Abstract: The disclosure herein describes platform-level migration for deep learning training (DLT) jobs from a checkpointed stated between a source node and a destination node. The checkpointing is performed through capturing GPU state (e.g., device state) and CPU state (e.g., host state). The GPU state includes GPU data (e.g., model parameters, optimizer state, etc.) that is located in the GPU and GPU context (e.g., the default stream in GPU, various handles created by libraries). Restoring the DLT job on the destination node involves resumption of processing of a destination GPU at the same checkpointed state.

Abstract: A network security device has at least one Fully Qualified Domain Name (FQDN) access policy that permits traffic to flow to at least one resource associated with at least one FQDN. The network security device receives, from a managed endpoint device, a packet directed to the at least one resource associated with the at least one FQDN. The network security device obtains DNS information associated with the managed endpoint device and, based on the domain name system (DNS) information, substitutes a network address of the at least one resource into the at least one FQDN access policy to open a traffic flow to the at least one resource associated with the at least one FQDN. The network security device then provides the packet to the at least one resource associated with the at least one FQDN.

Abstract: Systems and methods described herein can relate to a web browser-based application programming interface (API) that can control the authorization of data transmissions within a networked environment. The system enables content items displayed in a web page to make a call to the browser to request that the browser register the content item with the browser. The web browser can control the network transmissions to content items approved or authorized by the web browser. The system can reduce the number of transmissions that do not result in content for the client device—saving bandwidth and computational resources for the client device.

Abstract: Embodiments of the present disclosure provide improved data observability mechanisms. Specifically, embodiments provide for managing event-structured observability data in a federated network and database platform. The improved mechanisms enable improved analysis of data-driven errors and/or storage of associated data for purposes of data observability processing actions and/or rendering for user analysis. Example embodiments are configured for receiving a data stream representing operational engagement of an event-structured service hosted by the federated network and database platform. Some example embodiments are further configured for generating event-structured observability data from the data stream utilizing an event processing pipeline. The example embodiments are further configured for storing the event-structured observability data in at least one event-structured observability data repository of the federated network and database platform.

Abstract: Some embodiments provide a method for a network controller that manages several logical networks. The method receives a specification of a logical network that includes at least one logical forwarding element attached to a logical service (e.g., DHCP). The method selects at least one host machine to host the specified logical service from several host machines designated for hosting logical services. The method generates logical service configuration information for distribution to the selected host machine. In some embodiments, the method selects a master host machine and a backup host machine for hosting logical service. In some embodiments, a particular one of the designated host machines hosts at least two DHCP services for two different logical networks as separate processes operating on the particular host machine.

Abstract: Techniques for Qprints using telemetry-based similarity for DNS are provided. In some embodiments, a system/process/computer program product for Qprints using telemetry-based similarity for DNS in accordance with some embodiments includes aggregating a set of network related event data, wherein the set of network related event data includes Domain Name System (DNS) related query data; clustering the DNS related query data; and generating similarity clusters for domains based on their DNS related query data. For example, the set of network related event data can include passive DNS (pDNS) data aggregated over a period of time to express pDNS data at-scale, and similarity of the pDNS data aggregated over the period of time is quantified, within and across networks based on telemetry-based similarity for DNS using a statistical model.

Abstract: A distributed computer system including a story application for a mobile computing device enables presentation of media using multiple different real or virtual social media apps executed using the mobile computing device. Under control of a programmed story script, and in response to input signals to advance a media presentation, the story application selectively invokes a particular social media app, from among a plurality of different social media apps installed on the mobile device and instructs the particular social media app to present or play a particular story element of the media presentation. Successive story elements are presented serially in response to successive input signals, without interruption of other apps that may be running on the mobile device. In this manner, media presentations such as entertainment programs or stories may be presented using multiple different social media apps rather than via passive video playback.

Abstract: Described are various embodiments of a device and method for computing statistics of various network monitoring metrics. In one embodiment, the method includes constructing a first histogram of network traffic monitoring data acquired over a designated sampling period from said network; identifying one or more bins of said first histogram comprising each at least one of one or more desired percentile values; for each identified bin: building a second histogram centered on said identified bin, said second histogram comprising a second bin size that is smaller than said first bin size; calculating one or more bins of said second histogram comprising each at least one of said one or more desired percentile values and the values associated therewith; and converting said values associated therewith into percentile values representative of the range defined between said lower order of magnitude to said higher order of magnitude.

Abstract: Disclosed is a new approach for client-side contextual engagement with a website visitor. A browser loads a page containing a reference to a script file implementing a visitor client. The visitor client may store visitor client data including site rules and a visitor profile locally on the client device. The visitor client may include a context monitor for calling the decision engine, a decision engine for evaluating the site rules relative to a context defined at least partially by events occurring on the page and the visitor profile, and a widget manager for managing visitor client data persisted on the client device. The content monitor may call the decision engine when the context changes. The decision engine may determine, in real time and relative to the context, that proactive or reactive engagement(s) may be appropriate and call the widget manager to launch and display corresponding engagement channel(s) on the page.

Abstract: A method includes determining a set of test locations based on a network parameter; and determining a subset of the test locations that do not have a point test equipment. The method includes determining a drive test route to cover an area serviced by each the subset of the test locations. The method includes receiving a notification from a drive test equipment in response to the drive test equipment arriving at a first test location on the drive test route. The method includes sending test parameters to the drive test equipment; and receiving a drive test result from the drive test equipment. The method includes determining whether the drive test results meet a drive test specification. The method includes requesting the drive test equipment to relocate to a second test location in response to a determination that the drive test results meet the drive test specification.

