Method and system for calculating latency in a data flow

1. A system for calculating latency in a distributed data flow environment, the system comprising:

a. a plurality of hop servers; and

b. a central registry server;

wherein each hop server of the plurality of hop servers is configured to:

a. receive a plurality of data packets across multiple time windows, wherein each of the data packets of the plurality of data packets includes a header;

b. bucket each of the data packets of the plurality of data packets received across the time window into multiple buckets according to the header, wherein at least two of the buckets correspond to time windows representing different periods of time and referenced to a common originating data packet origin time; and

c. transmit a count of the plurality of the data packets in the buckets and window identifiers associated with the time windows to the central registry server; and

wherein the central registry server is configured to:

a. receive a plurality of counts of the data packets in each of the one or more buckets and a plurality of the window identifiers associated with the time windows corresponding to the buckets to the central registry from the plurality of hop servers;

b. receive a latency query; and

c. calculate latency based on the plurality of counts for each bucket and the window identifier, the time window, and referenced sending time for each bucket.

2. The system of claim 1 , wherein the plurality of hop servers further comprises one or more data agents, each data agent of the one or more data agents is configured to parse the header of the each data packet of the plurality of data packets.

3. The system of claim 2 , wherein the header of each data packet of the plurality of data packets comprises a timestamp at which the each data packet of the plurality of data packets originated.

4. The method of claim 1 , wherein to calculate latency based on the plurality of counts received by a first one of the hop servers from a second one of the hop servers and the plurality of window identifiers comprises to determine a weighted average of maximum latency times for each count based window identifier, the time window, and referenced sending time for each bucket.

5. The method of claim 1 , wherein at least one of the hop servers is an advertisement server.

6. A method of transmitting data packet information to calculate latency over a distributed data flow environment, the method comprising:

a. receiving a plurality of data packets across a time window, each of the data packet of the plurality of data packets comprising a header, wherein the header of each data packet of the plurality of data packets includes a timestamp representing an origin time at which each of the data packets originated;

b. bucketing each of the data packets of the plurality of data packets, received across the time window, into a plurality of buckets according to the header wherein at least two of the buckets correspond to respective time windows indexed to an originating data packet sending time; and

c. transmitting counts of the plurality of the data packets in the buckets and identifiers of the time windows to a central registry server to calculate latency based on the plurality of counts for each bucket and the plurality of window identifiers, the timestamp, and the time window for each bucket.

7. The method of claim 6 , the header of each data packet of the plurality of data packets comprises a timestamp at which the each data packet of the plurality of data packets originated.

8. The method of claim 6 , wherein to calculate latency based on the plurality of counts received by a first one of the hop servers from a second one of the hop servers and the plurality of window identifiers comprises to determine a weighted average of maximum latency times for each count based window identifier, the time window, and referenced sending time for each bucket.

9. The method of claim 6 , wherein data flow environment comprises multiple hop servers coupled together, and at least one of the hop servers is an advertisement server.

10. A method of calculating latency in a distributed data flow environment, the method comprising:

a. receiving a plurality of counts and a plurality of identifiers of time windows from a plurality of hop servers, wherein the plurality of counts indicates plurality of data packets in multiple buckets, and wherein each of the buckets comprises a plurality of data packets grouped together based on a header of each of the data packet from the plurality of the data packets, wherein the header of each data packet of the plurality of data packets includes a timestamp representing a time at which each of the data packets originated and at least two of the buckets correspond to respective time windows representing different periods of time and referenced to a common originating data packet origin time;

b. receiving a latency query; and

c. calculating latency based on the buckets and the identifier of the time windows, the referenced origin time, and the time windows for each bucket.

11. The method of claim 10 , wherein the header of each data packet of the plurality of data packets comprises a timestamp at which the each data packet of the plurality of data packets originated.

12. The method of claim 10 , wherein calculating latency based on the plurality of counts received by a first one of the hop servers from a second one of the hop servers and the plurality of window identifiers comprises to determine a weighted average of maximum latency times for each count based window identifier, the time window, and referenced sending time for each bucket.

13. The method of claim 10 , wherein at least one of the hop servers is an advertisement server.