Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Claims 1-20 are pending in this application.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claim(s) 1-20 are rejected under 35 U.S.C. 103 as being unpatentable over Gillen et al. (US Pub. No. 2012/0303670 A1 hereinafter Gillen) in view of Funke et al. (US Patent No. 11,615,117 B2 hereinafter Funke) in view of Wang et al. (US Pub. No. 2024/0143407 A1 hereinafter Wang).
As per claim 1, Gillen teaches a method comprising: performing one or more operations in a first user session using a cluster of machines in a distributed graph processing system (¶ [0015], “Data processing and/or storage functions may be organized in a graph or tree structure which may be distributed across an unbounded number of physical machines, for example, in a cloud computing environment. A process may comprise a number of steps implemented by software and/or hardware and may, for example, utilize applications, algorithms and code. One or more processes may manipulate and/or store data within the graph or tree structure and may also control the environment on which the processes are running, based on their own needs. The processes may be operable to dynamically modify and grow the graph or tree structure beyond physical machine boundaries while continuing to perform data storage and/or data processing functions therein.” ¶ [0017], “In some instances, the cloud computing platform 100 may provide vast amounts of computing assets and storage availability to the process that modifies and grows the graph or tree structure among a plurality of physical machines and performs data storage and/or data processing functions via the graph or tree structure. In this regard, portions of the graph or tree may respectively reside on a plurality of different physical nodes or machines in the cloud computing platform 100.”), wherein: the cluster of machines has a first number of machines providing a first available amount of a resource (¶ [0049]-[0050], “The following steps comprise an illustrative process for scaling a graph or tree as it grows beyond what can be stored in physical memory of a primary device on which it resides…The process virtual machine P1 210 may be running on device 110 and may send information to the service manager virtual machine SM1 216. The information may indicate that a graph or tree scaling process may begin. The process VM P1 210 may communicate to the service manager VM SM1 216 that memory usage has reached a specified level or threshold. For example, P1 210 may indicate that 70% of available RAM from memory 114 has been consumed.”), performing the one or more operations comprises: extending the first user session to the one or more new machines, and performing the one or more operations in the extended first user session, wherein the method is performed by one or more computing devices (¶ [0033], “The VM P1 210 may be operable to retrieve data or tasks and may process the tasks and/or store data, utilizing a graph or tree structure. P1 210 may also be operable to scale the graph or tree structure such that a branch of the graph or tree structure may reside and/or grow on a different physical machine and different virtual machine(s). The scaling process may begin, for example, when RAM memory or processing capacity becomes limiting on a physical machine which supports the VM P1 210. For example, the VM P1 210 may run on the device 110 shown in FIG. 1 and may utilize a graph or tree structure. When the memory 114 is utilized to a specified capacity, the system may begin branching the graph or tree structure to the device 120 such that a first portion of the tree resides on the device 110 while a second portion of the tree resides on the device 120. A new processing virtual machine P2 218 may be created on the device 120 and a portion of the graph or tree from the VM P1 210 may be copied to the device 120 and may be utilized and grown in a similar manner by the processing virtual machine P2 218.” ¶ [0050], “SM1 216 may begin a process to scale a portion of the graph or tree 300 to a second device 120. The service manager VM SM1 216 may communicate with APIs in the cloud infrastructure 100 to instantiate a new process VM P2 218, based on the process image IMG1 214. The new process VM instance may become P2 218 and it may be instantiated on the device 120. Once P2 218 becomes online, it may begin to poll the service manager VM SM1 216 on a regular basis for a transaction queue to monitor. If none are available, the processing node P2 218 may sleep for a period of time and then may check again.”).
Gillen fails to teach estimating the amount of resources one or more operations are expected to use and adding new machines in order to meet the expected amount of resources.
