DYNAMIC CODE DEPLOYMENT AND VERSIONING

[0001] The present application's Applicant is concurrently filing the following U.S. patent applications on Sep. 30, 2014:

Attorney
Docket No.	Title
SEAZN.982A	MESSAGE-BASED COMPUTATION REQUEST
	SCHEDULING
SEAZN.983A	LOW LATENCY COMPUTATIONAL CAPACITY
	PROVISIONING
SEAZN.984A	AUTOMATIC MANAGEMENT OF LOW LATENCY
	COMPUTATIONAL CAPACITY
SEAZN.989A	THREADING AS A SERVICE
SEAZN.990A	PROGRAMMATIC EVENT DETECTION AND
	MESSAGE GENERATION FOR REQUESTS TO
	EXECUTE PROGRAM CODE
SEAZN.991A	PROCESSING EVENT MESSAGES FOR USER
	REQUESTS TO EXECUTE PROGRAM CODE

[0002] The disclosures of the above-referenced applications are hereby incorporated by reference in their entireties.

[0003] Generally described, computing devices utilize a communication network, or a series of communication networks, to exchange data. Companies and organizations operate computer networks that interconnect a number of computing devices to support operations or provide services to third parties. The computing systems can be located in a single geographic location or located in multiple, distinct geographic locations (e.g., interconnected via private or public communication networks). Specifically, data centers or data processing centers, herein generally referred to as a “data center,” may include a number of interconnected computing systems to provide computing resources to users of the data center. The data centers may be private data centers operated on behalf of an organization or public data centers operated on behalf, or for the benefit of, the general public.

[0004] To facilitate increased utilization of data center resources, virtualization technologies may allow a single physical computing device to host one or more instances of virtual machines that appear and operate as independent computing devices to users of a data center. With virtualization, the single physical computing device can create, maintain, delete, or otherwise manage virtual machines in a dynamic manner. In turn, users can request computer resources from a data center, including single computing devices or a configuration of networked computing devices, and be provided with varying numbers of virtual machine resources.

[0005] In some scenarios, virtual machine instances may be configured according to a number of virtual machine instance types to provide specific functionality. For example, various computing devices may be associated with different combinations of operating systems or operating system configurations, virtualized hardware resources and software applications to enable a computing device to provide different desired functionalities, or to provide similar functionalities more efficiently. These virtual machine instance type configurations are often contained within a device image, which includes static data containing the software (e.g., the OS and applications together with their configuration and data files, etc.) that the virtual machine will run once started. The device image is typically stored on the disk used to create or initialize the instance. Thus, a computing device may process the device image in order to implement the desired software configuration.

[0011] Companies and organizations no longer need to acquire and manage their own data centers in order to perform computing operations (e.g., execute code, including threads, programs, software, routines, subroutines, processes, etc.). With the advent of cloud computing, storage space and compute power traditionally provided by hardware computing devices can now be obtained and configured in minutes over the Internet. Thus, developers can quickly purchase a desired amount of computing resources without having to worry about acquiring physical machines. Such computing resources are typically purchased in the form of virtual computing resources, or virtual machine instances. These instances of virtual machines, which are hosted on physical computing devices with their own operating systems and other software components, can be utilized in the same manner as physical computers.

[0012] However, even when virtual computing resources are purchased, developers still have to decide how many and what type of virtual machine instances to purchase, and how long to keep them. For example, the costs of using the virtual machine instances may vary depending on the type and the number of hours they are rented. In addition, the minimum time a virtual machine may be rented is typically on the order of hours. Further, developers have to specify the hardware and software resources (e.g., type of operating systems and language runtimes, etc.) to install on the virtual machines. Other concerns that they might have include over-utilization (e.g., acquiring too little computing resources and suffering performance issues), under-utilization (e.g., acquiring more computing resources than necessary to run the codes, and thus overpaying), prediction of change in traffic (e.g., so that they know when to scale up or down), and instance and language runtime startup delay, which can take 3-10 minutes, or longer, even though users may desire computing capacity on the order of seconds or even milliseconds. Thus, an improved method of allowing users to take advantage of the virtual machine instances provided by service providers is desired.

[0013] According to aspects of the present disclosure, by dynamically deploying code in response to receiving code execution requests, the delay (sometimes referred to as latency) associated with executing the code (e.g., instance and language runtime startup time) can be significantly reduced.

[0014] Generally described, aspects of the present disclosure relate to the acquisition of user code and the deployment of the user code onto the virtual compute system (e.g., internal storage, virtual machine instances, and/or containers therein). Specifically, systems and methods are disclosed which facilitate management of user code within the virtual compute system. The virtual compute system maintains a pool of virtual machine instances that have one or more software components (e.g., operating systems, language runtimes, libraries, etc.) loaded thereon. The virtual machine instances in the pool can be designated to service user requests to execute program codes. The program codes can be executed in isolated containers that are created on the virtual machine instances. Since the virtual machine instances in the pool have already been booted and loaded with particular operating systems and language runtimes by the time the requests are received, the delay associated with finding compute capacity that can handle the requests (e.g., by executing the user code in one or more containers created on the virtual machine instances) is significantly reduced.

[0015] In another aspect, a virtual compute system may determine that the user code associated with an incoming request is an updated version of the code that has already been loaded onto the virtual compute system. Based on the nature of the incoming request and the state of the virtual compute system, the virtual compute system may determine where the code should be placed and which version of the code should be used to service which request.

[0016] Specific embodiments and example applications of the present disclosure will now be described with reference to the drawings. These embodiments and example applications are intended to illustrate, and not limit, the present disclosure.

[0017] With reference to FIG. 1, a block diagram illustrating an embodiment of a virtual environment 100 will be described. The example shown in FIG. 1 includes a virtual environment 100 in which users (e.g., developers, etc.) of user computing devices 102 may run various program codes using the virtual computing resources provided by a virtual compute system 110.

