Automatically recommending point of presence centers

Patent metadata

Field	Value
Publication number	US-9900215-B2
Application number	US-201414540234-A
Country	US
Kind code	B2
Filing date	Nov 13, 2014
Priority date	Nov 13, 2014
Publication date	Feb 20, 2018
Grant date	Feb 20, 2018

Abstract

Techniques for automatically recommending a new data center, such as a point-of-presence (POP) center are provided. In one technique, multiple candidate locations for a new POP center are considered. An impact of adding a new POP to each candidate location is estimated and a score is generated. Each candidate location is ranked based on the score. In a related technique, an impact score for a candidate location is based on a prediction of whether and how much a new POP center at the candidate location would reduce response times of clients that would connect to the new POP center. The prediction may be based on a model that is generated based on response data generated by clients that are connecting to existing POP centers. The model may take into account multiple factors, such as geographic distance, network distance, type of browser executing on the clients, type of operating system, etc.

First claim

What is claimed is:

1. A method comprising:
receiving, from a plurality of client devices, response data that indicates one or more measures of responsiveness of a plurality of existing data centers;
based on the response data, generating a prediction model that is used to predict, for each possible location of multiple possible locations, an estimated time reduction if a data center is deployed at said each possible location;
generating, based on the prediction model and a plurality of candidate locations, a plurality of estimated time reductions;
wherein generating the plurality of estimated time reductions comprises, for each candidate location of the plurality of candidate locations:
for each client device of the plurality of client devices:
determining, based on the prediction model, an estimated response time associated with said each client device if a data center is deployed at said each candidate location,
determining an estimated time reduction for said each client device based on a difference between the estimated response time and an actual response time associated with said each client device with respect to an existing data center;

selecting, based on the plurality of estimated time reductions, a particular location of the plurality of candidate locations to recommend as a location in which to add a new data center;
causing the particular location to be presented on a computing device for display;
wherein the method is performed by one or more computing devices.

2. The method of claim 1 , wherein the response data indicates one or more of a connection time, a first byte time, or a page download time.

3. The method of claim 1 , wherein:
selecting comprises, for each candidate location of the plurality of candidate locations, generating an impact score of adding a data center at said each candidate location;
selecting comprises selecting based on the impact score of said each candidate location.

4. The method of claim 3 , wherein generating the impact score comprises:
generating the impact score based on a percentage of users that would use a data center at said each candidate location and an estimated time reduction if the data center is deployed at said each candidate location.

5. The method of claim 3 , wherein generating the impact score comprises generating an impact score for candidate locations where a percentage of client devices or instances of user access, in a geographical region, that would benefit from a data center deployed in said candidate location exceeds a particular threshold.

6. The method of claim 1 , wherein each estimated time reduction in the plurality of estimated time reductions pertains to one or more of a connection time, a first byte time, or a page download time.

7. The method of claim 1 , wherein generating the prediction model comprises generating the prediction model based on multiple features, including one or more of the following features:
geographical distance between the plurality of client devices and a first subset of the plurality of existing data centers;
autonomous system numbers (ASN) of the plurality of client devices and a second subset of the plurality of existing data centers; or
network distance between the plurality of client devices and a third subset of the plurality of existing data centers.

8. The method of claim 1 , wherein generating the prediction model comprises generating the prediction model based on one or more of the following features:
type of operating system of the plurality of client devices;
type of browser installed on the plurality of client devices;
data center that services the plurality of client devices;
country or geographic region in which each of the plurality of client devices resides;
time or date of one or more visits by the plurality of client devices;
time or date of registration of the plurality of client devices with a particular website.

9. The method of claim 1 , further comprising:
prior to generating the prediction model, applying a transformation to a plurality of values, indicated in the response data, to generate a plurality of transformed values, wherein the plurality of values correspond to one or more features;
wherein generating the prediction model comprises generating the prediction model based on the plurality of transformed values.

10. The method of claim 9 , further comprising:
after the generating the prediction model, applying the transformation to a second plurality of values to generate a second plurality of transformed values;
for each transformed value of the second plurality, inputting said each transformed value into the prediction model to predict a particular estimated time reduction for a client device that corresponds to said each transformed value.

11. The method of claim 1 , wherein the plurality of existing data centers is a plurality of point-of-presence (POP) centers.

12. The method of claim 1 , wherein the prediction model is generated using regression analysis.

13. The method of claim 1 , wherein:
a first subset of the plurality of estimated reduction times correspond to a first candidate location in the plurality of candidate locations;
a second subset of the plurality of estimated reduction times correspond to a second candidate location in the plurality of candidate locations;
the method further comprising:
generating a first score based on the first subset;
generating a second score based on the second subset;

selecting the particular location is based on the first score and the second score.

14. A method comprising:
receiving, from a plurality of client devices, response data that indicates one or more measures of responsiveness of a plurality of existing data centers;
storing region data that indicates a plurality of regions;
storing region association data that associates, for each client device of the plurality of client devices, said each client device with one of the plurality of regions;
for each candidate location of a plurality of candidate locations:
for each region in a subset of the plurality of regions:
identifying, based on the region association data, a first number of client devices that are associated with said each region;
for each client device in the first number of client devices, identifying a predicted response time if said each candidate location is selected as a next location in which to deploy a data center;
identifying a second number of client devices that are associated with said each region and that are associated with a predicted response time that is less than an actual response time indicated in the response data;

selecting, based on the second number of client devices identified for each region, a particular location of the plurality of candidate locations to recommend as a location in which to add a new data center;
causing the particular location to be presented on a display of a computing device.

15. A system comprising:
one or more processors;
one or more storage media storing instructions which, when executed by the one or more computing devices, cause:
receiving, from a plurality of client devices, response data that indicates one or more measures of responsiveness of a plurality of existing data centers;
based on the response data, generating a prediction model that is used to predict, for each possible location of multiple possible locations, an estimated time reduction if a data center is deployed at said each possible location;
generating, based on the prediction model and a plurality of candidate locations, a plurality of estimated time reductions;
wherein generating

Assignees

• Linkedin Corp
• Microsoft Technology Licensing Llc

Inventors

• Yang Yang
• Zhang Liang
• Kahn Zaid A
• Maheshwari Ritesh
• Dubey Sanjay S
• Shieh James C

Classifications

• H04L41/147

  for predicting network behaviour · CPC title

• H04L67/18

  Electricity · mapped topic

• H04L41/12

  Discovery or management of network topologies · CPC title

• H04L41/145

  involving simulating, designing, planning or modelling of a network · CPC title

• H04L41/0823 Primary

  characterised by the purposes of a change of settings, e.g. optimising configuration for enhancing reliability (for optimising operational conditions of wireless networks H04W24/02) · CPC title