Abstract: The present embodiments relate to identifying a ransomware attack. One embodiment relates to a method comprising configuring an operating system to collect metrics related to a hardware component. A message can be received from a user space library to validate an instruction detected in a cache, the instruction being associated with the hardware component. A metric can be compared to a threshold metric. The metric can be associated with the hardware component. A likelihood of a ransomware attack can be determined based at least in part on the comparison. A message can be transmitted to the user space library comprising the determination of the likelihood of the ransomware.

Abstract: A system obtains telemetry data associated with a fabric. The fabric includes a set of interconnected switches, organized according to a fabric topology, that provide multiple redundant pathways for routing data between a set of computing devices linked to the fabric. Based at least on the fabric topology, the system identifies a portion of the telemetry data corresponding to a first subsection of the fabric. Based at least on the fabric topology and the portion of the telemetry data, the system determines a performance metric corresponding to the first subsection of the fabric. Based at least on the performance metric, the system steers at least a portion of network traffic from the first subsection of the fabric to a second subsection of the fabric.

Abstract: Techniques for identity domain snapshotting and snapshot consumption using versioning are disclosed. A snapshot of an updated identity domain is received and applied to an identity domain to change the active version of the identity domain from a prior current version to the updated version. The snapshot is consumed by loading data objects in a hierarchical manner into an updated identity domain. While data objects belonging to the snapshot are loaded during consumption of the snapshot, client traffic continues to be served by the prior, non-updated version of the identity domain. Once consumption of the snapshot is completed successfully, the updated version of the identity domain becomes the active version. The prior version is then recorded in a cleanup table denoting that cleanup operations shall be performed for entities related to the data objects and associated with the prior version.

Abstract: Continual learning techniques are described for extending the capabilities of a base model, which is trained to predict a set of existing or base classes, to generate a target model that is capable of making predictions for both the existing or base classes and additionally for making predictions for new or custom classes. The techniques described herein enable the target model to be trained such that the model can make predictions involving both base classes and custom classes with high levels of accuracy.

Abstract: A system receives a candidate scenario that includes a proposed modification to a routing configuration of a fabric. The fabric includes a set of interconnected switches, organized according to a fabric topology, that provide multiple redundant pathways for routing data between a set of computing devices linked to the fabric. The system determines a performance metric corresponding to the candidate scenario based at least on the fabric topology and telemetry data associated with the fabric. Based at least on the performance metric, the system determines whether a criterion for proceeding with implementing the candidate scenario is satisfied. In response to determining that the criterion is satisfied, the system generates an instruction for implementing the candidate scenario including the proposed modification to the routing configuration of the fabric.

Abstract: Techniques discussed herein relate to automatically detecting and managing resource drift (e.g., differences between a requested/actual state and a predefined or approved state) corresponding to a set of resources of a cloud computing environment (e.g., resources of a landing zone environment). Requested changes may be intercepted before applied and a determination of whether to allow or reject the changes may be made based at least in part on contextual data corresponding to the user/entity/resource to which the requested change applies. Previously-applied changes may be detected (e.g., based on current/approved state data comparisons) and a determination of whether to allow the change to persist or to remediate the change may be similarly made based at least in part on contextual data. User interfaces may be provided to view the resource drift across a cloud computing environment. User input for allowing/rejecting/reverting resource changes may be obtained via these interfaces.

Abstract: Techniques are disclosed for building a region at a prefab factory. A manager service can implement a virtual bootstrap environment at a data center communicatively connected to a prefab factory data center. The manager service can deploy software resources to physical resources using the virtual bootstrap environment. The manager service can configure the physical resources for shipping to a destination site by at least generating a network configuration corresponding to a network topology of the physical resources in the prefab factory data center.

