Request management for hierarchical cache

Abstract

A computer implemented cache management system and method is provided for use with a service provider configured to communicate with one or more client devices and with a content provider. The system includes a cache hierarchy comprising multiple cache levels that maintain at least some resources for the content provider, and one or more request managers for processing client requests for resources and retrieving the resources from the cache hierarchy. In response to a resource request, the request manager selects a cache level from the cache hierarchy based on a popularity associated with the requested resource, and attempts to retrieve the resource from the selected cache level while bypassing cache level(s) inferior to the selected level.

Description

BACKGROUND

Generally described, computing devices and communication networks can be utilized to exchange information. In a common application, a device can request content from another device via the communication network. For example, a user at a personal computing device can utilize a software browser application to request a Web page from a server device via the Internet. For another example, a user of a content providing service can utilize a display device (e.g., television, computer, mobile device, smartphone, tablet) to request content (e.g., an audio presentation, video presentation, audiovisual presentation, movie, television show, video, podcast) to be streamed from a server device via the Internet to the display device. In such embodiments, the user computing device or display device can be referred to as a client device and the server device can be referred to as a content provider.

Content providers are generally motivated to provide requested content to client devices often with consideration of efficient transmission of the requested content to the client device and/or consideration of a cost associated with the transmission of the content. For larger scale implementations, a content provider may receive content requests from a high volume of client devices which can place a strain on the content provider's computing resources. Additionally, the content requested by the client devices may have a number of components, which can further place additional strain on the content provider's computing resources.

With reference to an illustrative example, content to be streamed to the display device to be presented to the user may be identified by one or more embedded resource identifiers, such as uniform resource locators (“URLs”). In turn, software on the client devices typically processes embedded resource identifiers to generate requests for the content. Often, the resource identifiers reference a computing device associated with the content provider such that the client device would transmit the request for the content to the referenced content provider computing device. Accordingly, in order to satisfy a content request, the content provider would provide client devices data associated with the content.

Some content providers attempt to facilitate the delivery of requested content through the utilization of a content delivery network (“CDN”) service provider. A CDN server provider typically maintains a number of computing devices in a communication network that can maintain content from various content providers. In turn, content providers can instruct, or otherwise suggest to, client devices to request some, or all, of the content provider's content from the CDN service provider's computing devices. The CDN service provider may store at least some portion of the content for different content providers in one or more hierarchical cache structures and respond to client content requests by retrieving data from the hierarchical cache structure.

BRIEF DESCRIPTION OF THE DRAWINGS

The foregoing aspects and many of the attendant advantages of this invention will become more readily appreciated as the same become better understood by reference to the following detailed description, when taken in conjunction with the accompanying drawings, wherein:

FIG. 1 is a block diagram illustrative of an example content delivery environment compatible with certain embodiments described herein, the example content delivery environment including a number of client devices, a content provider, and a content delivery network service provider.

FIG. 2 schematically illustrates an example cache server hierarchy of a resource cache component compatible with certain embodiments described herein.

FIG. 3 is a flow diagram of an example computer implemented resource request processing method for retrieving resource from a cache server hierarchy in response to a resource request based at least partly on a measure of resource popularity.

FIG. 4 is a flow diagram of an example computer implemented resource popularity updating method for updating resource popularity and its relationship with a cache server hierarchy for resource request management.

DETAILED DESCRIPTION

Generally described, the present disclosure is directed to managing components of a content delivery network (“CDN”) configured to respond to resource requests from a client device. Specifically, one aspect of the disclosure will be described with respect to managing resource requests received at point of presence (“POP”) locations of a CDN and routing the requests to resource cache component comprising a plurality of cache servers organized in a cache server hierarchy based on resource popularity measures. Another aspect of the disclosure will be described with respect to determining and updating resource popularity measures by different request managers within POPs to serve as a basis for the managing and routing of resource requests.

In accordance with an illustrative embodiment, a CDN POP includes one or more request managers that process resource requests from client devices. Each request manager may maintain a resource tracking index that tracks the respective popularity of different resources that are requested, based on the frequency, recency, CPU processing time, network bandwidth, or other attributes of resource requests received at the request manager. The request manager may map different value ranges of resource popularity to different cache levels in a cache hierarchy associated with the CDN POP.

Upon receipt of a resource request, the request manager may look up the resource tracking index for popularity information associated with the requested resource. The request manager then attempts to retrieve the requested resource from a cache level based on the popularity of the requested resource, in accordance with the mapping between resource popularity value ranges and cache levels. By doing so, lower levels of the cache hierarchy may be skipped or bypassed in the processing of certain requests for resources that are less popular.