However, Funke teaches the first available amount of resources includes a first free amount of resources that is not allocated, used, or reserved by one or more previous user operations or data objects in the distributed graph processing system (Col. 5 & 6, lines 57-67 & 1-13, “If a user specifies a maximum cluster count that is greater than a minimum cluster count, the resource manager 102 may automatically manage the number of currently active clusters based on the workload to satisfy the throughput criteria and to be cost-effective. So, whenever the warehouse is running, at least a minimum cluster count (minClusterCount) of clusters are active, and at most a maximum cluster count (maxClusterCount) of clusters are active. The resource manager 102 may decide how many clusters are required to handle the current workload given the specified performance criteria in terms of memory load and concurrency level.” Col. 8, lines 22-43, “The warehouse 302 includes a plurality of clusters (Cluster 1, Cluster 2, Cluster N) that each include a plurality of server nodes. In one embodiment, each of the clusters includes the same number of servers although this may be different in different embodiments. In one embodiment, each server in a cluster belong to the same availability zone but different clusters may be placed in different availability zones. The concept of availability of the warehouse may be based on overall availability percentage of the warehouse. For example, the availability for a specific cluster within the warehouse 302 may be the percentage of servers which are available (e.g., in an operational state) relatively to the cluster size. However, when that percentage goes below the minimum (e.g., 50%) required to run a query 0% availability may be determined for that cluster and no queries may be assigned until the warehouse 302, or some of the servers in the warehouse 302, is repaired. As discussed herein, the number of clusters in the warehouse 302 may be adjusted dynamically based on workload, server failures in the clusters, or the like.”), the one or more operations are estimated to use an expected amount of the resource (Col. 8, lines 44-56, “In one embodiment, the query scheduler and coordinator 218 weights each query (e.g., SQL statement or portion of a SQL statement) based on its projected resource consumption. For example, some queries may take significantly more memory to perform while other queries may take significantly more processing resources to perform. Similarly, some queries may have high or low consumption for both memory and processing.”), a resource manager adds one or more new machines to the cluster of machines to form a second number of machines in response to the expected amount of the resource being greater than the first free amount of the resource, and the second number of machines provide a second available amount of the resource including a second free amount of the resource (Col. 3, lines 22-33, “In one embodiment, a method for a multi-cluster warehouse may include dynamically adding compute clusters to the virtual warehouse based on the workload. The method may include determining whether a query can be processed while meeting a performance metric for the query. If the query in combination with a current workload does not allow one or more currently allocated compute clusters to meet the performance metric, the method may include triggering startup of a new compute cluster.” Col. 13, lines 28-34, “The system triggers 704 startup of a new compute cluster in response to determining that the query in combination with a current workload does not allow one or more currently allocated compute clusters to meet the performance metric. In one embodiment, the system may only trigger 704 startup if the number of currently active clusters is less than a predetermined maximum number of compute clusters.” See also Col. 5, lines 41-56.).
Gillen and Funke are considered to be analogous to the claimed invention because they are in the same field of dynamic resource allocation. Therefore, it would have be obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the method of Gillen with the techniques for estimating resource needs and adding resources to meet such needs as taught by Funke to arrive at the claimed invention. The motivation to modify Gillen with the teachings of Funke is that the support of dynamic changes allows a processing platform to scale quickly in response to changing demands from the users on the systems and components within data processing platform (See Funke Col. 5, lines 24-40.).
Although Gillen and Funke teach a generic resource manager that adds new machines to the cluster of machines, they fail to explicitly teach a control plane that adds new machines to the cluster of machines.
Accordingly, Wang teaches a control plane adds one or more new machines to the cluster of machines to form a second number of machines (¶ [0017], “A control plane may be used to manage the nodes and the pods in the cluster. Pods can be atomic units, i.e., smallest size of deployable units that can be created an managed within a system. In a production environment, a control plane may run across multiple computers and a cluster may run multiple nodes in order to provide fault-tolerance and availability.” ¶ [0026], “As shown, the control plane 150 may receive a software application request 210 for backend services provided via the API server 130 of FIG. 1. The proxy 220 is deployed to receive and/or intercept such a request 210 and interacts with the control plane scaler 230 to manage resource allocations in the API server 130 (or other control plane components). For example, the proxy 220 may instruct the control plane scaler 230 regarding a proper amount of resources to be assigned to the backend services (e.g., services provided by any of the control plane 150 components, such as the API server 130 and others) based on context-aware heuristics. In some cases, the control plane scaler 230 includes a vertical and/or autoscaler (e.g., with corresponding monitoring, determination, and adjustment components to change computational resources such as CPU cores, frequencies, memory, and bandwidths, and a number of replicas in horizontal scaling).”).