Abstract: The present disclosure relates to network connectivity and reachability monitoring in cloud environments. In an example, a computer receives, from a device of a customer, input indicating a destination that belongs to a virtual cloud network and a set of vantage points for testing the destination. Based on the input, a first source is deployed to a first network location associated with a first vantage point and a second source is deployed to a second network location associated with a second vantage point. Each source is configured to send probes of different types to the destination. The computer system receives, probe data from each source based on probes sent by the source to the destination. Based on the received probe data, the computer system generates test data indicating at least one of connectivity or reachability of the destination. The test data is presented at a user interface.

Abstract: Systems and methods described herein related to real-time management of service deployment systems are disclosed. A management system may monitor an infrastructure environment repository in real-time. Differences in configurations stored within the repository may be identified and used to generate configuration data. The configuration data may be used to update service operating parameters and service deployment configurations. The configuration data may be used to generate deployment data comprising location information of the updated infrastructure environment. The deployment data may be utilized to deploy one or more services on an updated infrastructure using the updated services.

Abstract: In some embodiments, a system, article, and computer-implemented method are disclosed for analyzing and addressing issues. A natural language description of a particular issue is accessed and encoded to a particular issue vector embedding. A database is accessed which includes a plurality of issues associated with a plurality of different issue pattern vector embeddings. Distances are determined for pairs of the particular issue vector embedding and one or more individual issue pattern vector embeddings of the plurality of different issue pattern vector embeddings to determine a similarity. Information identifying the one or more issues are displayed with one or more sets of responsive actions associated with the one or more issues. Feedback is received indicating that a selected issue of the one or more issues is the particular issue and that selected actions are responsive to the particular issue. Weights for the selected actions are adjusted.

Abstract: Techniques for reducing the number of high-risk deployments that negatively impact critical resources of a computer system are disclosed. One or more embodiments provide two different deployment workflows for deploying containerized applications to a container orchestration platform. A first deployment workflow requires a first set of approvers. A second deployment workflow requires a second set of approvers. One or more embodiments train a machine learning model to evaluate a deployment manifest for a target deployment and determine if the deployment is a high-risk deployment. If the deployment is not high risk, then the system executes the first deployment workflow. If the deployment is high risk, then the system executes the second deployment workflow.

Abstract: Techniques are described for managing secure connections (e.g., tunnels) between different endpoints using a pod of servers. Instead of computing devices connecting to a single server at a service IP address, the connections are spread among the different servers in the pod that can be reached using a public IP address.

Abstract: In a computer-implemented embodiment, an interaction machine learning model is trained based on many interactions on many resources. A context lexical token is inferred that represents a current operational context of a user. The context lexical token is inserted into a sequence of other inferred lexical tokens. From the context lexical token within the sequence of tokens, the interaction machine learning model infers a predicted resource that will be accessed next. In an embodiment, accelerated matchmaking entails suitability measurement by a dot product of a) a dynamically inferred user embedding that is based on the context lexical token and b) a statically inferred item embedding.

Abstract: The present is directed to systems, methods, and devices for holistic rendering of cloud network configuration. The method can include receiving data characterizing a plurality of devices in a computing network. The method can include generating with the inventory processor a data file characterizing each of the plurality of devices in the computing network. This data file can be generated based on the received data and on a set of static overrides. The method can include generating a configuration file for each of the plurality of devices in the computing network via iterative selection and application of templates to portions of the data file.

Abstract: Techniques are described for performing different types of restart operations for a file storage replication between a source file system and a target file system in different cloud infrastructure regions. In certain embodiments, the disclosed techniques perform a restart operation to terminate a current cross-region replication by synchronizing resource cleanup operations in the source file system and the target file system, respectively. In other embodiments, disclosed techniques perform a restart operation to allow a customer to reuse the source file system by identifying a restartable base snapshot in the source file system without dependency on the target file system.

