Machine learning language based dynamic network discovery

1 . A computer-implemented method in a network device manager for using machine learning to dynamically set TTL (Time-To-Live) values for connection events to a plurality of devices managed by the network device manager over a data communication network, the method comprising:

generating an adaptive TTL model from connection events, based on varying flight delay times for connecting the device manager to a plurality of managed devices, including:

collecting training data by detecting successful discoveries of any of the plurality of managed devices at various locations on the data communication network, and storing TTL values associated with the successful discoveries at various locations in a TTL database, wherein a TTL value is set to account for a flight delay time and a log-in time prior to expiration of a data packet, the flight delay time comprising a period for data packets to traverse over the data communication network between the device manager and any of the plurality of managed devices and a period for logging in of the network device manager to any of the plurality of managed device;

collecting training data by detecting failed discoveries of any of the plurality of managed devices at various locations, and detecting iterations of a TTL value needed to reach a successful discovery, and storing the TTL value iterations at various locations in the TTL database;

responsive to the successful discoveries, determining a flight time portion for the successful discovery using a ping or a traceroute, wherein the ping or the traceroute includes the flight delay time portion of successful discoveries without a login portion of successful discoveries, and storing the flight delay time portion in the TTL database; and

using machine learning to predict TTL values needed for connection events over various flight delay times at various locations; and

during a connection event for any of the plurality of managed devices, automatically updating a TTL value for the connection event from the adaptive TTL model, including:

determining a current flight delay time between the device manager and a specific managed device

determining a specific location of the specific managed device; and

setting a TTL value for the connection event from predicted TTL values, using the current flight delay time and the specific location.

2 . The method of claim 1 , further comprising:

setting a maximum TTL for the TTL iterations; and

responsive to exceeding the maximum TTL, labeling a managed device of the plurality of managed devices as unreachable.

3 . The method of claim 1 , wherein the TTL value is set in real-time with the connection event.

4 . The method of claim 1 , wherein the TTL value is set periodically prior to the connection event.

5 . The method of claim 1 , wherein the flight delay time comprises half of a time of a round-trip for the ping or the traceroute.

6 . The method of claim 1 , wherein the log-in portion of the TTL value is constant and the flight time portion is variable.

7 . The method of claim 1 , wherein the log-in portion of the TTL value is variable.

8 . The method of claim 1 , wherein the connection event comprises one of an initial discovery connection, a reconnection, and a keep alive connection.

9 . The method of claim 1 , wherein at least one of the plurality of managed devices is located remotely from an enterprise network including the device manager.

10 . The method of claim 1 , wherein at least one of the plurality of managed devices is located on a different continent from the device manager.

11 . The method of claim 1 , wherein at least one of the plurality of managed devices is located within a LAN, and the device manager is also located on the LAN.

12 . A non-transitory computer-readable medium in a network device manager for using machine learning to dynamically set TTL (Time-To-Live) values for connection events to a plurality of devices managed by the network device manager over a data communication network, the method comprising:

generating an adaptive TTL model from connection events, based on varying flight delay times for connecting the device manager to a plurality of managed devices, including:

collecting training data by detecting successful discoveries of any of the plurality of managed devices at various locations on the data communication network, and storing TTL values associated with the successful discoveries at various locations in a TTL database, wherein a TTL value is set to account for a flight delay time and a log-in time prior to expiration of a data packet, the flight delay time comprising a period for data packets to traverse over the data communication network between the device manager and any of the plurality of managed devices and a period for logging in of the network device manager to any of the plurality of managed device;

collecting training data by detecting failed discoveries of any of the plurality of managed devices at various locations, and detecting iterations of a TTL value needed to reach a successful discovery, and storing the TTL value iterations at various locations in the TTL database;

responsive to the successful discoveries, determining a flight time portion for the successful discovery using a ping or a traceroute, wherein the ping or the traceroute includes the flight delay time portion of successful discoveries without a login portion of successful discoveries, and storing the flight delay time portion in the TTL database; and

using machine learning to predict TTL values needed for connection events over various flight delay times at various locations; and

during a connection event for any of the plurality of managed devices, automatically updating a TTL value for the connection event from the adaptive TTL model, including:

determining a current flight delay time between the device manager and a specific managed device

determining a specific location of the specific managed device; and

setting a TTL value for the connection event from predicted TTL values, using the current flight delay time and the specific location.

13 . A network device manager for using machine learning to dynamically set TTL (Time-To-Live) values for connection events to a plurality of devices managed by the network device manager over a data communication network, the network device manager comprising:

a processor;

a network interface communicatively coupled to the processor and to the WLAN; and

a memory, communicatively coupled to the processor and storing:

a TTL modeling module, to generate an adaptive TTL model from connection events, based on varying flight delay times for connecting the device manager to a plurality of managed devices, including:

a training module to collect training data by detecting successful discoveries of any of the plurality of managed devices at various locations on the data communication network, and storing TTL values associated with the successful discoveries at various locations in a TTL database, wherein a TTL value is set to account for a flight delay time and a log-in time prior to expiration of a data packet, the flight delay time comprising a period for data packets to traverse over the data communication network between the device manager and any of the plurality of managed devices and a period for logging in of the network device manager to any of the plurality of managed device,

wherein the training module collects training data by detecting failed discoveries of any of the plurality of managed devices at various locations, and detecting iterations of a TTL value needed to reach a successful discovery, and storing the TTL value iterations at various locations in the TTL database;

a flight time module to, responsive to the successful discoveries, determine a flight time portion for the successful discovery using a ping or a traceroute, wherein the ping or the traceroute includes the flight delay time portion of successful discoveries without a login portion of successful discoveries, and storing the flight delay time portion in the TTL database;

a machine learning module to, use machine learning for predicting TTL values needed for connection events over various flight delay times at various locations; and

a TTL updating module, during a connection event for any of the plurality of managed devices, automatically updates a TTL value for the connection event from the adaptive TTL model, by determining a current flight delay time between the device manager and a specific managed device, determining a specific location of the specific managed device, and setting a TTL value for the connection event from predicted TTL values, using the current flight delay time and the specific location.