Thus, cache misses at certain lower cache levels because of requests for unpopular resources may be avoided. As a person of skill in the art would understand, cache misses at the lowest or lower cache levels that are triggered by requests for unpopular resources may cause insertion of the unpopular resources into the lowest or lower cache levels, thus taking up valuable cache storage space in the lowest or other lower cache levels. The insertion of unpopular resources may cause premature flush or purge of certain cached items that are likely more popular (e.g., more likely to be requested) from the lowest or lower level caches, and consequently decrease cache hit rate, increase cache flush/purge rate or overhead, or otherwise negatively impact cache performance. By attempting to retrieve the requested resource from a cache level based on the popularity of the requested resource, unpopular resources are less likely to be inserted into lower level caches, and the storage space of lower level caches (which are likely limited and expensive) may be utilized more efficiently.

In another aspect, request managers can be classified or clustered into different groups based on geographic proximity, network topology, or pattern of resource requests that have been received. Request managers within a same group may share resource popularity information with one another so that any request manager within the group may update its locally maintained resource tracking index based on newly observed resource request patterns by other request managers within the group. Request managers may further adjust the mapping between resource popularity value ranges and cache levels based on updated resource popularity in order to utilize the cache hierarchy efficiently.

Although various aspects of the disclosure will be described with regard to illustrative examples and embodiments, one skilled in the art will appreciate that the disclosed embodiments and examples should not be construed as limiting.

FIG. 1 is a block diagram illustrative of an example content delivery environment 100 for the management and processing of content requests, the example content delivery environment compatible with certain embodiments described herein. As illustrated in FIG. 1, the content delivery environment 100 includes a number of client devices 102 (generally referred to as clients) for requesting content from a plurality of providers including a content provider 104 and a CDN service provider 106. In an illustrative embodiment, the client devices 102 can correspond to a wide variety of devices including personal computing devices, laptop computing devices, hand-held computing devices, terminal computing devices, mobile devices (e.g., smartphones, tablets), wireless devices, various electronic devices (e.g., televisions) and appliances, and the like. In an illustrative embodiment, the client devices 102 include necessary hardware and software components for establishing communications over a communication network 108, such as a wide area network or local area network. For example, the client devices 102 may be equipped with networking equipment and browser software applications that facilitate communications via the Internet or an intranet.

As also illustrated in FIG. 1, the content delivery environment 100 can also include the content provider 104 configured to communicate with the one or more client devices 102 via the communication network 108. The content provider 104 illustrated in FIG. 1 comprises a logical association of one or more computing devices associated with the content provider. By way of example, the content provider 104 can include a web server component 110 corresponding to one or more server computing devices for obtaining and processing requests for content from the client devices 102. The content provider 104 can further include an origin server component 112 and associated storage component 114 corresponding to one or more computing devices for obtaining and processing requests for network resources from the CDN service provider 106.

One skilled in the relevant art will appreciate that the content provider 104 can be associated with various additional computing resources, such additional computing devices for administration of content and resources, DNS name servers, and the like. For example, although not illustrated in FIG. 1, the content provider 104 can be associated with one or more DNS name server components that would be authoritative to resolve client device DNS queries corresponding to a domain of the content provider 104 (e.g., return an IP address responsive to the DNS query). A DNS name server component is considered to be authoritative to a DNS query if the DNS name server can completely resolve the query by providing a responsive IP address. Additionally, the content provider 104 may omit some of the components illustrated in FIG. 1, such as the origin server 112 or provide content to client computing devices 102 via components other than the web server 110.

With continued reference to FIG. 1, the content delivery environment 100 can further include a CDN service provider 106 configured to communicate with the one or more client devices 102 and with the content provider 104 via the communication network 108. The CDN service provider 106 illustrated in FIG. 1 comprises a logical association of one or more computing devices associated with the CDN service provider 106. Specifically, the CDN service provider 106 can comprise a number of Point of Presence (“POP”) locations 116, 122, 128, which also can be generally referred to as Point of Delivery locations or “PODs,” that correspond to nodes on the communication network 108. Each CDN POP 116, 122, 128 also comprises at least one resource cache component 120, 126, 132 having a plurality of cache server computer devices organized in a cache server hierarchy.