Abstract: A database system generates a prompt for an LLM or other machine learning (ML) model to narrow the search space to highly relevant information about a database. A distinct instance of a classifier, a clustering algorithm, or a topic modeling model can be trained based on information from ML automation within the database system, respectively for each column or table in the database. Model instances can then be used during generative LLM inferencing to identify relevant sources of data to answer the user's question. Thus, the prompt generation combines ML automation and other ML models or an LLM for topic modeling and schema description.

Abstract: Techniques for extracting information from unstructured documents that enable an ML model to be trained such that the model can accurately distinguish in-distribution (“in-D”) elements and out-of-distribution (“OO-D”) elements within an unstructured document. Novel training techniques are used that train an ML model using a combination of a regular training dataset and an enhanced augmented training dataset. The regular training dataset is used to train an ML model to identify in-D elements, i.e., to classify an element extracted from a document as belonging to one of the in-D classes contained in the regular training dataset. The augmented training dataset, which is generated based upon the regular training dataset may contain one or more augmented elements which are used to train the model to identify OO-D elements, i.e., to classify an augmented element extracted from a document as belonging to an OO-D class instead of to an in-D class.

Abstract: A database-aware storage server provides instant creation of snapshots without the need to create an intermediate test master database. During the snapshot creation time, the source database stays read-write and completes ongoing reads and writes. The database-aware storage server allows creation of layers of writable snapshots in a hierarchy. All these multiple databases share common data blocks. Any new writes performed by the database post snapshot are stored in blocks of sparse files. This promotes space sharing and reduces the total amount of space used by all these related databases. The allocations for the source and all new snapshot databases share the same common pool of storage. The newly created snapshot databases can access the data store directly without going through an intermediate layer.

Abstract: Here is schema generation from a data table and reference documentation, including inferential analysis and generation of natural language. A large language model (LLM) inferentially generates an initial column summary that consists of natural language that describes exactly one column of a data table. From a vector store that contains many fixed-size portions of many reference documents, a few nearest neighbor portions are dense retrieved that are semantically similar to the initial column summary. From the initial column summary and the nearest neighbor portions, the LLM inferentially generates an expanded column summary that describes the column, including inferring that some nearest neighbor portions should be excluded. From the expanded column summary, the LLM inferentially generates a) a data schema that consists of semi-structured text that describes the data table, including a column comment that contains natural language that describes the column.

Abstract: Techniques are disclosed for network pathway diagnosis. The system accesses sets of training data. The sets of training data define associations between issues of computer networks and network pathways of the computer networks. The training data is used to train a machine learning model to select for diagnosis network pathways in computer networks. An issue is detected in a computer network that incorporates a cluster of computing nodes configured for executing containerized applications. The trained machine learning model is applied to select a target network pathway in the computer network for diagnosis based on a target set of characteristics that are associated with the issue. Diagnosing the target network pathway reveals that the target network pathway is dysfunctional, why the target network pathway is dysfunctional, and/or actions that potentially resolve the dysfunction of the target network pathway.

Abstract: A method for network analytics data director (NADD)-informed configuration of a 3gpp-Sbi-Max-Rsp-Time header value includes receiving, at the NADD and from network functions (NFs), NF configuration details and copies of service-based interface (SBI) messages transmitted to and received by the NFs, determining service operation processing times of the NFs, and communicating, to an NF service consumer, NF analytics data including the service operation processing times and the NF configuration details. The method further includes automatically determining, by the NF service consumer and using the NF analytics data, a 3gpp-Sbi-Max-Rsp-Time header value for an SBI request message, adding, by the NF service consumer, the 3gpp-Sbi-Max-Rsp-Time header value to the SBI request message, and transmitting, by the NF service consumer, the SBI request message to a destination.

Abstract: The present disclosure relates to systems and methods for automatic rule generation based on natural language input. Natural language input can be received. The natural language input can be tokenized. First tokens can be mapped to a first condition of a rule, and second tokens can be mapped to a second condition of the rule. A graph representation of the natural language input can be generated. A pre-generated, tenant-specific graph can be selected that corresponds to the graph representation of the natural language input. A rule can be generated based on the tenant-specific graph. The rule can be provided to facilitate implementation of the rule.