[0018] By way of illustration, various example user computing devices 102 are shown in communication with the virtual compute system 110, including a desktop computer, laptop, and a mobile phone. In general, the user computing devices 102 can be any computing device such as a desktop, laptop, mobile phone (or smartphone), tablet, kiosk, wireless device, and other electronic devices. In addition, the user computing devices 102 may include web services running on the same or different data centers, where, for example, different web services may programmatically communicate with each other to perform one or more techniques described herein. Further, the user computing devices 102 may include Internet of Things (IoT) devices such as Internet appliances and connected devices. The virtual compute system 110 may provide the user computing devices 102 with one or more user interfaces, command-line interfaces (CLI), application programing interfaces (API), and/or other programmatic interfaces for generating and uploading user codes, invoking the user codes (e.g., submitting a request to execute the user codes on the virtual compute system 110), scheduling event-based jobs or timed jobs, tracking the user codes, and/or viewing other logging or monitoring information related to their requests and/or user codes. Although one or more embodiments may be described herein as using a user interface, it should be appreciated that such embodiments may, additionally or alternatively, use any CLIs, APIs, or other programmatic interfaces.

[0019] The user computing devices 102 access the virtual compute system 110 over a network 104. The network 104 may be any wired network, wireless network, or combination thereof. In addition, the network 104 may be a personal area network, local area network, wide area network, over-the-air broadcast network (e.g., for radio or television), cable network, satellite network, cellular telephone network, or combination thereof. For example, the network 104 may be a publicly accessible network of linked networks, possibly operated by various distinct parties, such as the Internet. In some embodiments, the network 104 may be a private or semi-private network, such as a corporate or university intranet. The network 104 may include one or more wireless networks, such as a Global System for Mobile Communications (GSM) network, a Code Division Multiple Access (CDMA) network, a Long Term Evolution (LTE) network, or any other type of wireless network. The network 104 can use protocols and components for communicating via the Internet or any of the other aforementioned types of networks. For example, the protocols used by the network 104 may include Hypertext Transfer Protocol (HTTP), HTTP Secure (HTTPS), Message Queue Telemetry Transport (MQTT), Constrained Application Protocol (CoAP), and the like. Protocols and components for communicating via the Internet or any of the other aforementioned types of communication networks are well known to those skilled in the art and, thus, are not described in more detail herein.

[0020] The virtual compute system 110 is depicted in FIG. 1 as operating in a distributed computing environment including several computer systems that are interconnected using one or more computer networks. The virtual compute system 110 could also operate within a computing environment having a fewer or greater number of devices than are illustrated in FIG. 1. Thus, the depiction of the virtual compute system 110 in FIG. 1 should be taken as illustrative and not limiting to the present disclosure. For example, the virtual compute system 110 or various constituents thereof could implement various Web services components, hosted or “cloud” computing environments, and/or peer-to-peer network configurations to implement at least a portion of the processes described herein.

[0021] Further, the virtual compute system 110 may be implemented in hardware and/or software and may, for instance, include one or more physical or virtual servers implemented on physical computer hardware configured to execute computer executable instructions for performing various features that will be described herein. The one or more servers may be geographically dispersed or geographically co-located, for instance, in one or more data centers.

[0022] In the environment illustrated FIG. 1, the virtual environment 100 includes a virtual compute system 110, which includes a frontend 120, a warming pool manager 130, a worker manager 140, a versioning and deployment manager 150, and an internal data store 160. In the depicted example, virtual machine instances (“instances”) 152, 154 are shown in a warming pool 130A managed by the warming pool manager 130, and instances 156, 158 are shown in an active pool 140A managed by the worker manager 140, The illustration of the various components within the virtual compute system 110 is logical in nature and one or more of the components can be implemented by a single computing device or multiple computing devices. For example, the instances 152, 154, 156, 158 can be implemented on one or more physical computing devices in different various geographic regions. Similarly, each of the frontend 120, the warming pool manager 130, the worker manager 140, the versioning and deployment manager 150, and the internal data store 160 can be implemented across multiple physical computing devices. Alternatively, one or more of the frontend 120, the warming pool manager 130, the worker manager 140, the versioning and deployment manager 150, and the internal data store 160 can be implemented on a single physical computing device. In some embodiments, the virtual compute system 110 may comprise multiple frontends, multiple warming pool managers, multiple worker managers, multiple deployment managers, and/or multiple internal data stores. Although four virtual machine instances are shown in the example of FIG. 1, the embodiments described herein are not limited as such, and one skilled in the art will appreciate that the virtual compute system 110 may comprise any number of virtual machine instances implemented using any number of physical computing devices. Similarly, although a single warming pool and a single active pool are shown in the example of FIG. 1, the embodiments described herein are not limited as such, and one skilled in the art will appreciate that the virtual compute system 110 may comprise any number of warming pools and active pools.

[0023] In the example of FIG. 1, the virtual compute system 110 is illustrated as being connected to the network 104. In some embodiments, any of the components within the virtual compute system 110 can communicate with other components (e.g., the user computing devices 102 and auxiliary services 106, which may include monitoring/logging/billing services 107, storage service 108, an instance provisioning service 109, and/or other services that may communicate with the virtual compute system 110) of the virtual environment 100 via the network 104. In other embodiments, not all components of the virtual compute system 110 are capable of communicating with other components of the virtual environment 100. In one example, only the frontend 120 may be connected to the network 104, and other components of the virtual compute system 110 may communicate with other components of the virtual environment 100 via the frontend 120.

[0024] Users may use the virtual compute system 110 to execute user code thereon. For example, a user may wish to run a piece of code in connection with a web or mobile application that the user has developed. One way of running the code would be to acquire virtual machine instances from service providers who provide infrastructure as a service, configure the virtual machine instances to suit the user's needs, and use the configured virtual machine instances to run the code. Alternatively, the user may send a code execution request to the virtual compute system 110. The virtual compute system 110 can handle the acquisition and configuration of compute capacity (e.g., containers, instances, etc., which are described in greater detail below) based on the code execution request, and execute the code using the compute capacity. The virtual compute system 110 may automatically scale up and down based on the volume, thereby relieving the user from the burden of having to worry about over-utilization (e.g., acquiring too little computing resources and suffering performance issues) or under-utilization (e.g., acquiring more computing resources than necessary to run the codes, and thus overpaying).

[0025] The frontend 120 processes all the requests to execute user code on the virtual compute system 110. In some embodiments, the frontend 120 serves as a front door to all the other services provided by the virtual compute system 110. The frontend 120 processes the requests and makes sure that the requests are properly authorized. For example, the frontend 120 may determine whether the user associated with the request is authorized to access the user code specified in the request.