In an illustrative embodiment, although the CDN POPs 116, 122, 128 are illustrated in FIG. 1 as logically associated with the CDN service provider 106, the CDN POPs can be geographically distributed throughout the communication network 108 in a manner to best serve various demographics of client devices 102. Additionally, one skilled in the relevant art will appreciate that the CDN service provider 106 can be associated with various additional computing resources, such as additional computing devices for administration of content and resources, and the like. Still further, although the resource cache components 120, 126, 132 are discussed as being associated with the CDN service provider 106, one skilled in the relevant art will appreciate that the resource cache components 120, 126, 132 may be provided by different service providers, independently or on behalf of content provider 104 or CDN server provider 106. In some embodiments, the resource cache components 120, 126, 132 may include hardware or software components for tracking the expiration (e.g., time-to-live values) of resources stored on behalf of one or more various content providers 104.

FIG. 2 schematically illustrates an example cache server hierarchy of a resource cache component 120, compatible with certain embodiments described herein. The resource cache component 120 can be configured to process requests for resources and transmit the requested resource to the client device 102. The resource cache component 120 is also configured to communicate with one or more request managers 210 of the CDN service provider 106 and comprises a plurality of cache servers 220 configured to store resources from content providers 104 and transmit various requested resources to various client devices 102.

The request manager 210 may be implemented within a corresponding CDN POP 116, 122, or 128, and can include a plurality of routers (e.g., border routers) that can be configured to provide connectivity to or communicate with external devices (e.g., client devices 102), other request managers 210, and the cache servers 220 of one or more of the resource cache components of a same or different POP(s). The request manager 210 may receive requests for resource (e.g., from one or more client devices 102), determine popularity associated with the requested resource, map or route the resource request to a cache level, retrieve or attempt to retrieve the requested resource from the cache level, or implement other functionalities disclosed herein. The cache servers 220 can comprise endpoints (e.g., hosts, network interface controllers or NICs) that are operationally coupled to the communication network 108.

In certain embodiments, the cache servers 220 of the resource cache component 120 may be logically organized in a cache server hierarchy. For example, in the example cache server hierarchy of FIG. 2, the resource cache component 120 comprises a plurality of cache servers 220 (which for clarity, will be referred to as cache servers 220′, 220″, 220′″, etc.) that are logically organized in levels. The cache servers 220 located closer to client devices 102 (e.g., edge cache server 220′) populate the lower levels of the cache server hierarchy and the cache servers 220 located further from the client devices 102 (e.g., cache servers 220″, 220′″) populate the higher levels of the cache server hierarchy.

Cache servers 220 residing on the same logical level (e.g., the four cache servers 220′, the three cache servers 220″, or the two cache servers 220′″) can be referred to as peers of one another. Cache servers 220 residing at a higher logical level than another cache server can be referred to as hierarchically superior or parent cache servers of the other cache server (e.g., cache servers 220″ are hierarchically superior or parent cache servers of each of cache servers 220′, and cache servers 220′″ are hierarchically superior or parent cache servers of each of cache servers 220″ and each of cache servers 220′). Cache servers 220 residing at a lower logical level than another cache server can be referred to as hierarchically inferior or child cache servers of the other cache server (e.g., cache servers 220″ are hierarchically inferior or child cache servers of each of cache servers 220′″, and cache servers 220′ are hierarchically inferior or child cache servers of each of cache servers 220′″ and each of cache servers 220″). The origin server 112 of the content provider 104 can be at the top of the hierarchy, as any resource to be maintained by one or more cache servers 220 of the resource cache component hierarchy may ultimately be retrieved from the origin server 112, in the event that the resource is not maintained by a peer or parent cache server. The concepts of peer cache servers, hierarchically superior cache servers, and hierarchically inferior cache servers can apply to cache servers 220 across multiple resource cache components 120, 126, 132 and across multiple CDN POPs 116, 122, 128. For example, a cache server 220″ of one resource cache component 120 of one CDN POP 116 can be considered to be a peer of a corresponding cache server 220″ of another resource cache component 126 of CDN POP 122, can be considered to be hierarchically superior to a corresponding cache server 220′ of another resource cache component 132 of CDN POP 128, and can be considered to be hierarchically inferior to a corresponding cache server 220′″ of another resource cache component 132 of CDN POP 128.