Gillen, Funke, and Wang are all considered to be analogous to the claimed invention because they are all in the same field of dynamic resource allocation. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to substitute the control plane of Wang for the resource manager of Gillen and Funke to arrive at the claimed invention. This substitution would have been reasonable and yielded predictable results under MPEP § 2143 as all references dynamically change the amount of allocated resources in order to meet demands of the processing system.
As per claim 2, Gillen, Funke, and Wang teach the method of claim 1. Funke teaches wherein performing the one or more operations comprises: pausing a given operation within the one or more operations in response to the expected amount of the resource being greater than the first free amount of the resource; and resuming the given operation in the extended first user session in response to the second free amount of the resource being greater than the expected amount of the resource (Col. 3, lines 8-21, “Automatically resuming or starting a new or suspended cluster may be performed when the warehouse cannot handle the workload and would have to queue queries (or queue queries longer than an accepted length of time). Queries can get queued because the total resource consumption on the cluster has exceeded a threshold. For example, the resource consumption may include parameters for memory load as well as computing or processing load. In one embodiment, a parameter controls for how long a query may be queued before a new cluster should be resumed or provisioned. As soon as the new cluster has resumed, queries can be scheduled to execute on the new cluster. This applies to new queries as well as already queued queries.” Col. 7, lines 9-22, “Workload considerations include at least two things. First, workload considerations may account for memory usage. When queries are scheduled and are queued because all clusters are at their maximum memory capacity, the virtual warehouse manager 220 will resume one or more clusters so that queueing can be avoided, or shortened. Queuing may still occur if new clusters need to be resumed since resuming a cluster may take a bit of time, for example in minutes. However, the virtual warehouse manager 220 may also make sure that there is a free pool of several free servers so that queries can be put on the free pool during the starting of the new cluster.” See also Col. 8 & 9, lines 57-67 & 1-14.).
Refer to claim 1 for reason to combine.
As per claim 3, Gillen, Funke, and Wang teach the method of claim 2. Wang teaches wherein the resource comprises memory of disk storage (¶ [0051], “The amount of resources to be assigned to or updated in the backend service may include at least one of: a number of central processing unit (CPU) cores; a respective operating frequency of the number of CPU cores; an amount of random access memory (RAM); or a network bandwidth for the backend service.”).
Gillen, Funke, and Wang are all considered to be analogous to the claimed invention because they are all in the same field of dynamic resource allocation. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Gillen and Funke with the ability to add more memory to a processing system as taught by Wang to arrive at the claimed invention. The motivation to modify Gillen and Funke with the teachings of Wang is that dynamically adjusting the amount of memory for a processing system improves resource utilization efficiency and avoids underutilizing resources (See Wang para. 0014.).
As per claim 4, Gillen, Funke, and Wang teach the method of claim 2. Gillen teaches wherein the one or more operations include a second operation that is performed without interruption (¶ [0042], “In some cases, the backend 430A may be scaled vertically by updating the number of assigned CPU cores, the frequencies thereof, and RAM capacities. In adjusting bandwidth, the backend 430A may be apportioned with different number of antennas and software or hardware resources limiting data transfer rates. The backend 430A may be updated to the backend 430B with scaled resources to meet the demands by the backend service request 410.”).
As per claim 5, Gillen, Funke, and Wang teach the method of claim 1. Funke teaches wherein the distributed graph processing system communicates the first free amount of the resource and the expected amount of the resource to the control plane (Col. 8, lines 22-56, “The concept of availability of the warehouse may be based on overall availability percentage of the warehouse. For example, the availability for a specific cluster within the warehouse 302 may be the percentage of servers which are available (e.g., in an operational state) relatively to the cluster size. However, when that percentage goes below the minimum (e.g., 50%) required to run a query 0% availability may be determined for that cluster and no queries may be assigned until the warehouse 302, or some of the servers in the warehouse 302, is repaired…In one embodiment, the query scheduler and coordinator 218 weights each query (e.g., SQL statement or portion of a SQL statement) based on its projected resource consumption. For example, some queries may take significantly more memory to perform while other queries may take significantly more processing resources to perform. Similarly, some queries may have high or low consumption for both memory and processing.”).