[0026] The user code as used herein may refer to any program code (e.g., a program, routine, subroutine, thread, etc.) written in a specific program language. In the present disclosure, the terms “code,” “user code,” and “program code,” may be used interchangeably. Such user code may be executed to achieve a specific task, for example, in connection with a particular web application or mobile application developed by the user. For example, the user codes may be written in JavaScript (node.js), Java, Python, and/or Ruby. The request may include the user code (or the location thereof) and one or more arguments to be used for executing the user code. For example, the user may provide the user code along with the request to execute the user code. In another example, the request may identify a previously uploaded program code (e.g., using the API for uploading the code) by its name or its unique ID. In yet another example, the code may be included in the request as well as uploaded in a separate location (e.g., the storage service 108 or a storage system internal to the virtual compute system 110) prior to the request is received by the virtual compute system 110. The virtual compute system 110 may vary its code execution strategy based on where the code is available at the time the request is processed.

[0027] The frontend 120 may receive the request to execute such user codes in response to Hypertext Transfer Protocol Secure (HTTPS) requests from a user. Also, any information (e.g., headers and parameters) included in the HTTPS request may also be processed and utilized when executing the user code. As discussed above, any other protocols, including, for example, HTTP, MQTT, and CoAP, may be used to transfer the message containing the code execution request to the frontend 120. The frontend 120 may also receive the request to execute such user codes when an event is detected, such as an event that the user has registered to trigger automatic request generation. For example, the user may have registered the user code with an auxiliary service 106 and specified that whenever a particular event occurs (e.g., a new file is uploaded), the request to execute the user code is sent to the frontend 120. Alternatively, the user may have registered a timed job (e.g., execute the user code every 24 hours). In such an example, when the scheduled time arrives for the timed job, the request to execute the user code may be sent to the frontend 120. In yet another example, the frontend 120 may have a queue of incoming code execution requests, and when the user's batch job is removed from the virtual compute system's work queue, the frontend 120 may process the user request. In yet another example, the request may originate from another component within the virtual compute system 110 or other servers or services not illustrated in FIG. 1.

[0028] A user request may specify one or more third-party libraries (including native libraries) to be used along with the user code. In one embodiment, the user request is a ZIP file containing the user code and any libraries (and/or identifications of storage locations thereof). In some embodiments, the user request includes metadata that indicates the program code to be executed, the language in which the program code is written, the user associated with the request, and/or the computing resources (e.g., memory, etc.) to be reserved for executing the program code. For example, the program code may be provided with the request, previously uploaded by the user, provided by the virtual compute system 110 (e.g., standard routines), and/or provided by third parties. In some embodiments, such resource-level constraints (e.g., how much memory is to be allocated for executing a particular user code) are specified for the particular user code, and may not vary over each execution of the user code. In such cases, the virtual compute system 110 may have access to such resource-level constraints before each individual request is received, and the individual requests may not specify such resource-level constraints. In some embodiments, the user request may specify other constraints such as permission data that indicates what kind of permissions that the request has to execute the user code. Such permission data may be used by the virtual compute system 110 to access private resources (e.g., on a private network).

[0029] In some embodiments, the user request may specify the behavior that should be adopted for handling the user request. In such embodiments, the user request may include an indicator for enabling one or more execution modes in which the user code associated with the user request is to be executed. For example, the request may include a flag or a header for indicating whether the user code should be executed in a debug mode in which the debugging and/or logging output that may be generated in connection with the execution of the user code is provided back to the user (e.g., via a console user interface). In such an example, the virtual compute system 110 may inspect the request and look for the flag or the header, and if it is present, the virtual compute system 110 may modify the behavior (e.g., logging facilities) of the container in which the user code is executed, and cause the output data to be provided back to the user. In some embodiments, the behavior/mode indicators are added to the request by the user interface provided to the user by the virtual compute system 110. Other features such as source code profiling, remote debugging, etc. may also be enabled or disabled based on the indication provided in the request.

[0030] In some embodiments, the virtual compute system 110 may include multiple frontends 120. In such embodiments, a load balancer may be provided to distribute the incoming requests to the multiple frontends 120, for example, in a round-robin fashion. In some embodiments, the manner in which the load balancer distributes incoming requests to the multiple frontends 120 may be based on the state of the warming pool 130A and/or the active pool 140A. For example, if the capacity in the warming pool 130A is deemed to be sufficient, the requests may be distributed to the multiple frontends 120 based on the individual capacities of the frontends 120 (e.g., based on one or more load balancing restrictions). On the other hand, if the capacity in the warming pool 130A is less than a threshold amount, one or more of such load balancing restrictions may be removed such that the requests may be distributed to the multiple frontends 120 in a manner that reduces or minimizes the number of virtual machine instances taken from the warming pool 130A. For example, even if, according to a load balancing restriction, a request is to be routed to Frontend A, if Frontend A needs to take an instance out of the warming pool 130A to service the request but Frontend B can use one of the instances in its active pool to service the same request, the request may be routed to Frontend B.

[0031] The warming pool manager 130 ensures that virtual machine instances are ready to be used by the worker manager 140 when the virtual compute system 110 receives a request to execute user code on the virtual compute system 110. In the example illustrated in FIG. 1, the warming pool manager 130 manages the warming pool 130A, which is a group (sometimes referred to as a pool) of pre-initialized and pre-configured virtual machine instances that may be used to service incoming user code execution requests. In some embodiments, the warming pool manager 130 causes virtual machine instances to be booted up on one or more physical computing machines within the virtual compute system 110 and added to the warming pool 130A. In other embodiments, the warming pool manager 130 communicates with an auxiliary service (e.g., the instance provisioning service 109 of FIG. 1) to create and add new instances to the warming pool 130A. In some embodiments, the warming pool manager 130 may utilize both physical computing devices within the virtual compute system 110 and one or more virtual machine instance services to acquire and maintain compute capacity that can be used to service code execution requests received by the frontend 120. In some embodiments, the virtual compute system 110 may comprise one or more logical knobs or switches for controlling (e.g., increasing or decreasing) the available capacity in the warming pool 130A. For example, a system administrator may use such a knob or switch to increase the capacity available (e.g., the number of pre-booted instances) in the warming pool 130A during peak hours. In some embodiments, virtual machine instances in the warming pool 130A can be configured based on a predetermined set of configurations independent from a specific user request to execute a user's code. The predetermined set of configurations can correspond to various types of virtual machine instances to execute user codes. The warming pool manager 130 can optimize types and numbers of virtual machine instances in the warming pool 130A based on one or more metrics related to current or previous user code executions.

