow every step, as previously described, to resolve a domain name, including querying the root name server and then the host name resolution server. However, embodiments of the invention may be configured to utilize name resolution services on their proximal name resolution servers. At step 540, the name resolution server returns the canonical name (as previously described) and optionally a numerical address of the resource server. The resource server parses the canonical name and extracts the information that indicates the storage location of the resource. For example, the canonical name may be, for the domain name “www.joe.ext”, “jsmith.pages1.isp.ext”. The resource server interprets the canonical name as pointing to a storage location (e.g., /www/users/jsmith). At step 550, the server fetches the resource from the storage location and serves it to the client.
In an embodiment of the invention, when a client 100 requires a name resolution, it makes a query 610 to a proximal name resolution server 180 that conducts the rest of the name resolution on behalf of the client. For example, a user on client 100 attempts to access a resource served on a server referred to as “www.joe.ext”. The proximal name resolution server may send a response 626 to the query providing the canonical name and/or the numerical address associated with the domain name if the response exists in the proximal name resolution server cache. In embodiments of the invention, the proximal name resolution server is enabled to iterate through the domain name, and attempts to resolve significant portions of the domain name. If the response is not available in the proximal name resolution server cache, the proximal name resolution server sends a query 612 to the root server 140. The root server 140 returns a response 613 to the proximal name resolution server. The response from the root server typically contains the numerical address and textual reference of the host name resolution server 130. In embodiments of the invention, the textual reference of the host name resolution server contains multiple parts. For example, a request to resolve the domain name “www.joe.ext” directed to the root server may cause the root server to return a host name resolution reference that contains at least “isp.ext.jsmith.pages1.ns.joe.ext”. The proximal name resolution server 180 utilizes the numerical address obtained from the root server to send a query 615 to the host DNS name resolution server 130. Name resolution protocols do not necessarily support sending the textual information about the name resolution server. The host name resolution server may send a query 617 to the root DNS name server 140 and request a resolution for the name that was submitted for resolution. The root name resolution server 140 returns a response 618 containing the host name resolution server's reference. The host DNS name resolution server 130 parses the textual reference and generates a canonical name corresponding to the domain name that was submitted. For example, if the root name server returns “isp.ext.jsmith.pages1.ns.joe.ext”, the host name resolution server 130 may generate a canonical name “jsmith.pages.isp.ext”. The canonical name is returned in a response 620 to the proximal name resolution server. The proximal name resolution server checks the response 621 to determine whether it contains a numerical address. If the response does not contain a numerical address, the proximal name resolution server submits another request 622 to the host name resolution server to resolve the canonical name into a numerical address 623. The host name resolution server parses the canonical name and determines the numerical address associated with the canonical name. In embodiments of the invention, canonical names may not all have a stored mapping. The host name resolution server examines the canonical name and may determine that the resource host belongs to a zone of the network that is associated with one or more numerical addresses. For example, a host name resolution receives a name “jsmith.pages.isp.ext” that does not have a stored explicit mapping. The host name resolution server is capable of parsing the previously generated canonical name and determining that the numerical address corresponds to the numerical address of the server named “pages.isp.ext”. In this event, it returns a response 624 to the proximal name resolution server 180 that, in turn, returns the resolution result 626 to the client.
In embodiments of the invention, the client submits a query 630 to access a resource stored on a resource server 120. Following the above example, the query requests access to the domain name “www.joe.ext/resourcename.html”. The resource server is enabled to retrieve information about the domain name from the name resolution servers 180. In the latter example, the resource server submits a request 632 to the host name resolution server 130 (e.g. host DNS server). The name resolution server checks 633 to determine whether the result exists in a cache. The host name resolution server may issue a query 634 to the root server 140. The root server then returns a reference 635 to the name resolution server associated with the domain name. The host name resolution server then dynamically generates 636 a canonical name and returns it 637 to the resource server 120. The resource server then parses the name to extract information about the resource location. For example, the resource server may determine, following the previous example, that the path to the resource on the storage device is “/www/users/jsmith”. The resource server retrieves the resource and serves it the client 638.
Thus, a method and apparatus has been disclosed for resolving network location names by dynamically generating a canonical name containing specific information about one or more resources on a resource server, without requiring the use a database of mappings between the generated names and numerical addresses.