Abstract: Techniques for presenting a graphical user interface (GUI) for configuring a cloud service workstation are disclosed. The system presents a GUI that presents a plurality of possible workstation configurations and the costs associated with each respective workstation configuration, prior to creation of a workstation. The GUI updates the cost associated with a workstation configuration responsive to receiving a selection to modify the workstation configuration from a user. The user may request a different configuration based on a single user input, without specifying which resources to modify. The GUI may recommend a workstation configuration based on one or more user inputs such as a budget, an application service domain, a duration, or a processing power requirement.

Abstract: According to an aspect, a system performs a first machine learning (ML) technique and a second ML technique on a large set of data items. The first ML technique identifies a primary cluster containing data items classified as non-anomalous, with the rest of the data items classified as anomalous. The second ML technique identifies a set of secondary clusters. Upon receiving feedback on the classification of at least some of the data items, the system maintains a feedback data associating the feedback with each secondary cluster. Upon receiving an indication to find a classification of a data item, the system identifies as a result, whether the data item is non-anomalous or anomalous based on whether the data item belongs to the primary cluster or not. The system also determines based on a secondary cluster to which the data item belongs and the feedback associated with the secondary cluster, whether the result is to be modified.

Abstract: Embodiments determine a final occupancy prediction for a check-in date for a plurality of hotel rooms. Embodiments receive historical reservation data including a plurality of booking curves for the hotel rooms corresponding to a plurality of reservation windows, the historical reservation data including a plurality of features. Based on the historical reservation data, embodiments generate a first occupancy prediction for the check-in date using a first model and generate a second occupancy prediction for the check-in date using a second model. Embodiments determine a best performing model from at least the first model and the second model uses a corresponding occupancy prediction corresponding to the best performing model as the final occupancy prediction for the check-in date.

Abstract: Techniques for generating natural language (NL) queries are disclosed. The system generates NL queries that are schema-aligned, thereby mitigating linking errors during the conversion from NL queries to SQL queries. The system receives a request to generate an NL query that comprises a reference SQL query and identifies a schema. To generate the NL query from the SQL query, the system determines schema elements corresponding to the target schema and uses those schema elements to generate a schema-aligned NL query. The system provides the reference SQL query and a relevant portion of the target schema to a SQL-to-NL machine learning model that generates the NL query, taking into consideration the schema elements.

Abstract: Techniques for garbage collection using detail cards are disclosed. A card in a card table and a detail card in a detail card table are associated with a heap segment. Bits in the detail card are associated, respectively, with heap subsegments of the heap segment. A garbage collector marks a first bit in the detail card as dirty. During a sweep cycle, the garbage collector determines that the card is clean, the first bit is dirty, and a second bit in the detail card is clean. During the sweep cycle, based on the card being clean, the first bit being dirty, and the second bit being clean, the garbage collector (a) scans a heap subsegment associated with the first bit for intergenerational references and (b) refrains from scanning a heap subsegment associated with the second bit for intergenerational references.

Abstract: Techniques for providing a private service access (PSA) object for accessing a cloud service from a virtual cloud network (VCN) using a private internet protocol (IP) address are provided. In an example method, a cloud computing environment associated with a cloud service provider (CSP) provides a PSA object for accessing a cloud service from a VCN, the PSA object being associated with a virtual network interface card (VNIC), the VNIC being associated with a private IP address, in which the IP address is from a range of private addresses associated with the VCN. The cloud computing environment receives, by the VNIC associated with the PSA object, a request to access the cloud service from a client connected to the VCN. The cloud computing environment routes the request to the cloud service using the VNIC and the PSA object.