[0032] As shown in FIG. 1, instances may have operating systems (OS) and/or language runtimes loaded thereon. For example, the warming pool 130A managed by the warming pool manager 130 comprises instances 152, 154. The instance 152 includes an OS 152A and a runtime 152B. The instance 154 includes an OS 154A. In some embodiments, the instances in the warming pool 130A may also include containers (which may further contain copies of operating systems, runtimes, user codes, etc.), which are described in greater detail below. Although the instance 152 is shown in FIG. 1 to include a single runtime, in other embodiments, the instances depicted in FIG. 1 may include two or more runtimes, each of which may be used for running a different user code. In some embodiments, the warming pool manager 130 may maintain a list of instances in the warming pool 130A. The list of instances may further specify the configuration (e.g., OS, runtime, container, etc.) of the instances.

[0033] In some embodiments, the virtual machine instances in the warming pool 130A may be used to serve any user's request. In one embodiment, all the virtual machine instances in the warming pool 130A are configured in the same or substantially similar manner. In another embodiment, the virtual machine instances in the warming pool 130A may be configured differently to suit the needs of different users. For example, the virtual machine instances may have different operating systems, different language runtimes, and/or different libraries loaded thereon. In yet another embodiment, the virtual machine instances in the warming pool 130A may be configured in the same or substantially similar manner (e.g., with the same OS, language runtimes, and/or libraries), but some of those instances may have different container configurations. For example, two instances may have runtimes for both Python and Ruby, but one instance may have a container configured to run Python code, and the other instance may have a container configured to run Ruby code. In some embodiments, multiple warming pools 130A, each having identically-configured virtual machine instances, are provided.

[0034] The warming pool manager 130 may pre-configure the virtual machine instances in the warming pool 130A, such that each virtual machine instance is configured to satisfy at least one of the operating conditions that may be requested or specified by the user request to execute program code on the virtual compute system 110. In one embodiment, the operating conditions may include program languages in which the potential user codes may be written. For example, such languages may include Java, JavaScript, Python, Ruby, and the like. In some embodiments, the set of languages that the user codes may be written in may be limited to a predetermined set (e.g., set of 4 languages, although in some embodiments sets of more or less than four languages are provided) in order to facilitate pre-initialization of the virtual machine instances that can satisfy requests to execute user codes. For example, when the user is configuring a request via a user interface provided by the virtual compute system 110, the user interface may prompt the user to specify one of the predetermined operating conditions for executing the user code. In another example, the service-level agreement (SLA) for utilizing the services provided by the virtual compute system 110 may specify a set of conditions (e.g., programming languages, computing resources, etc.) that user requests should satisfy, and the virtual compute system 110 may assume that the requests satisfy the set of conditions in handling the requests. In another example, operating conditions specified in the request may include: the amount of compute power to be used for processing the request; the type of the request (e.g., HTTP vs. a triggered event); the timeout for the request (e.g., threshold time after which the request may be terminated); security policies (e.g., may control which instances in the warming pool 130A are usable by which user); and etc.

[0035] The worker manager 140 manages the instances used for servicing incoming code execution requests. In the example illustrated in FIG. 1, the worker manager 140 manages the active pool 140A, which is a group (sometimes referred to as a pool) of virtual machine instances that are currently assigned to one or more users. Although the virtual machine instances are described here as being assigned to a particular user, in some embodiments, the instances may be assigned to a group of users, such that the instance is tied to the group of users and any member of the group can utilize resources on the instance. For example, the users in the same group may belong to the same security group (e.g., based on their security credentials) such that executing one member's code in a container on a particular instance after another member's code has been executed in another container on the same instance does not pose security risks. Similarly, the worker manager 140 may assign the instances and the containers according to one or more policies that dictate which requests can be executed in which containers and which instances can be assigned to which users. An example policy may specify that instances are assigned to collections of users who share the same account (e.g., account for accessing the services provided by the virtual compute system 110). In some embodiments, the requests associated with the same user group may share the same containers (e.g., if the user codes associated therewith are identical). In some embodiments, a request does not differentiate between the different users of the group and simply indicates the group to which the users associated with the requests belong.

[0036] In the example illustrated in FIG. 1, user codes are executed in isolated compute systems referred to as containers. Containers are logical units created within a virtual machine instance using the resources available on that instance. For example, the worker manager 140 may, based on information specified in the request to execute user code, create a new container or locate an existing container in one of the instances in the active pool 140A and assigns the container to the request to handle the execution of the user code associated with the request. In one embodiment, such containers are implemented as Linux containers. The virtual machine instances in the active pool 140A may have one or more containers created thereon and have one or more program codes associated with the user loaded thereon (e.g., either in one of the containers or in a local cache of the instance). Each container may have credential information made available therein, so that user codes executing on the container have access to whatever the corresponding credential information allows them to access.

[0037] As shown in FIG. 1, instances may have operating systems (OS), language runtimes, and containers. The containers may have individual copies of the OS and the language runtimes and user codes loaded thereon. In the example of FIG. 1, the active pool 140A managed by the worker manager 140 includes the instances 156, 158. The instance 156 has containers 156A, 156B. The container 156A has OS 156A-1, runtime 156A-2, and code 156A-3 loaded therein. In the depicted example, the container 156A has its own OS, runtime, and code loaded therein. In one embodiment, the OS 156A-1 (e.g., the kernel thereof), runtime 156A-2, and/or code 156A-3 are shared among the containers 156A, 156B (and any other containers not illustrated in FIG. 1). In another embodiment, the OS 156A-1 (e.g., any code running outside the kernel), runtime 156A-2, and/or code 156A-3 are independent copies that are created for the container 156A and are not shared with other containers on the instance 156. In yet another embodiment, some portions of the OS 156A-1, runtime 156A-2, and/or code 156A-3 are shared among the containers on the instance 156, and other portions thereof are independent copies that are specific to the container 156A. The instance 158 includes containers 158A, 158B and a code cache 159C for storing code executed in any of the containers on the instance 158.