In some embodiments, the latency or communication cost for request managers 210 to access cache servers 220 increases with the level of cache server in the hierarchy. For example, a lowest level cache server 220′ may be implemented with one or more high-performance processors and memory on a same host computing device that implements a request manager 210 and thus has a negligible latency; a medium level cache server 220″ may be communicating with the request manager 210 via some dedicated high-speed communication channel (e.g., optical fiber connection within a data center) and thus has a relatively low latency; and a higher level cache server 220′″ may be implemented on one or more slower computing systems and require multiple hops via the Internet for the request manager 210 to access and thus has a relatively high latency for data access. In some embodiments, the average data storage cost, such as technical standard or qualification of storage repository, decreases with the level of cache servers in the hierarchy and the average size of cache increases with the level of cache servers. For example, a lowest level cache server 220′ may implement its cache on some expensive fast-access memory having limited storage space; a medium level cache server 220″ may implement its cache on one or more hard drives; and a higher level cache server 220′″ may implement its cache on certain low-cost mass storage device.

In some embodiments, a cache miss at one level cache server 220′ of a resource cache component of a CDN POP can result in a request to another cache server that is hierarchically superior to the cache server 220′ (e.g., a hierarchically superior cache server 220″ at the same resource cache component of the same CDN POP, a hierarchically superior cache server 220″ at a different resource cache component of a different CDN POP, or the origin server 112). This cache miss at the cache server 220′, which is on the edge and thus also referred to as an edge cache server 220′, may, for example, result from an unavailability or nonexistence of a cached copy of the resource at the edge cache server 220′. Accordingly, if the edge cache server 220′ receives a request for the resource, the edge cache server 220′ may have to request the resource from a parent cache server (e.g., one of the cache servers 220″). Should the requested resource not be available from the parent cache server 220″, further requests for the resource will be transmitted up the cache server hierarchy to other parent cache servers (e.g., one of the cache servers 220′″), or even to the origin server 112, until a computing device that maintains the requested resource in storage is found. Upon identification of the cache server 220 maintaining the requested resource, the requested resource can be propagated to hierarchically inferior or child cache servers (including the edge cache server 220′), the requested resource may also be returned to a client device 102 that requested the resource.

In certain embodiments, a request manager 210 tracks the popularity of different resources based on the recency, frequency, CPU processing time, network bandwidth, or other attributes associated with the requests for resources that are received at the request manager 210. For example, the request manager 210 may use a resource tracking index to track each resource that has been requested during a recent period of time (e.g., the most recent week). The resource tracking index may be implemented as a table, list, queue, heap, or other applicable data structure that is maintained by a computer memory or other data storage device. Each entry of the tracking index may include any one or more of the following: an identifier of a resource that has been recently requested, frequency measure(s) of requests for the resource during one or more predefined time periods, weights or time-decay factors that can be assigned to requests received at different time periods, one or more resource popularity measures for the resource based on predefined algorithm(s) or formula(e) applicable to various attributes associated with the requests for the resource, expiration time of the entry (e.g., time-to-live), and the like. The tracking index may be updated periodically (e.g., every hour, every day, etc.) and locally by each request manager 210. The tracking index and resource popularities may also be updated when triggered by certain events (e.g., after receipt of a predefined number of resource requests, upon detection of heavy incoming resource request traffic, etc.). As will be described in more detail with respect to FIG. 4, the tracking index and resource popularities may be updated based on resource popularity information that is shared among certain groupings of request managers 210. As will also be discussed in more detail with respect to FIG. 3, the request manager 210 may manage, route, or otherwise process resource requests with respect to different levels of cache servers based on the popularity of the requested resource.

FIG. 3 is a flow diagram of an example computer implemented resource request processing method for retrieving a resource from a cache server hierarchy in response to a resource request based at least partly on a measure of resource popularity. The method of FIG. 3 may be implemented by a request manager 210 of one of the CDN POPs 116, 122, or 128.

At block 310, the request manager 210 receives a request for a resource from a client device 102. In accordance with an illustrative embodiment, the client device 102 may generate an initial content request corresponding to a resource identifier (e.g., in accordance with common network protocols). The request may then be received and processed by the content provider 104 to identify (e.g., in accordance with DNS routing principles) the appropriate responsive content and the CDN POP 116, 122, or 128 associated with the appropriate CDN service provider 106 from which the responsive content may be obtained. For example, the CDN service provider 106 can provide the content provider 104 with an identification of applicable domains of the CDN service provider 106 (unless it has been previously provided) and any additional information. In turn, the content provider 104 can then process the information received from the CDN service provider 106 and can transmit appropriate information, including embedded resource identifiers, to the client device 102 that made the content request. Upon receipt of the information from the content provider 104, the client device 102 can transmit a resource request to a request manager 210 of a selected CDN POP 116, 122, or 128, in order to acquire the resources identified by the embedded resource identifiers (e.g., using its local DNS resolver).