Refer to claim 1 for reason to combine.
As per claim 6, Gillen, Funke, and Wang teach the method of claim 1. Funke teaches wherein: the control plane communicates a maximum amount of the resource in a pool of machines, and the distributed graph processing system determines that a sum of the first available amount of the resource and an unavailable portion of the expected amount of the resource does not exceed the maximum amount of the resource (Col. 5 & 6, lines 57-67 & 1-13, “If a user specifies a maximum cluster count that is greater than a minimum cluster count, the resource manager 102 may automatically manage the number of currently active clusters based on the workload to satisfy the throughput criteria and to be cost-effective. So, whenever the warehouse is running, at least a minimum cluster count (minClusterCount) of clusters are active, and at most a maximum cluster count (maxClusterCount) of clusters are active.” Col. 13, lines 28-34, “The system triggers 704 startup of a new compute cluster in response to determining that the query in combination with a current workload does not allow one or more currently allocated compute clusters to meet the performance metric. In one embodiment, the system may only trigger 704 startup if the number of currently active clusters is less than a predetermined maximum number of compute clusters.”).
Refer to claim 1 for reason to combine.
As per claim 7, Gillen, Funke, and Wang teach the method of claim 1. Gillen teaches wherein each operation within the one or more operations is a graph loading operation, a graph query operation, or a graph processing algorithm (¶ [0027], “Virtual machines running on one or more of the devices 110, 120 and 130 may comprise processes that may create a graph and/or tree structure and may process and/or store data in nodes of the graph or tree structure. Nodes of the graph or tree structure may be different than physical nodes in the cloud computing platform, for example, a single physical device or node in the cloud computing platform may comprise a graph or tree comprising a plurality of nodes.” See also para. 0062.).
As per claim 8, Gillen, Funke, and Wang teach the method of claim 1. Funke teaches wherein the one or more previous user operations are performed in a second user session (Col. 4 & 5, lines 50-67 & 1-40, “The execution platform 112 may include one or more compute clusters which may be dynamically allocated or suspended for specific warehouses, based on the query workload provided by the users 104-108 to a specific warehouse…In some embodiments, the communication links between resource manager 102 and users 104-108, metadata 110, and execution platform 112 are implemented via one or more data communication networks and may be assigned various tasks such that user requests can be optimized…As shown in FIG. 1, data storage devices 116, 118, and 120 are decoupled from the computing resources associated with execution platform 112. This architecture supports dynamic changes to the data processing platform 100 based on the changing data storage/retrieval needs, computing needs, as well as the changing needs of the users and systems accessing data processing platform 100. The support of dynamic changes allows data processing platform 100 to scale quickly in response to changing demands on the systems and components within data processing platform 100.”).
Refer to claim 1 for reason to combine.
As per claim 9, Gillen, Funke, and Wang teach the method of claim 1. Gillen teaches wherein: a given operation within the one or more operations has a graph object associated with the first user session and one or more dependent objects that depend on the graph object, and performing the one or more operations in the extended first user session comprises extending the graph object and the one or more dependent objects to machines within the cluster of machines based on a hierarchy of dependencies (¶ [0020], “The processors 112, 122 and/or 132 may control and/or process growth of the scalable graph or tree structures and may handle data storage and/or data processing functions utilizing the graph or tree structure which may extend to and reside on a plurality of the physical machines.” ¶ [0027], “The graph or tree structure may comprise a root node and one or child or descendant nodes embodied on a first physical device. Processes of one or more virtual machine in the first device may scale the graph or tree so that a branch may be embodied on another physical device. Thus, if a root node is embodied on the device 110, a portion of the nodes of the graph or tree below the root node may reside on the same physical device and a portion may be scaled or graphed to one or more other devices, such as the device 120.” See also para. 0045-0047.).
As per claim 10, Gillen, Funke, and Wang teach the method of claim 1. Wang teaches wherein the resource comprises processor cores or network bandwidth (¶ [0051], “The amount of resources to be assigned to or updated in the backend service may include at least one of: a number of central processing unit (CPU) cores; a respective operating frequency of the number of CPU cores; an amount of random access memory (RAM); or a network bandwidth for the backend service.”).