[0038] In the example of FIG. 1, the sizes of the containers depicted in FIG. 1 may be proportional to the actual size of the containers. For example, the container 156A may occupy more space than the container 156B on the instance 156. Similarly, the containers 158A, 158B may be equally sized. The dotted boxes labeled “C” shown in the instance 158 indicate the space remaining on the instances that may be used to create new containers. In some embodiments, the sizes of the containers may be 64 MB or any multiples thereof. In other embodiments, the sizes of the containers may be any arbitrary size smaller than or equal to the size of the instances in which the containers are created. In some embodiments, the sizes of the containers may be any arbitrary size smaller than, equal to, or larger than the size of the instances in which the containers are created. By how much the sizes of the containers can exceed the size of the instance may be determined based on how likely that those containers might be utilized beyond the capacity provided by the instance.

[0039] Although the components inside the containers 156B, 158A are not illustrated in the example of FIG. 1, each of these containers may have various operating systems, language runtimes, libraries, and/or user code. In some embodiments, instances may have user codes loaded thereon (e.g., in an instance-level cache such as the code cache 159C), and containers within those instances may also have user codes loaded therein (e.g., container 156A). In some embodiments, the worker manager 140 may maintain a list of instances in the active pool 140A. The list of instances may further specify the configuration (e.g., OS, runtime, container, etc.) of the instances. In some embodiments, the worker manager 140 may have access to a list of instances in the warming pool 130A (e.g., including the number and type of instances). In other embodiments, the worker manager 140 requests compute capacity from the warming pool manager 130 without having knowledge of the virtual machine instances in the warming pool 130A.

[0040] After a request has been successfully processed by the frontend 120, the worker manager 140 finds capacity to service the request to execute user code on the virtual compute system 110. For example, if there exists a particular virtual machine instance in the active pool 140A that has a container with the same user code loaded therein (e.g., code 156A-3 shown in the container 156A), the worker manager 140 may assign the container to the request and cause the user code to be executed in the container. Alternatively, if the user code is available in the local cache of one of the virtual machine instances (e.g., stored in the code cache 159C of the instance 158 but do not belong to any individual containers), the worker manager 140 may create a new container on such an instance, assign the container to the request, and cause the user code to be loaded and executed in the container.

[0041] If the worker manager 140 determines that the user code associated with the request is not found on any of the instances (e.g., either in a container or the local cache of an instance) in the active pool 140A, the worker manager 140 may determine whether any of the instances in the active pool 140A is currently assigned to the user associated with the request and has compute capacity to handle the current request. If there is such an instance, the worker manager 140 may create a new container on the instance and assign the container to the request. Alternatively, the worker manager 140 may further configure an existing container on the instance assigned to the user, and assign the container to the request. For example, the worker manager 140 may determine that the existing container may be used to execute the user code if a particular library demanded by the current user request is loaded thereon. In such a case, the worker manager 140 may load the particular library and the user code onto the container and use the container to execute the user code.

[0042] If the active pool 140A does not contain any instances currently assigned to the user, the worker manager 140 pulls a new virtual machine instance from the warming pool 130A, assigns the instance to the user associated with the request, creates a new container on the instance, assigns the container to the request, and causes the user code to be downloaded and executed on the container.

[0043] In some embodiments, the virtual compute system 110 is adapted to begin execution of the user code shortly after it is received (e.g., by the frontend 120). A time period can be determined as the difference in time between initiating execution of the user code (e.g., in a container on a virtual machine instance associated with the user) and receiving a request to execute the user code (e.g., received by a frontend). The virtual compute system 110 is adapted to begin execution of the user code within a time period that is less than a predetermined duration. In one embodiment, the predetermined duration is 500 ms. In another embodiment, the predetermined duration is 300 ms. In another embodiment, the predetermined duration is 100 ms. In another embodiment, the predetermined duration is 50 ms. In another embodiment, the predetermined duration is 10 ms. In another embodiment, the predetermined duration may be any value chosen from the range of 10 ms to 500 ms. In some embodiments, the virtual compute system 110 is adapted to begin execution of the user code within a time period that is less than a predetermined duration if one or more conditions are satisfied. For example, the one or more conditions may include any one of: (1) the user code is loaded on a container in the active pool 140A at the time the request is received; (2) the user code is stored in the code cache of an instance in the active pool 140A at the time the request is received; (3) the active pool 140A contains an instance assigned to the user associated with the request at the time the request is received; or (4) the warming pool 130A has capacity to handle the request at the time the request is received.

[0044] The user code may be downloaded from an auxiliary service 106 such as the storage service 108 of FIG. 1. Data 108A illustrated in FIG. 1 may comprise user codes uploaded by one or more users, metadata associated with such user codes, or any other data utilized by the virtual compute system 110 to perform one or more techniques described herein. Although only the storage service 108 is illustrated in the example of FIG. 1, the virtual environment 100 may include other levels of storage systems from which the user code may be downloaded. For example, each instance may have one or more storage systems either physically (e.g., a local storage resident on the physical computing system on which the instance is running) or logically (e.g., a network-attached storage system in network communication with the instance and provided within or outside of the virtual compute system 110) associated with the instance on which the container is created. Alternatively, the code may be downloaded from a web-based data store provided by the storage service 108.

[0045] Once the worker manager 140 locates one of the virtual machine instances in the warming pool 130A that can be used to serve the user code execution request, the warming pool manager 130 or the worker manger 140 takes the instance out of the warming pool 130A and assigns it to the user associated with the request. The assigned virtual machine instance is taken out of the warming pool 130A and placed in the active pool 140A. In some embodiments, once the virtual machine instance has been assigned to a particular user, the same virtual machine instance cannot be used to service requests of any other user. This provides security benefits to users by preventing possible co-mingling of user resources. Alternatively, in some embodiments, multiple containers belonging to different users (or assigned to requests associated with different users) may co-exist on a single virtual machine instance. Such an approach may improve utilization of the available compute capacity. In some embodiments, the virtual compute system 110 may maintain a separate cache in which user codes are stored to serve as an intermediate level of caching system between the local cache of the virtual machine instances and a web-based network storage (e.g., accessible via the network 104).