At block 320, the request manager 210 determines the popularity of the requested resource. Illustratively, the request manager 210 may look up the requested resource in a locally maintained resource tracking index. If the resource tracking index includes the requested resource, the request manager 210 may retrieve one or more popularity measures associated with the requested resource. In some embodiments, the request manager 210 may update the resource tracking index based on attributes associated with the currently received resource request either prior to or after the index lookup. For example, based on the time of receipt of the resource request, the interval between the receipt of the current and prior requests for the resource, or other characteristics of the current or prior requests, the request manager 210 may recalculate one or more popularity measures for the resource requested. Consequently, corresponding existing values in the resource tracking index may be replaced by the newly recalculated values.

At block 330, the request manager 210 selects an appropriate cache level from the cache server hierarchy of the resource cache component 120, 126, or 132, for retrieving the requested resource. The request manager 210 may or may not select the lowest level cache server 220′ (e.g., a cache level having the lowest communication latency as compared to other cache levels) and may make the selection based on the one or more popularity measures associated with the requested resource. In some embodiments, the request manager 210 may maintain or reference a mapping or other relationship between different value ranges of resource popularity and cache levels of the cache server hierarchy. The request manager 210 may determine that the popularity measure of the requested resource falls into certain value range (x, y), which is mapped to certain cache level θ. Therefore, the request manager 210 may select cache level θ for retrieval of the requested resource. Illustratively, resource popularity values may be divided into several predefined or dynamically generated value ranges (e.g., brackets,) each mapped deterministically or probabilistically to a corresponding cache level of the cache server hierarchy. In some embodiments, higher popularity values may be mapped to lower cache levels, which may have limited and expensive data storage space but may facilitate faster access to resources that are more likely to be requested. On the other hand, lower popularity values may be mapped to higher cache levels, which may have larger and cheaper data storage space but may be slower to access.

At block 340, the request manager 210 attempts to retrieve the requested resource from the selected cache level. In some embodiments, the request manager 210 may route the resource request to a cache server on the selected cache level. In other embodiments, the request manager 210 may generate a resource retrieval request to access the cache server on the selected cache level based on the resource request. For example, the popularity associated with the requested resource be relatively low (e.g., the resource is not frequently requested) and the request manager 210 selects a high level cache server 220′″ for retrieving the requested resource. In this case, the request manager 210 routes the resource request or otherwise communicates with the high level cache server 220′″ for retrieval of the resource, without communicating with any cache servers (e.g., cache servers 220″ or 220′) inferior to the high level cache server 220′″. Accordingly, any cache server inferior to the high level cache server 220′″ will not incur a cache miss due to the current resource request, and the requested resource will not be propagated to the inferior cache server.

Upon receiving the resource retrieval request, at block 350, the high level cache server 220′″ can determine whether the requested resource is available (e.g., cached locally at the high level cache server 220′″). If the requested resource is available at the selected cache level, the method of FIG. 3 may proceed to block 370.

If the high level cache server 220′″ does not have a copy of the requested resource, the method of FIG. 3 may proceed to block 360 where the high level cache server 220′″ can attempt to obtain the requested resource from another source (e.g., by sending a resource retrieval request to one or more hierarchically superior cache servers of the same resource cache component, to one or more hierarchically superior cache servers of a different resource cache component, or if the resource is also absent from these hierarchically superior cache servers, the origin server 112). In certain embodiments, the resource retrieval requests can be sent from a cache server on a first level to one or more cache servers that are on a second level that is hierarchically superior, but adjacent to the first level. In certain embodiments, the resource retrieval requests can be sent from a cache server on a first level to one or more cache servers on a second level that is not hierarchically adjacent to the first level (which can be considered as jumping hierarchical levels, e.g., from a mid-level cache server to the origin server 112). In some configurations, each cache server 220 of a resource cache component is constrained to only send resource retrieval requests to hierarchically superior cache servers of the same resource cache component. However, as described above, in certain embodiments, each cache server of a resource cache component is not constrained to only send resource retrieval requests to hierarchically superior cache servers of the same resource cache component, but can also send resource retrieval requests to hierarchically superior cache servers of other resource cache components of the CDN service provider 106.

At block 370, the resource is returned to the client device 102 that requested the resource. In some embodiments, the request manager 210 obtains the requested resource from the cache server 220′″ on the selected cache level and forwards the resource to the client device 102. In other embodiments, the responding cache server 220′″ on the selected cache level transmits the requested resource to the client device 102. In still other embodiments, one of the cache servers superior to the selected cache level that had received a resource retrieval request transmits the requested resource to the client device 102.