Abstract: Techniques for managing requests to modify a data set at future times are disclosed. The system receives a request to modify the data set at a future time, such as in a future time window. The system generates a representation of the data set at the future time. The system determines the future state of the data set based on the current state of the data set and requested future modifications to the data set. The system determines whether a conflict exists between the request to modify the data set at the future time and another scheduled request to modify the data set at the future time. Based on determining whether a conflict exists, the system performs one or more of scheduling the request to modify the data set at the future time and performing a remediation action to address the conflict.

Abstract: Techniques are disclosed herein for knowledge augmentation in supervised instruction fine-tuning of generative models such as LLMs. The techniques include using a generative model to create augmented training examples to fine-tune a pre-trained generative model. The process begins with original training examples, where each example has a coding prompt and a corresponding gold code solution. Using a documentation dataset containing API or component references, the generative model generates new, enriched training examples having a prompt that instructs the pre-trained generative model to describe the relevant components in addition to performing the code task, and a solution that includes the component descriptions and their roles followed by the original gold code solution. The pre-trained model is then fine-tuned using a combined set of original and augmented training examples, resulting in a more accurate and capable final model.

Abstract: Computer-implemented techniques manage credential data across multiple cloud regions by implementing replication and failover mechanisms. The techniques store different types of credential data, such as authentication tokens and encryption keys, in separate storage systems within a primary cloud region. Changes to credential data trigger replication processes to additional cloud regions with destination regions determined by associated tenant identifiers. During replication, the credential data undergoes transformation to meet region-specific requirements before storage in corresponding systems. The techniques enable failover capabilities when a primary region becomes unavailable, allowing operations to transition to a secondary region including synchronized credential data. This synchronization maintains operational continuity while preserving security and access control policies. The architecture ensures consistent credential management across distributed cloud environments.

Abstract: Techniques for enforcing restrictions for software assurance within a cloud environment are disclosed. A customer section is established within a cloud environment, the customer section including a plurality of cloud resources. One or more restriction policies are caused to be stored within the customer section. One or more operations of the customer section are restricted. In an example, restricting the one or more operations of the customer section includes routing traffic to and from the customer section through a restricting section of the cloud environment. In an example, restricting the one or more operations of the customer section further includes performing the restricting of the one or more operations of the customer section at least in part in accordance with the one or more restriction policies.

Abstract: A cloud infrastructure system is provided for storing cross-tenancy authorization policies for authorizing different users from different tenancies to have different levels of access to bastion functionality that impacts the different tenancies. Stored cross-tenancy authorization policies include, for a first tenancy, policies that authorize a first set of users for bastion service creation and a second set of users for bastion service access, and, for a second tenancy, policies that authorize the second set of users for bastion session creation and the first set of users for deleting a portion of the second tenancy from which bastion sessions may be accessed. Based on the policies, the system authorizes creation of a bastion service that is configured to use a recording destination that is not modifiable by the second user, and then uses the bastion service to create a bastion session for securely accessing resource(s) of the first tenancy.

Abstract: An example method comprises accessing a set of code for which to generate an index for code retrieval, and determining a hierarchy of candidate subsets of the set of code for which to make a determination whether to separately store vector embeddings of the candidate subsets in the index. The method also comprises determining, for a particular subset of the candidate subsets, an ancestor that is closest to the particular subset and for which a determination has been made to separately store a vector embedding in the index. The method also comprises determining a relative amount of content of the ancestor that is shared by the particular subset. The method also comprises generating and storing, in the index for code retrieval from the set of code, a particular vector embedding of the particular subset based at least in part on determining that the relative amount does not satisfy a threshold.

