Amazon Technologies, Inc.

Zhuo Zhang

Muhammad Ali Siddiqui

Swaminathan Subramanian

Joshua M Goodspeed

Mark Evans Brighton

Shane Anil Pereira

Sameer R Rajyaguru

Charles Drummond Swan

Ivan Eduardo Gonzalez

Patrick Ludvig Bajari

Harry J Paarsch

Stephen Chen

Devesh R Raval

Brent E Edwards

Systems for providing a multi-tenant threat intelligence service are provided. The system receives threat information from a user including IP addresses, and universal threat information including IP addresses. Modify an in-memory IP address tree using IP addresses received from the user and included in the universal threat information. Compare IP addresses from logs of network activity associated with the user to the in-memory IP address tree, and identify IP addresses included in the IP address tree. Cause matching IP addresses to be sent to the user as representing potentially malicious network activity.

Multi-tenant threat intelligence service

Techniques are described for performing automated predictions of program execution capacity or other capacity of computing-related hardware resources that will be used to execute software programs in the future, such as for a group of computing nodes that execute one or more programs for a user. The predictions that are performed may in at least some situations be based on historical data regarding corresponding prior actual usage of execution-related capacity (e.g., for one or more prior years), and may include long-term predictions for particular future time periods that are multiple months or years into the future. In addition, the predictions of the execution-related capacity for particular future time periods may be used in various manners, including to manage execution-related capacity at or before those future time periods, such as to prepare sufficient execution-related capacity to be available at those future time periods.

Predicting long-term computing resource usage

Techniques are described for managing program execution capacity or other capacity of computing-related hardware resources used to execute software programs, such as for a group of computing nodes that is in use executing one or more programs for a user. Dynamic modifications to the program execution capacity of the group may include adding or removing computing nodes, such as in response to automated determinations that previously specified triggers are currently satisfied, and may be automatically governed at particular times based on automatically generated predictions of program execution capacity that will be used at those times by the group, such as to verify that requested dynamic execution capacity modifications at a time are within the predicted execution capacity values for that time. In some situations, the techniques are used in conjunction with a fee-based program execution service that executes multiple programs on behalf of multiple users of the service.

Predictive governing of dynamic modification of program execution capacity

This disclosure is directed to measuring test effects using adjusted outlier data. Test data and control data may include some outlier data (i.e., right-side tails of distribution curves), which may bias the resultant data. The outlier data may be adjusted to reduce bias. A cutoff point is selected along the distribution of data. Data below the cutoff is maintained and used to determine an effect of the data below the cutoff point. The effect of the data above the cutoff may be processed as follows. Predictor data is identified from the data below, but near, the cutoff point. The predictor data may then be used determine the effect of the outlier data that is above the cutoff point. In some embodiments, the predictor data may be weighted and combined with a weighted portion of the outlier data to determine an effect of the data above the cutoff point.

Measuring test effects using adjusted outlier data

Advertising revenue may be optimized by selecting an advertisement to display to a user based at least in part on the user's browsing history and a revenue model associated with the advertisement. For example, if the user has a relatively high clicking frequency on advertisements, then pay-per-click advertisements may be selected for presentation to the user rather than pay-per-impression advertisements. Revenue model-based advertisement selection may be combined with other advertising techniques, such as targeting.

Selecting advertisements based on advertising revenue model

Growing community of inventors

Mercer Island, WA, United States of America

Zhuo Zhang