By implementing the method of FIG. 3, lower levels of the cache hierarchy may be skipped or bypassed in the processing of certain requests for resources that are less popular. Accordingly, cache misses at certain lower cache levels because of requests for unpopular resources may be avoided, and valuable cache storage space in the lowest or other lower cache levels can be preserved for more popular resources (e.g., more likely to be requested). Additionally, the method of FIG. 3 may reduce the likelihood of premature flush or purge of certain popular resources from the lowest or lower level caches, and consequently increase cache hit rate, reduce cache processing overhead, or otherwise improve cache performance.

FIG. 4 is a flow diagram of an example computer implemented resource popularity updating method for updating resource popularity and its relationship with a cache server hierarchy for resource request management. The method of FIG. 4 may be implemented by a request manager 210 of one of the CDN POPs 116, 122, or 128.

For example, it might be difficult for a request manager 210 to determine whether a particular level two cache server 220″ has a cached copy of certain requested resource, since content provider sites or CND POPs may be serving thousands of resource requests each and there may be hundreds of cache servers associated with a site or POP. Nonetheless, a lossy approach to approximate this information can be employed. For instance, updates to resource popularity can be achieved by a broadcast network, whereby a customer facing request manager 210 who just got notified of a cache hit for a target resource from the level two cache server 220″ may broadcast to all the other customer facing POPs or associated request managers 210 this information. If the target resource has been requested multiple times across multiple POPs or request managers recently, similar information may be broadcasted to and received by other request managers. In turn, the other request managers may elevate the popularity measure(s) of the target resource in their respective (e.g., local) resource tracking indices, so that requests for the target resource will be mapped or routed to a level one cache server(s) 220′.

At block 410, the request manager 210 receives resource popularity information from one or more external sources. Illustratively, request managers 210 of a CDN POP may be classified or clustered into different groups, for example, based on geographic proximity, network topology, or pattern of resource requests received, and resource popularity information may be shared among request managers 210 of a same group. For example, once a first request manager within a group has received a threshold number of resource requests or completed an update of its locally maintained resource tracking index, the first request manager may broadcast or otherwise send any updated resource popularity information to other request managers of the same group. In some embodiments, each group of request managers 210 may maintain a shared resource tracking information repository, and each request manager 210 of the group may access the shared resource tracking information repository to enter or retrieve updated resource popularity information.

At block 420, the request manager 210 updates resource popularity in its locally maintained resource tracking index based on the information received from external source(s), such as another request manager 210 of the same POP, a request manager 210 of a different POP, a cache server, an information broadcasting or transmittal service associated with the CDN service provider 106 or the content service provider 104, or the like. In some embodiments, the request manager 210 updates certain entries in its locally maintained resource tracking index with the resource popularity information received. For example, the locally maintained resource tracking index may include a relatively low popularity value for a target resource due to an insufficient frequency or recency of locally received requests for the target resource at the request manager 210. However, resource popularity information received from the external source may include an entry for the same target resource that indicates a substantially higher popularity value (e.g., exceeding a predefined threshold value by comparison), because other request manager(s) of the same group has recently received a sufficient number of requests for the target resource within a short or predetermined, or threshold period of time. In this case, the request manager 210 may adjust an entry of the target resource in its locally maintained resource tracking index to reflect the high popularity value.

In some embodiments, the popularity values or other information received from the external source may be discounted or weighted before combining with the information of the locally maintained resource tracking index. In some embodiments, the information received from the external source includes raw data on counts or timestamps associated with resource requests received at other request managers, and the request manager 210 receiving the information may calculate new popularity values based on the raw data as well as locally maintained logs of resource requests.

At block 430, the request manager 210 optionally adjusts the mapping or other relationship between resource popularity and cache levels in the cache server hierarchy. Illustratively, the request manager 210 may analyze and evaluate the distribution of updated resource popularities for resources that reflects a more recent resource request pattern, and adjust the value ranges that can be mapped to different cache levels. The value ranges may be shifted, merged, split, narrowed, or expanded in a way to properly accommodate the updated distribution of resource popularities. For example, if the updated distribution of resource popularities reflect a significant decrease in popularity values (e.g., lower than a threshold when compared to prior popularity values) for a majority of resources, then most of the prior value ranges will need to be shifted lower. By adjusting the value ranges or their mapping to cache levels, the request manager 210 may avoid mapping too many or too few resources to certain cache levels. In some embodiments, the request manager 210 may adjust the value ranges so that the number of resources mapped to different cache levels would conform with a defined distribution among the cache levels (e.g., a distribution based on the cache size at different levels). This way, the cache hierarchy may be fully and efficiently utilized in accordance with the more recent resource request pattern.