Abstract: Operations include: presenting a Graphical User Interface (GUI) displaying a viewport of a scrollable container associated with a set of interface elements; receiving, by the GUI, user input to initiate a scrolling operation in relation to the scrollable container; responsive to determining that the interface element conversion criterion is met, converting a restricted-scroll interface element to a fully-scrollable interface element; executing the scrolling operation at least by: removing the converted fully-scrollable interface element from the viewport of the scrollable container and scrolling a second fully-scrollable interface element into the viewport.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for selectively excluding information from responses generated by a response generation system are disclosed. In one or more implementations, a system obtains documents encompassing information to be excluded, analyzes this information to identify concepts for exclusion, and generates metadata characterizing these concepts. When processing a query, the system obtains a first response from the response generation system, then analyzes this response using the metadata to identify content that should be excluded. A second response is generated that excludes the identified content. The system evaluates the second response using a language model to determine a score based on the presence of excluded concepts characterized by the metadata. The second response is transmitted if the score meets a specified criterion, ensuring effective exclusion of unwanted information.

Abstract: The illustrative embodiments provide a framework for executing data-centric workflows within the database session, thus achieving workflow execution with no additional process creation, virtualization, or network overhead. The user-provided serverless functions that comprise the application workflow are compiled into native-image binary executables and deployed as stored procedures. Using native-image stored procedures greatly reduces memory and execution time overhead compared to traditional serverless computing frameworks. An application workflow is configured using workflow metadata, which specifies the native-image stored procedures that make up the application workflow and transaction boundaries at runtime. The framework ensures low overhead fault tolerance using run-to-completion and exactly-once semantics.

Abstract: A computer-implemented method includes receiving signals via a network at ports on the network, the signals corresponding to scanning activity at the ports by a plurality of sources on the network; the sources are located at a plurality of geographical bins. The method also includes determining a popularity score for each of the ports, based on a number of geographical bins sending signals to the in a first time period; calculating, for each of the geographical bins, a probability of scanning activity occurring at a port in a second time period, resulting in a plurality of probabilities for that port; and calculating, for each of the ports, a surprisingness index based on the plurality of probabilities. The method further includes estimating, in accordance with the popularity score and the surprisingness index for each of the ports, a likelihood that the port is experiencing suspicious scanning activity.

Abstract: Techniques for monitoring activity during an unlocked state of one or more security policies associated with a security zone of a tenancy of a cloud environment are disclosed. A request to temporarily unlock the one or more security policies is received. The plurality of security policies are applicable to the security zone that encompasses a plurality of cloud resources of the tenancy. The one or more security policies are unlocked. While the one or more security policies are unlocked, a plurality of application programming interface (API) calls destined for the plurality of cloud resources within the security zone are logged. In an example, the logging is performed in response to, or simultaneous with unlocking the one or more security policies, such that prior to unlocking the one or more security policies, API calls destined for the plurality of cloud resources within the security zone are not logged.

Abstract: Techniques for providing recommended attribute value pairs for clustering a set of users are disclosed. The system may provide an administrator with attributes and attribute values prior to executing the clustering. The administrator may select some combinations of attribute value pairs, which the system may then use for execution of the clustering. Other techniques are disclosed for enabling an administrator to apply administrator-defined constraints to a list of recommended actions generated by a machine learning model. In some cases, the recommended actions may be specific to a particular group of users identified by execution of the administrator-informed clustering process.

Abstract: Techniques for creating and executing custom AI-driven actions in a data analysis interface are disclosed. One or more embodiments receive user input that defines a custom action by specifying an AI prompt with a natural language query and context variables. This definition is stored for later use. One or more embodiments present a graphical user interface (GUI) displaying data items, allowing users to select a subset. Based on the stored definition of a custom action, an option to execute the custom action appears. When the option is chosen, one or more embodiments assign metadata values from the selected items to the context variables, creating initialized context variables. The complete prompt, including the query and initialized variables, is submitted to an AI agent. The AI's response is displayed in the GUI, providing context-aware insights within the data analysis workflow.

Abstract: Techniques are provided for updating in-memory property graphs in a fast manner, while minimizing memory consumption. A graph is represented as delta compressed sparse rows (CSR), in which its data structure stores forward edge offsets that map reverse edges to forward edges, enabling fast traversals of graph edges in forward and reverse directions. To support fast graph updates, delta logs are used to store changes to the graph. In an embodiment, a base version of the graph data structure is initially loaded or created, and subsequent versions of the graph are created from the reference to the initial graph and a delta log data structure that records the changes compared to the base version of the graph.