Gillen, Funke, and Wang are all considered to be analogous to the claimed invention because they are all in the same field of dynamic resource allocation. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Gillen and Funke with the ability to add more memory to a processing system as taught by Wang to arrive at the claimed invention. The motivation to modify Gillen and Funke with the teachings of Wang is that dynamically adjusting the amount of memory for a processing system improves resource utilization efficiency and avoids underutilizing resources (See Wang para. 0014.).
As per claim 11, it is a method claim comprising similar limitations to claim 1, so it is rejected for similar reasons. Wang also teaches monitoring a processing system by a control plane (¶ [0017], “A cluster may include a set of nodes (e.g., worker machines) that run containerized applications (e.g., packaged with runtime, libraries, etc.). The nodes can host pods that are components of the application workload. A control plane may be used to manage the nodes and the pods in the cluster. Pods can be atomic units, i.e., smallest size of deployable units that can be created an managed within a system. In a production environment, a control plane may run across multiple computers and a cluster may run multiple nodes in order to provide fault-tolerance and availability. Control plane components, such as etcd (e.g., a persistent storage), API aggregator, and control manager, often stay alive all of the time during operation in existing deployments, regardless of actual workload demanded by the nodes.”).
Refer to claim 1 for reason to combine.
As per claim 12, Gillen, Funke, and Wang teach the method of claim 11. Funke teaches wherein monitoring the distributed graph processing system further comprises removing a given machine from the cluster of machines in response to the first available amount of the resource being greater than the first free amount of the resource by an amount of the resource in the given machine (Col. 13, lines 35-67, “The method 800 begins and a system determines 802 whether a current workload is serviceable by one fewer than the plurality of compute clusters while meeting a performance metric. In one embodiment, determining 802 whether the current workload is serviceable by one fewer than the plurality of compute clusters may include determining whether a historical workload for a time period leading up to the current time was serviceable by one fewer than the plurality of clusters while meeting the performance metric…The system decommissions 804 (or inactivating) at least one compute cluster of the plurality of compute clusters in response to determining that the workload is serviceable by one fewer than the plurality of compute clusters. The system may only decommission 804 or remove a compute cluster if the current number of active clusters is less than a predetermined minimum number of compute clusters.”).
Refer to claim 1 for reason to combine.
As per claim 13, it is a non-transitory storage media claim comprising similar limitations to claim 1, so it is rejected for similar reasons. Gillen also teaches a non-transitory storage media storing instructions (¶ [0083], “The software may be embodied in any computer-readable medium or signal-bearing medium, for use by, or in connection with an instruction executable system, apparatus, and device, resident to system that may maintain a persistent or non-persistent connection with a destination.”).
As per claim 14, it is a non-transitory storage media claim comprising similar limitations to claim 2, so it is rejected for similar reasons.
As per claim 15, it is a non-transitory storage media claim comprising similar limitations to claim 3, so it is rejected for similar reasons.
As per claim 16, it is a non-transitory storage media claim comprising similar limitations to claim 4, so it is rejected for similar reasons.
As per claim 17, it is a non-transitory storage media claim comprising similar limitations to claim 5, so it is rejected for similar reasons.
As per claim 18, it is a non-transitory storage media claim comprising similar limitations to claim 6, so it is rejected for similar reasons.
As per claim 19, it is a non-transitory storage media claim comprising similar limitations to claim 7, so it is rejected for similar reasons.
As per claim 20, it is a non-transitory storage media claim comprising similar limitations to claim 9, so it is rejected for similar reasons.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure. Tang et al. (US Patent No. 10,412,022 B1) teaches identifying a resource capacity and scaling a system to meet the identified capacity.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to JOHN ROBERT DAKITA EWALD whose telephone number is (703)756-1845. The examiner can normally be reached Monday-Friday: 9:00-5:30 ET.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Lewis Bullock can be reached at (571)272-3759. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/J.D.E./Examiner, Art Unit 2199
/LEWIS A BULLOCK JR/Supervisory Patent Examiner, Art Unit 2199
Prosecution Insights