At block 440, the request manager 210 processes subsequent resource requests based on the updated resource popularity. As discussed above, the request manger 210 may look up its locally maintained resource tracking index for any subsequently requested resource, determine one or more popularity values associated with the subsequently requested resource, and identify a cache level of the cache hierarchy for retrieval of the subsequently requested resource.

Depending on the embodiment, certain acts, events, or functions of any of the methods described herein can be performed in a different sequence, can be added, merged, or left out altogether (e.g., not all described acts or events are necessary for the practice of the algorithm). Moreover, in certain embodiments, acts or events can be performed concurrently, e.g., through multi-threaded processing, interrupt processing, or multiple processors or processor cores or on other parallel architectures, rather than sequentially.

The various illustrative logical blocks, modules and method elements described in connection with the embodiments disclosed herein can be implemented as electronic hardware (e.g., application-specific logic circuitry), computer software executed by hardware, or combinations of both. To clearly illustrate this interchangeability of hardware and software, various illustrative components, blocks, modules, and steps have been described above generally in terms of their functionality. Whether such functionality is implemented as hardware, or as software executed by hardware, depends upon the particular application and design constraints imposed on the overall system. The described functionality can be implemented in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the disclosure.

The various illustrative logical blocks and modules described in connection with the embodiments disclosed herein can be implemented or performed by a computing device, such as a general purpose processor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein. A general purpose processor can be a microprocessor, but in the alternative, the processor can be a controller, microcontroller, or state machine, combinations of the same, or the like. A processor can also be implemented as a combination of computing devices, e.g., a combination of a DSP and a microprocessor, a plurality of microprocessors, one or more microprocessors in conjunction with a DSP core, or any other such configuration.

The elements of a method, process, or algorithm described in connection with the embodiments disclosed herein can be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. A software module can reside in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disk, a removable disk, a CD-ROM or any other form of computer-readable storage medium known in the art. A storage medium can be coupled to the processor such that the processor can read information from, and write information to, the storage medium. In the alternative, the storage medium can be integral to the processor. The processor and the storage medium can reside in an ASIC. The ASIC can reside in a user terminal. In the alternative, the processor and the storage medium can reside as discrete components in a user terminal.

Conditional language used herein, such as, among others, “can,” “might,” “may,” “e.g.,” and the like, unless specifically stated otherwise, or otherwise understood within the context as used, is generally intended to convey that certain embodiments include, while other embodiments do not include, certain features, elements, and/or states. Thus, such conditional language is not generally intended to imply that features, elements and/or states are in any way required for one or more embodiments or that one or more embodiments necessarily include logic for deciding, with or without author input or prompting, whether these features, elements and/or states are included or are to be performed in any particular embodiment. The terms “comprising,” “including,” “having,” “involving,” and the like are synonymous and are used inclusively, in an open-ended fashion, and do not exclude additional elements, features, acts, operations and so forth. Also, the term “or” is used in its inclusive sense (and not in its exclusive sense) so that when used, for example, to connect a list of elements, the term “or” means one, some, or all of the elements in the list.

Disjunctive language such as the phrase “at least one of X, Y, or Z,” unless specifically stated otherwise, is otherwise understood with the context as used in general to present that an item, term, etc., may be either X, Y, or Z, or any combination thereof (e.g., X, Y, and/or Z). Thus, such disjunctive language is not generally intended to, and should not, imply that certain embodiments require at least one of X, at least one of Y or at least one of Z to each be present.

Unless otherwise explicitly stated, articles such as “a” or “an” should generally be interpreted to include one or more described items. Accordingly, phrases such as “a device configured to” are intended to include one or more recited devices. Such one or more recited devices can also be collectively configured to carry out the stated recitations. For example, “a processor configured to carry out recitations A, B, and C” can include a first processor configured to carry out recitation A working in conjunction with a second processor configured to carry out recitations B and C.

While the above detailed description has shown, described, and pointed out novel features as applied to various embodiments, it will be understood that various omissions, substitutions, and changes in the form and details of the devices or algorithms illustrated can be made without departing from the spirit of the disclosure. As will be recognized, certain embodiments described herein can be embodied within a form that does not provide all of the features and benefits set forth herein, as some features can be used or practiced separately from others. All changes which come within the meaning and range of equivalency of the claims are to be embraced within their scope.