[0046] After the user code has been executed, the worker manager 140 may tear down the container used to execute the user code to free up the resources it occupied to be used for other containers in the instance. Alternatively, the worker manager 140 may keep the container running to use it to service additional requests from the same user. For example, if another request associated with the same user code that has already been loaded in the container, the request can be assigned to the same container, thereby eliminating the delay associated with creating a new container and loading the user code in the container. In some embodiments, the worker manager 140 may tear down the instance in which the container used to execute the user code was created. Alternatively, the worker manager 140 may keep the instance running to use it to service additional requests from the same user. The determination of whether to keep the container and/or the instance running after the user code is done executing may be based on a threshold time, the type of the user, average request volume of the user, and/or other operating conditions. For example, after a threshold time has passed (e.g., 5 minutes, 30 minutes, 1 hour, 24 hours, 30 days, etc.) without any activity (e.g., running of the code), the container and/or the virtual machine instance is shutdown (e.g., deleted, terminated, etc.), and resources allocated thereto are released. In some embodiments, the threshold time passed before a container is torn down is shorter than the threshold time passed before an instance is torn down.

[0047] In some embodiments, the virtual compute system 110 may provide data to one or more of the auxiliary services 106 as it services incoming code execution requests. For example, the virtual compute system 110 may communicate with the monitoring/logging/billing services 107. The monitoring/logging/billing services 107 may include: a monitoring service for managing monitoring information received from the virtual compute system 110, such as statuses of containers and instances on the virtual compute system 110; a logging service for managing logging information received from the virtual compute system 110, such as activities performed by containers and instances on the virtual compute system 110; and a billing service for generating billing information associated with executing user code on the virtual compute system 110 (e.g., based on the monitoring information and/or the logging information managed by the monitoring service and the logging service). In addition to the system-level activities that may be performed by the monitoring/logging/billing services 107 (e.g., on behalf of the virtual compute system 110) as described above, the monitoring/logging/billing services 107 may provide application-level services on behalf of the user code executed on the virtual compute system 110. For example, the monitoring/logging/billing services 107 may monitor and/or log various inputs, outputs, or other data and parameters on behalf of the user code being executed on the virtual compute system 110. Although shown as a single block, the monitoring, logging, and billing services 107 may be provided as separate services.

[0048] In some embodiments, the worker manager 140 may perform health checks on the instances and containers managed by the worker manager 140 (e.g., those in the active pool 140A). For example, the health checks performed by the worker manager 140 may include determining whether the instances and the containers managed by the worker manager 140 have any issues of (1) misconfigured networking and/or startup configuration, (2) exhausted memory, (3) corrupted file system, (4) incompatible kernel, and/or any other problems that may impair the performance of the instances and the containers. In one embodiment, the worker manager 140 performs the health checks periodically (e.g., every 5 minutes, every 30 minutes, every hour, every 24 hours, etc.). In some embodiments, the frequency of the health checks may be adjusted automatically based on the result of the health checks. In other embodiments, the frequency of the health checks may be adjusted based on user requests. In some embodiments, the worker manager 140 may perform similar health checks on the instances and/or containers in the warming pool 130A. The instances and/or the containers in the warming pool 130A may be managed either together with those instances and containers in the active pool 140A or separately. In some embodiments, in the case where the health of the instances and/or the containers in the warming pool 130A is managed separately from the active pool 140A, the warming pool manager 130, instead of the worker manager 140, may perform the health checks described above on the instances and/or the containers in the warming pool 130A.

[0049] The versioning and deployment manager 150 manages the deployment of user code on the virtual compute system 110. For example, the versioning and deployment manager 150 may communicate with the frontend 120, the warming pool manager 130, the worker manager 140, and/or the internal data store 160 to manage the deployment of user code onto any internal data store, instance-level code cache, and/or containers on the virtual compute system 110. Although the versioning and deployment manager 150 is illustrated as a distinct component within the virtual compute system 110, part or all of the functionalities of the versioning and deployment manager 150 may be performed by the frontend 120, the warming pool manager 130, the worker manager 140, and/or the internal data store 160. For example, the versioning and deployment manager 150 may be implemented entirely within one of the other components of the virtual compute system 110 or in a distributed manner across the other components of the virtual compute system 110. In the example of FIG. 1, the versioning and deployment manager 150 includes code deployment data 150A. The code deployment data 150A may include data regarding the history of incoming requests, versions of the user code executed on the virtual compute system 110, and any other metric that may be used by the versioning and deployment manager 150 to adjust and/or optimize the deployment of the user code associated with the incoming code execution requests. The code deployment data 150A may also include any management policies specified by the users or determined by the versioning and deployment manager 150 for deploying their code (e.g., versioning preferences, etc.) on the virtual compute system 110.

[0050] Throughout the lifecycle of a user code, various updates may be made to the code. In some embodiments, the versioning and deployment manager 150 maintains a list of all the user codes executing on the virtual compute system 110, and when the versioning and deployment manager 150 determines that one or more of the users codes have been updated, the versioning and deployment manager 150 causes the updated user codes to be used (instead of the older versions thereof) in connection with subsequent code execution requests received by the virtual compute system 110. For example, when a user updates a particular user code using one API and makes requests associated with the user code using another API, the virtual compute system 110 may programmatically determine when the requests associated with the user code should be processed with the new version (e.g., based on the size of the new version, availability of the older version in the active pool 130A, etc.). In some embodiments, the request may include an indication that the user code associated with the request has been updated. For example, the user may specify that the code has been updated. In another example, the user may specify the version of the code that he or she wishes to use, and the versioning and deployment manager 150 may determine, for each request, whether the version specified by the user is different from the one or more versions of the code that might be running on the virtual compute system 110. In yet another example, the request may include an identifier that is unique to the code (e.g., date of creation, date of modification, hash value, etc.). In other embodiments, the versioning and deployment manager 150 may automatically determine, based on the user code received along with the request, whether there has been any updates to the user code. For example, the versioning and deployment manager 150 may calculate a hash value or a checksum of the code and determine whether the code is different from the one or more versions of the code that might be running on the virtual compute system 110.