Abstract: A processing system may obtain a first sampled flow record for a first flow in a network, comprising information regarding selected packets of the first flow, derive, from the first sampled flow record, a data volume and a duration of the first flow, and determine a first flow metric for the first flow that is calculated from the data volume and the duration, where the first flow metric is one of a plurality of flow metrics for a plurality of flows, and where the plurality of flow metrics is determined from the plurality of sampled flow records associated with the plurality of flows. The processing system may then classify the first flow into one of at least two classes, based upon the first flow metric and at least a first flow metric threshold.

Abstract: A message is received, that represents an action to be performed in a service. The message includes a natural language portion and an endpoint to interact with. The natural language portion is parsed to identify the action and the endpoint is interacted with to perform the action.

Abstract: The described technology is generally directed towards latency diagnostics for multiparty systems, such as a system comprising a communication network, a content delivery network (CDN), and one or more internet service providers (ISPs). A latency analyzer component can process latency data in response to a latency diagnostic trigger, such as an alert from a video quality monitoring system. The latency analyzer can determine whether latency is attributable to the communication network. If not, the latency analyzer can determine whether the latency is attributable to the CDN. If the latency is not attributable to the communication network or the CDN, the latency analyzer can determine that the latency is attributable to the ISP, and the latency analyzer can identify the ISP and generate appropriate reports and notifications.

Abstract: Disclosed is a first device and a second device each sending BFD echo request packets in an initial stage of establishing communication between the two devices. A method can include determining that a certain mode is detected, such as a low bandwidth mode. The method can be practiced by one or both of the devices or a separate network controller. Based on the detection of the certain mode or the communication link being in a certain mode, the method includes the first device or the second device electing to be a master or a slave. Upon establishing that one of the devices is the master and the other of the devices is a slave, only the master will send BFD echo request packets.

Abstract: A verified method of high-value, person-to-person communication is provided. The method comprises creating a unique dynamic messaging link recognizable to a selected intended recipient; creating a transmissible personal hypermedia message to which the link is assigned; and storing the hypermedia message on a device accessible to a network. The link comprises a recognizable proprietary domain name and a random hash code. The individual is sent a direct message and the device is monitored for transmission thereto of the link, analyzing and storing accompanying metadata. The direct message contains an invitation to click on the link over the imprimatur of a person known to the recipient. The invitation contains descriptive material relating to the hypermedia message. Upon activation of the link, without redirection, the hypermedia message is transmitted to the recipient and the sender is notified. The sender is alerted if the link has not been activated within a predetermined period.

Abstract: Technology for automatic cross-device meeting authentication. An example method involving initializing, by a first computing device, a real-time meeting communication session, wherein the first computing device is connected to a media system; receiving information indicative of a mobile computing device of a user being in a physical presence of the first computing device, wherein the user is associated with the mobile device and a second computing device; granting control of the real-time meeting communication session from the mobile computing device of the user to the first computing device; transmitting an instruction that causes the second computing device to display a user interface element associated with the real-time meeting communication session, wherein the instruction is sent after granting the control; and granting control of the real-time meeting communication session to the second computing device of the user in response to a selection of the user interface element.

Abstract: Systems and methods are described for applying a plurality of data points of a time series data set representing values of a metric measuring performance of a cloud computing service to a machine learning model to predict a forecast of a most likely value of the metric at a selected future time. The method includes determining whether the plurality of data points of the time series data set are anomalies according to the machine learning model and the forecast and generating a collective anomaly from the anomalies when the plurality of data points is determined to be anomalies. The method further includes determining whether the collective anomaly does not meet one or more cloud computing service level objective (SLO) threshold requirements and sending an alert when the collective anomaly does not meet one or more cloud computing SLO threshold requirements.

Abstract: A system and method for multivariate testing of messages to a subgroup in a one-to-many messaging platform. A client text message is generated for transmission to a number of users via one or more messaging services. A subset of users is defined according to one or more attributes of the text message or the users, and the client text message is transmitted only to users in the subgroup. The transmission is analyzed for performance metrics, such as actions or reactions by users in the subgroup, and based on the performance metrics, the message is optimized for transmission to the larger group of users. Optimization happens rapidly.

Abstract: A method for enabling a performance measurement on packet flow transmitted through a communication network. A marking value is periodically switched in the packets with a marking period Tm. The packet flow is then divided into blocks of duration Ts (synchronization period). Each synchronization period comprises an integer number of marking periods. Two or more measurement points on the path of the packet flow may provide a performance parameter for each marking period and associate thereto a synchronization information generated based on their local clocks and relating to the synchronization period containing the marking period to which the performance parameter relates; and a sequence information indicating the marking period's position within the synchronization period. A management server may identify performance parameters provided by different measurement points and relating to a same marking period based on the synchronization information and the sequence information.

Abstract: Some embodiments provide a novel method for assessing the suitability of network links for connecting compute nodes located at different geographic sites. The method of some embodiments identifies and analyzes sample packets from a set of flows exchanged between first and second compute sites that are connected through a first network link in order to identify attributes of the sampled packets. The method also computes attributes of predicted packets between the identified samples in order to identify attributes of each flow in the set of flows. The method then uses the identified and computed attributes of each flow in the set of flows to emulate the set of flows passing between the two compute sites through the second network link in order to assess whether a second network link should be used for future flows (e.g., future flows exchanged between the first and second compute sites).