Claims

1. A computer implemented cache management system configured to be used in conjunction with one or more client devices and with a content provider, the cache management system comprising: a hierarchical cache structure comprising a plurality of cache levels, wherein a highest cache level is provided by the content provider; anda request management unit comprising a physical processor, the request management unit configured to: receive, from a client device, a request for a target resource associated with the content provider;determine a popularity measure associated with the target resource;select a cache level from the plurality of cache levels based at least partly on the popularity measure associated with the target resource, wherein selecting the cache level includes bypassing at least one cache level in the plurality of cache levels; andretrieve the target resource from the selected cache level.

2. The computer implemented cache management system of claim 1, wherein each cache level of the plurality of cache levels comprises one or more cache servers.

3. The computer implemented cache management system of claim 1, wherein the highest cache level comprises one or more one or more origin servers of the content provider.

4. The computer implemented cache management system of claim 1, wherein the request management unit is further configured to maintain a resource tracking index for tracking popularity measures associated with resources.

5. The computer implemented cache management system of claim 4, wherein the request management unit is further configured to determine the popularity measure associated with the target resource based at least in part on resource popularity information included in the resource tracking index.

6. A computer implemented method for managing a cache of a service provider which is configured to communicate with one or more client devices and with a content provider, the method comprising: receiving a request for a resource from a requestor;selecting a target cache level from a cache hierarchy based at least partly on a popularity measure associated with the resource, wherein the cache hierarchy comprises a plurality of cache levels and wherein at least one of the plurality of cache levels or an origin maintains a cached copy of the resource; andtransmitting the resource to the requestor without causing the target cache level to retrieve the resource from the origin.

7. The computer implemented method of claim 6, wherein different cache levels of the plurality of cache levels are associated with different latencies.

8. The computer implemented method of claim 7, wherein a lowest cache level of the plurality of cache levels is associated with a lowest latency among the plurality of cache levels.

9. The computer implemented method of claim 6, wherein the target cache level corresponds to a cache level superior to a lowest cache level of the plurality of cache levels.

10. The computer implemented method of claim 6 further comprising bypassing one or more cache levels inferior to the target cache level for retrieval of the resource.

11. The computer implemented method of claim 6, wherein a second request for the resource is received at a cache server on the target cache level and wherein the cache server transmits a third request for the resource to a cache level superior to the target cache level.

12. The computer implemented method of claim 6 further comprising determining the popularity measure associated with the resource based at least partly on a measure of frequency or recency of requests for the resource.

13. The computer implemented method of claim 6, wherein selecting the target cache level from the cache hierarchy comprises selecting the target cache level in accordance with a relationship between popularity measures and the plurality of cache levels.

14. The computer implemented method of claim 13 further comprising updating the relationship in accordance with a pattern of resource requests received during a predefined recent period of time.

15. The computer implemented method of claim 6, wherein the resource corresponds to at least a portion of one of an audio presentation, video presentation, audiovisual presentation, movie, television show, video, or podcast.

16. Non-transitory computer storage having stored thereon instructions that, when executed by a computer system configured to communicate with one or more client devices and with a content provider, cause the computer system to: receive a first request for a resource from a requestor;select a first cache level from a cache hierarchy based at least partly on a popularity measure associated with the resource, wherein the cache hierarchy comprises a plurality of cache levels and wherein at least one of the plurality of cache levels or an origin maintains a cached copy of the resource;retrieve the resource from the first cache level, wherein the resource is provided by a cache level superior to a lowest cache level of the plurality of cache levels; andcause return of the resource to the requestor.

17. The non-transitory computer storage of claim 16, wherein the instructions further cause the computer system to receive resource popularity information from an external source.

18. The non-transitory computer storage of claim 17, wherein the popularity information is based at least partly on one or more requests for the resource received at the external source.

19. The non-transitory computer storage of claim 17, wherein the instructions further cause the computer system to determine a second popularity measure associated with the resource based at least in part on the resource popularity information.

20. The non-transitory computer storage of claim 17, wherein the instructions further cause the computer system to: update the popularity measure associated with the resource so that the updated popularity measure is mapped to a second cache level of the cache hierarchy; andin response to a second request for the resource from a second requestor, select the second cache level from the cache hierarchy based at least partly on the updated popularity measure associated with the resource.