[0051] In some embodiments, the versioning and deployment manager 150 may cause one or more of the requests that are received after the user code has been updated to be serviced using the older version of the code. For example, when a new version of the user code is detected, the versioning and deployment manager 150 may allow any containers that are in the middle of executing the older version of the user code to finish before loading the new version onto those containers. In some embodiments, the versioning and deployment manager 150 allows the older version of the code to be used while the new version is being downloaded onto an internal data store, a code cache of a particular instance, and/or a container. By using the older version of the code while the new version is being downloaded, any latency increase due to the change may be avoided by overlaying the procurement of the new version (e.g., latest corrected/request version) with the execution of the requests. In some embodiments, the versioning and deployment manager 150 may immediately start downloading the new version onto those containers having the older version loaded thereon (or onto new containers that are created on the instances on which those containers having the older version loaded thereon are created), In other embodiments, the versioning and deployment manager 150 may download the new version onto those containers having the older version loaded thereon (or onto new containers that are created on the instances on which those containers having the older version loaded thereon are created) after those containers become idle (e.g., not currently executing any user code). In some embodiments, the versioning and deployment manager 150 may determine how long it might take to download the new version associated with a particular request, and decide to service the particular request using the older version if the download time exceeds a threshold value. In some embodiments, the versioning and deployment manager 150 may determine how many requests associated with the code are being received, and decide to service one or more of the requests using the older version if one or more containers already have the older version loaded, and if not enough containers have the newer version loaded thereon to serve all of the requests.

[0052] In some embodiments, the versioning and deployment manager 150 determines, based on a user request, how quickly the older versions of the code should be removed from the virtual compute system 110. For example, the user associated with the code execution request may indicate that the update is a minor one and that the user would prefer the latency to be as low as possible. In such an example, the versioning and deployment manager 150 may keep running the older versions of the code and gradually phase in the newer version (e.g., when the instances running the older version have all been discarded or otherwise not available or have enough capacity to handle all the incoming requests associated with the code). In another example, the user may indicate that the older versions have a security bug that is exposing customers' credit card information and that all previous versions of the code should be killed immediately. In such an example, the versioning and deployment manager 150 may stop and/or terminate any containers running the older versions of the code and begin using the newer version immediately.

[0053] In some embodiments, the versioning and deployment manager 150, based on the history of the volume of requests received by the virtual compute system 110, may preemptively load a program code that is sufficiently frequently executed on the virtual compute system 110 onto one or more containers in the active pool 140A. In some embodiments, the versioning and deployment manager 150 causes certain codes to remain in the container and/or the instance if the code is anticipated to be executed in a cyclical manner. For example, if the versioning and deployment manager 150 determines that the virtual compute system 110 a particular code receives 90% of its requests between the hours of 7 PM and 8 PM, the versioning and deployment manager 150 may cause the particular code to be remain in the containers even after hours of inactivity. In some embodiments, the versioning and deployment manager 150 may preemptively load the new version onto one or more containers in the active pool 140A or the warming pool 130A, when the versioning and deployment manager 150 detects the new version, even before any request associated with the new version is received.

[0054] In the example of FIG. 1, the versioning and deployment manager 150 maintains the internal data store 160 that is used to store data accessed by one or more instances. For example, the versioning and deployment manager 150 may store user code onto the internal data store 160 so that the user code can be shared among multiple instances. In some embodiments, the data stored on the internal data store 160 in connection with such multiple instances (e.g., user codes executed in containers created on such instances) remains on the internal data store 160 for use by other instances even after the particular instance is shut down. In one example, downloading a code onto a container from the internal data store 160 is more than 10 times faster than downloading the same code onto a container from a data store external to the virtual compute system 110 (e.g., storage service 108). In some embodiments, the internal data store 160 is divided into isolated containers (which may provide additional security among such containers), and access to each container is restricted to one or more instances associated therewith.

[0055] The versioning and deployment manager 150 may include a code deployment unit for analyzing incoming code execution requests received by the virtual compute system 110 and determining where and how user code should be acquired and deployed. An example configuration of the versioning and deployment manager 150 is described in greater detail below with reference to FIG. 6.

[0056] With reference to FIGS. 2-5, an example versioning scheme for handling code execution requests after user code has been updated will be described. In the example of FIG. 2, the storage service 106 has code (1) 108B (“code (1)”) loaded thereon, and the instance 158 has a code cache 158C with the code (1) loaded thereon and four containers that are busy executing the code (1) 108B. The dotted boxes labeled “C” shown in the instance 158 indicate the space remaining on the instances that may be used to create new containers.

[0057] In FIG. 3, the code (1) previously stored on the storage service 108 has been updated to code (2) 108C (“code (2)”). FIG. 3 also shows that one of the containers has become idle, and the other three containers are still busy executing code (1) (e.g., in connection with existing requests associated with the code (1) or new requests associated with the code (2)). For example, the three containers are still executing the now-out-of-date code (1) even after the code has been updated to code (2), for example, to reduce the latency associated with servicing the request. The versioning and deployment manager 150 may have initiated the download of the code (2) at this point.

[0058] In FIG. 4, the code (2) has finished downloading onto the code cache 159C. The code (2) has also been loaded onto the previously idle container and two new containers. The first three containers are still executing the code (1) (e.g., servicing new requests associated with the code (2) using the code (1) loaded thereon).

[0059] FIG. 5 illustrates the configuration after all the containers have switched to the code (2) and are running the code (2) in connection with incoming code execution requests. In FIG. 5, the code (1) has also been removed from the code cache 159C, for example, by the versioning and deployment manager 150 after it has determined (e.g., based on the time elapsed since the code (1) was updated, or based on a user indication to eventually phase out the code (1)) that the code (1) is no longer needed. In some embodiments, after the code (2) has been fully phased in, assuming the level of incoming code execution requests does not change, the same number of containers (e.g., four in the example of FIGS. 2-5) may be able to handle the incoming code execution requests associated with a particular code, regardless of which version of the particular code is used.

[0060] Thus, by continuing to service incoming code execution requests using an older version of the code even after the code has been updated, existing containers having the older version of the code loaded thereon can be utilized to reduce the latency associated with processing the code execution requests.

[0061] FIG. 6 depicts a general architecture of a computing system (referenced as versioning and deployment manager 150) that manages the deployment of user code in the virtual compute system 110. The general architecture of the versioning and deployment manager 150 depicted in FIG. 6 includes an arrangement of computer hardware and software modules that may be used to implement aspects of the present disclosure. The versioning and deployment manager 150 may include many more (or fewer) elements than those shown in FIG. 6. It is not necessary, however, that all of these generally conventional elements be shown in order to provide an enabling disclosure. As illustrated, the versioning and deployment manager 150 includes a processing unit 190, a network interface 192, a computer readable medium drive 194, an input/output device interface 196, all of which may communicate with one another by way of a communication bus. The network interface 192 may provide connectivity to one or more networks or computing systems. The processing unit 190 may thus receive information and instructions from other computing systems or services via the network 104. The processing unit 190 may also communicate to and from memory 180 and further provide output information for an optional display (not shown) via the input/output device interface 196. The input/output device interface 196 may also accept input from an optional input device (not shown).