Abstract: Techniques include receiving, from a data store, a first request that comprises a node identifier identifying a node and indicating that a configuration of the node is to be updated to an updated configuration. The techniques include determining whether the configuration is different from the updated configuration. In accordance with determining the configuration is different: requesting the node be cordoned, requesting the node be drained, requesting to replace a boot volume of the node, associating the updated configuration with the node, and requesting the node be uncordoned. The techniques further including transmitting a status indicator to the data store to indicate that the first request was successful.

Abstract: Systems, articles, and computer-implemented methods are disclosed for generating natural language summaries of a multi-dimensional analysis of a detected anomaly within a member of multi-dimensional data by prompting a LLM with a prompt generated to include data about the anomaly in a manner understandable by the LLM. The prompt to the LLM includes a path to a member of the hierarchy containing an anomaly with a delimiter between the member and ancestor nodes. The delimiter allows the ancestral context of the member of the hierarchy to be understood by the LLM. The prompt also includes a metric defining a magnitude of the anomaly in relation to another value, such as an average, a value of the anomaly, a time corresponding to the anomaly, and one or more examples of other anomalies with included data about those anomalies matching the type of data provided for the detected anomaly.

Abstract: Techniques for creating, managing, and using SSH certificates with one or more target-specific principals are disclosed. A certificate authority receives a certificate signing request that includes both a user identifier and a resource identifier. The user identifier identifies a user, and the resource identifier represents one or more target hosts. The certificate authority forms a target-specific principal for use in creating the certificate. The target-specific principal indicates both the user and the resource identifier representing the resource(s) for which access is requested. The resource identifier may represent a host class associated with more than one host. Once the certificate authority verifies that the user is entitled to access the requested resource(s), it generates the certificate, signs it, and returns it to the requesting device.

Abstract: Techniques are described for creating a network-link between a first virtual network in a first cloud environment and a second virtual network in a second cloud environment. The first virtual network in the first cloud environment is created to enable a user associated with a customer tenancy in the second cloud environment to access one or more services provided in the first cloud environment. The network-link is created based on one or more link-enabling virtual networks being deployed in the first cloud environment and the second cloud environment.

Abstract: Techniques are described for failing over from a primary database to the replicated logical database of the primary database. Techniques are also described for the client-side object references to be preserved when failing over to the logical replica database, for AS OF and other queries preserved, and for versioning of checksums, signatures and structures across logical replicas.

Abstract: Techniques are described for providing session management functionalities using an access token (e.g., an Open Authorization (OAuth) access token). Upon successful user authentication, a session (e.g., a single sign-on session) is created for the user along with a user identity token that includes information identifying the session. The user identity token is presentable in an access token request sent to an access token issuer authority (e.g., an OAuth server). Upon receiving the access token request, the user identity token is parsed to identify and validate the session against information stored for the session. The validation can include various session management-related checks. If the validation is successful, the token issuer authority generates the access token. In this manner, the access token that is generated is linked to the session. The access token can then be used by an application to gain access to a protected resource.

Abstract: Disclosed herein are techniques for detecting invalid web traffic. A method described herein involves obtaining an input sequence. The method further includes inputting the input sequence to an encoder model configured to generate an embedding from the input sequence. The method includes generating, as an output of the encoder model, an embedding. The method includes inputting the embedding to a classifier model configured to predict a vector of probability values corresponding to a plurality of classes using the embedding. A first class of the plurality of classes may be attributed to variable traffic and a second class of the plurality of classes may be attributed to repetitive traffic. The method includes predicting, as an output of the classifier, the vector of probability values. The method includes determining a machine-actor score from the vector of probability values. The method also includes outputting the machine-actor score.