System and methods for accelerating data delivery

• US 20040199650A1
• Filed: 11/14/2003
• Published: 10/07/2004
• Est. Priority Date: 11/14/2002
• Status: Abandoned Application

• Current Assignee: Kasenna, Inc. (Enghouse Systems Limited)
• Original Assignee: Kasenna, Inc. (Enghouse Systems Limited)
• Inventors: Menon, Satish N., Howe, John E., Nava, Alberto, Muthukumarasamy, Jayakumar
• Application Number:
  US10/713,656
• Publication Number:
  US 20040199650A1
• Time in Patent Office: Days
• US Class Current: 709/230
• CPC Class Codes: H04L 65/1101 Session protocols
• CPC Class Codes: H04L 65/80 Responding to QoS
• CPC Class Codes: H04L 69/16 Implementation or adaptation of Internet protocol [IP], of transmission control protocol [TCP] or of user datagram protocol [UDP]
• CPC Class Codes: H04L 69/161 Implementation details of TCP/IP or UDP/IP stack architecture; Specification of modified or new header fields
• CPC Class Codes: H04L 69/162 involving adaptations of sockets based mechanisms secure socket layer H04L63/168

Abstract

Embodiments of the present invention provide for sending data over a generic network interface in such a way that the resources used by the host CPU for packetization is minimized, allowing more packets to be sent across the network than is currently possible using existing technologies. Data is sent in a timely manner, minimizing jitter associated with real-time data delivery, especially for multimedia applications. CPU resource utilization is achieved by using a number of different techniques, such as zero-copy data movement where the data from a storage system is made available directly to the network driver without creating intermediate copies that consume CPU resources to do the copying.

Claims

1. A method for reducing processor cycles required to send data over a communication link in packets having a packet size, the method comprising:

sending a write call comprising a destination and a quantity of data greater than said packet size to a driver through a socket;

performing a zero-copy write of said quantity of data to said driver; and

generating a plurality of packets less than or equal to said packet size.

2. A method according to claim 1, further comprising

grouping data from a plurality of streams into said write call.

3. A method according to claim 2, wherein said grouped data comprises all data from said plurality of streams to be sent in a time interval.

4. A method according to claim 3, wherein the time interval is selected based on a bit rate of at least one stream.

5. A method according to claim 1, wherein said zero-copy write comprises writing data to a translation buffer.

6. A method according to claim 5, wherein the translation buffer is setup with a translation between virtual memory and physical memory.

7. A method according to claim 1, further comprising generating an interrupt only after a last packet of said plurality of packets is transmitted to said communication link.

8. A method according to claim 1, wherein said packet size is a maximum packet size allowable by the communication link.

9. A method according to claim 1, wherein the communication link comprises a network.

10. A method according to claim 1, wherein said write call further comprises a plurality of destinations.

11. A method according to claim 1, wherein said quantity of data comprises at least a portion of a multimedia data file.

12. A method according to claim 11 wherein the multimedia data file requires real-time delivery.

13. A method according to claim 11, wherein the multimedia data file is a video file, an audio file, or a game file.

14. A method according to claim 1, wherein said quantity of data comprises at least a portion of a file having a format chosen from the group of formats consisting of MPEG-1, MPEG-2, MPEG-4, H.264, MP3, QuickTime, AVI, Audio/Video, real-time data in RTP format, and combinations thereof.

15. A method according to claim 1, further comprising:

communicating at least one of said packets to a network interface card.

16. A method according to claim 1, further comprising:

storing said quantity of data in memory registered with said driver.

17. A method according to claim 1, wherein said write call comprises a write vector.

18. A method according to claim 1, further comprising generating a single header comprising header information for a plurality of protocol layers and sending the single header to a queue for a NIC.

19. A computer program product for sending data over a communications link in packets having a packet size, the computer program product comprising:

a computer-readable medium comprising a program module, the program module including instructions for;

receiving a write call comprising a destination and a quantity of data greater than said packet size through a socket;

performing a zero-copy write of said quantity of data; and

generating a plurality of packets less than or equal to said packet size.

20. A computer program product according to claim 19, further comprising instructions for:

grouping data from a plurality of streams into said write call.