[0062] The memory 180 may contain computer program instructions (grouped as modules in some embodiments) that the processing unit 190 executes in order to implement one or more aspects of the present disclosure. The memory 180 generally includes RAM, ROM and/or other persistent, auxiliary or non-transitory computer-readable media. The memory 180 may store an operating system 184 that provides computer program instructions for use by the processing unit 190 in the general administration and operation of the versioning and deployment manager 150. The memory 180 may further include computer program instructions and other information for implementing aspects of the present disclosure. For example, in one embodiment, the memory 180 includes a user interface unit 182 that generates user interfaces (and/or instructions therefor) for display upon a computing device, e.g., via a navigation and/or browsing interface such as a browser or application installed on the computing device. In addition, the memory 180 may include and/or communicate with one or more data repositories (not shown), for example, to access user program codes and/or libraries.

[0063] In addition to and/or in combination with the user interface unit 182, the memory 180 may include a code deployment unit 186 that may be executed by the processing unit 190. In one embodiment, the user interface unit 182, and code deployment unit 186 individually or collectively implement various aspects of the present disclosure, e.g., analyzing incoming code execution requests received by the virtual compute system 110, determining where and how user code should be acquired and deployed, etc. as described further below.

[0064] The code deployment unit 186 analyzes incoming code execution requests received by the virtual compute system 110. For example, the code deployment unit 186 may determine whether the user code associated with an incoming request is a newer version of a code that is loaded on one or more of the containers of the virtual compute system 110. Based on the nature of the incoming request and the state of the virtual compute system 110, the code deployment unit 186 determines where the code should be placed and which code should be used to service which request.

[0065] While the code deployment unit 186 is shown in FIG. 6 as part of the versioning and deployment manager 150, in other embodiments, all or a portion of the code deployment unit 186 may be implemented by other components of the virtual compute system 110 and/or another computing device. For example, in certain embodiments of the present disclosure, another computing device in communication with the virtual compute system 110 may include several modules or components that operate similarly to the modules and components illustrated as part of the versioning and deployment manager 150.

[0066] Turning now to FIG. 7, a routine 700 implemented by one or more components of the virtual compute system 110 (e.g., the versioning and deployment manager 150) will be described. Although routine 700 is described with regard to implementation by the versioning and deployment manager 150, one skilled in the relevant art will appreciate that alternative components may implement routine 700 or that one or more of the blocks may be implemented by a different component or in a distributed manner.

[0067] At block 702 of the illustrative routine 700, the versioning and deployment manager 150 receives a code execution request associated a user code. For example, the versioning and deployment manager 150 may receive the request from the frontend 120 after the frontend has performed any initial processing on the request. As discussed above, the request may specify the code to be executed on the virtual compute system 110, and any operating conditions such as the amount of compute power to be used for processing the request, the type of the request (e.g., HTTP vs. a triggered event), the timeout for the request (e.g., threshold time after which the request may be terminated), security policies (e.g., may control which instances in the warming pool 130A are usable by which user), etc.

[0068] At block 704, the versioning and deployment manager 150 detects that the code associated with the request is an updated version of a code that has already been loaded onto the virtual compute system 110. For example, one or more containers may have the older version of the code associated with the request loaded thereon.

[0069] At block 706, the versioning and deployment manager 150 initiates a download of the updated version of the code onto the virtual compute system 110. For example, the versioning and deployment manager 150 cause the updated version of the code to be downloaded onto an internal data store of the virtual compute system 110 (e.g., internal data store 160 of FIG. 1), a code cache on one of the instances (e.g., code cache 158C of FIG. 1), or one or more containers created on the virtual compute system 110.

[0070] At block 708, the versioning and deployment manager 150 causes the code execution request associated with the updated version of the code to be processed with an older version of the code that was previously loaded on one of the containers before the code execution request was received at block 702.

[0071] While the routine 700 of FIG. 7 has been described above with reference to blocks 702-708, the embodiments described herein are not limited as such, and one or more blocks may be added, omitted, modified, or switched without departing from the spirit of the present disclosure. For example, the routine 700 may further include block 710, where the versioning and deployment manager 150 causes a subsequent code execution request associated with the updated version of the code to be processed with the updated version of the code loaded onto one of the containers after the download initiated at block 706 has completed.

[0072] It will be appreciated by those skilled in the art and others that all of the functions described in this disclosure may be embodied in software executed by one or more physical processors of the disclosed components and mobile communication devices. The software may be persistently stored in any type of non-volatile storage.

[0073] Conditional language, such as, among others, “can,” “could,” “might,” or “may,” 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 steps. Thus, such conditional language is not generally intended to imply that features, elements and/or steps are in any way required for one or more embodiments or that one or more embodiments necessarily include logic for deciding, with or without user input or prompting, whether these features, elements and/or steps are included or are to be performed in any particular embodiment.

[0074] Any process descriptions, elements, or blocks in the flow diagrams described herein and/or depicted in the attached figures should be understood as potentially representing modules, segments, or portions of code which include one or more executable instructions for implementing specific logical functions or steps in the process. Alternate implementations are included within the scope of the embodiments described herein in which elements or functions may be deleted, executed out of order from that shown or discussed, including substantially concurrently or in reverse order, depending on the functionality involved, as would be understood by those skilled in the art. It will further be appreciated that the data and/or components described above may be stored on a computer-readable medium and loaded into memory of the computing device using a drive mechanism associated with a computer readable storage medium storing the computer executable components such as a CD-ROM, DVD-ROM, or network interface. Further, the component and/or data can be included in a single device or distributed in any manner. Accordingly, general purpose computing devices may be configured to implement the processes, algorithms, and methodology of the present disclosure with the processing and/or execution of the various data and/or components described above.

[0075] It should be emphasized that many variations and modifications may be made to the above-described embodiments, the elements of which are to be understood as being among other acceptable examples. All such modifications and variations are intended to be included herein within the scope of this disclosure and protected by the following claims.