21. A computer program product according to claim 20, wherein said grouped data comprises all data from said plurality of streams to be sent in a time interval.

22. A computer program product according to claim 21, wherein the time interval is selected based on a bit rate of at least one stream.

23. A computer program product according to claim 19, wherein said zero-copy write comprises writing data to a translation buffer.

24. A computer program product according to claim 23, wherein the translation buffer is setup with a translation between virtual memory and physical memory.

25. A computer program product according to claim 19, further comprising generating an interrupt only after a last packet of said plurality of packets is transmitted to said communication link.

26. A computer program product according to claim 19, wherein said packet size is a maximum packet size allowable by the communication link.

27. A computer program product according to claim 19, wherein said write call further comprises a plurality of destinations.

28. A computer program product according to claim 19, wherein said quantity of data comprises at least a portion of a multimedia data file.

29. A computer program product according to claim 28 wherein the multimedia data file requires real-time delivery.

30. A computer program product according to claim 28, wherein the multimedia data file is a video file, an audio file, or a game file.

31. A computer program product according to claim 19, wherein said quantity of data comprises at least a portion of a file having a format chosen from the group of formats consisting of MPEG-1, MPEG-2, MPEG-4, H.264, MP3, QuickTime, AVI, Audio/Video, real-time data in RTP format, and combinations thereof.

32. A computer program product according to claim 19, further comprising:

communicating at least one of said packets to a network interface card.

33. A computer program product according to claim 19, further comprising:

storing said quantity of data in memory.

34. A computer program product according to claim 19, wherein said write call comprises a write vector.

35. A method for reducing buffering requirements on a network switch;

the method comprising;

receiving a write call from an application through a socket, the write call comprising a plurality of destinations, including a first destination and a second destination, and a first quantity of data destined for the first destination, and a second quantity of data destined for the second destination;

packetizing said first quantity of data into a plurality of packets less than or equal to a packet size, each packet destined for the first destination;

generating at least one packet comprising at least a portion of the second quantity of data destined for the second destination;

transmitting a first packet destined for the first destination to the network switch; and

transmitting the at least one packet destined for the second destination to the network switch, before transmitting a second packet destined for the first destination.

36. A method according to claim 35, wherein the first quantity of data comprises at least a portion of a video media file.

37. A method according to claim 35, further comprising generating a plurality of said packets comprising at least a portion of the second quantity of data destined for the second destination.

38. A method according to claim 35, further comprising:

transmitting the second packet destined for the first destination.

39. A method according to claim 35, wherein said write call further comprises a plurality of destinations.

40. A method according to claim 35, further comprising:

communicating at least one of said packets to a network interface card.

41. A method according to claim 35, further comprising:

performing a zero-copy write of said quantity of data.

42. A method according to claim 35, wherein said write call comprises a write vector.

43. A computer program product for balancing load on a network switch;

the computer program product comprising;

a computer-readable medium comprising a program module, the program module including instructions for;

receiving a write call from an application through a socket, the write call comprising a plurality of destinations, including a first destination and a second destination, and a first quantity of data destined for the first destination, and a second quantity of data destined for the second destination;

packetizing said first quantity of data into a plurality of packets less than or equal to a packet size, each packet destined for the first destination;

generating at least one packet comprising at least a portion of the second quantity of data destined for the second destination;

transmitting a first packet destined for the first destination to the network switch; and

transmitting the at least one packet destined for the second destination to the network switch, before transmitting a second packet destined for the first destination.

44. A system for sending data across a network, the system comprising:

a computer running an application configured to send a write call to a driver through a socket, the write call comprising a destination and a quantity of data greater than a packet size, the application further configured to perform a zero-copy write of said quantity of data to said driver;

the driver configured to packetize said data into a plurality of packets less than or equal to said packet size; and

a downstream device adapted to receive at least one of said packets.

45. A system according to claim 44, further comprising a network switch coupled to the computer and the downstream device, adapted to receive at least one of said packets and route the received packet to the downstream device.

46. A system according to claim 44, wherein the downstream device comprises a down stream device has a timing requirement for receipt of data.

47. A system according to claim 45, further comprising a plurality of said computers in communication with the network switch.

48. A system according to claim 44, further comprising a plurality of downstream devices in communication with the network switch.