Method and system for providing direct data placement support

• US 20060034283A1
• Filed: 08/13/2004
• Published: 02/16/2006
• Est. Priority Date: 08/13/2004
• Status: Abandoned Application

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Recio, Renato J., Sarkar, Prasenjit, Ko, Michael Anthony
• Application Number:
  US10/917,508
• Publication Number:
  US 20060034283A1
• Time in Patent Office: Days
• US Class Current: 370/392
• CPC Class Codes: H04L 63/12 Applying verification of the received information cryptographic mechanisms or cryptographic arrangements for data integrity or data verification H04L9/32
• 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

Abstract

A system and method for reducing the overhead associated with direct data placement is provided. Processing time overhead is reduced by implementing packet-processing logic in hardware. Storage space overhead is reduced by combining results of hardware-based packet-processing logic with ULP software support; parameters relevant to direct data placement are extracted during packet-processing and provided to a control structure instantiation. Subsequently, payload data received at a network adapter is directly placed in memory in accordance with parameters previously stored in a control structure. Additionally, packet-processing in hardware reduces interrupt overhead by issuing system interrupts in conjunction with packet boundaries. In this manner, wire-speed direct data placement is approached, zero copy is achieved, and per byte overhead is reduced with respect to the amount of data transferred over an individual network connection. Movement of ULP data between application-layer program memories is thereby accelerated without a fully offloaded TCP protocol stack implementation.

Claims

1. A method for reducing the overhead associated with the direct placement of packet data incoming to a network adapter over a network connection;

said method comprising;

a. receiving a header portion of at least one packet at said network adapter;

b. extracting and processing, via logic implemented in hardware, upper layer protocol (ULP) parameter values from said header portion of said at least one packet;

c. storing in software data structure, said ULP parameters values extracted from header portion of at least one packet; and

d. directly placing packet data received in a payload portion of said at least one packet;

said placement based on said stored ULP parameters values.

2. A method for reducing the overhead associated with the direct placement of packet data, as per claim 1, wherein said ULP is either of:

Internet Small Computer System Interface (iSCSI) or the iWARP protocol suite;

said iWARP protocol suite comprising Remote Direct Memory Access Protocol (RDMAP), Direct Data Placement Protocol (DDP), and Marker PDU Aligned Framing for TCP (MPA).

3. A method for reducing the overhead associated with the direct placement of packet data, as per claim 1, wherein said packet data is placed in a memory location specified by at least one of said stored ULP parameter values.

4. A method for reducing the overhead associated with the direct placement of packet data, as per claim 1, wherein said processing step comprises scheduling interrupts on boundaries of said at least one packet.

5. A method for reducing the overhead associated with the direct placement of packet data, as per claim 2, wherein said packet data is directly placed if said stored ULP parameter values satisfy conditions necessary for direct data placement.

6. A method for reducing the overhead associated with the direct placement of packet data, as per claim 2, wherein said logic implemented in hardware performs functions comprising:

verifying cyclic redundancy check (CRC) for RDMA messages, marker removal for RDMA messages, and interrupt-scheduling on RDMA message boundaries, if said ULP is the iWARP protocol suite; and

interrupt-scheduling on iSCSI Protocol Data Unit (PDU) boundaries, if said ULP is iSCSI.

7. A method for reducing the overhead associated with the direct placement of packet data, as per claim 5, wherein said conditions are comprised of:

determining an RDMA message is in sequence, determining that an RDMA message is properly aligned, checking that inbound RDMA write is enabled for an incoming RDMA write message, and checking that inbound RDMA read is enabled for an incoming RDMA read response message;

if said ULP is the iWARP protocol suite;

else if said ULP is iSCSI, said conditions are comprised of;

determining that iSCSI PDUs are received in order, determining correctness of iSCSI PDU header and data digests, and determining that TCP segments in said at least one packet are received in order.

8. A method for reducing the overhead associated with the direct placement of packet data, as per claim 6, wherein said logic implemented in hardware performs functions further comprising:

segmenting said payload portion of at least one packet, generating checksums for said at least one packet, inserting markers in said payload portion of at least one packet, performing header and data digests if said ULP is iSCSI, and generating CRCs if said ULP is the iWARP protocol suite.

9. A system for reducing the overhead associated with the direct placement of packet data incoming to a network adapter over a network connection;

said system comprising;

a. hardware receiving a header portion of at least one packet incoming to said network adapter;

said hardware extracting and processing upper layer protocol (ULP) parameter values from said header portion of at least one packet;

b. software storing said ULP parameters values extracted from said header portion of at least one packet; and

c. direct data placement of packet data received in a payload portion of said at least one packet;

said placement based on said stored ULP parameters values.

10. A system for reducing the overhead associated with the direct placement of packet, as per claim 9, wherein said ULP is either of:

the iWARP protocol suite or Internet Small Computer System Interface (iSCSI).

11. A system for reducing the overhead associated with the direct placement of packet data, as per claim 9, wherein said packet data is placed in a memory location specified by at least one of said stored ULP parameter values.

12. A system for reducing the overhead associated with the direct placement of packet data, as per claim 9, wherein said processing step comprises scheduling interrupts on boundaries of said at least one packet.

13. An article of manufacture comprising a computer usable medium having computer readable program code embodied therein which implements a reduction of the overhead associated with the direct placement of packet data incoming to a network adapter over a network connection;

said medium comprising of modules for;

a. receiving a header portion of at least one packet at said network adapter;

b. extracting and processing, via logic implemented in hardware, upper layer protocol (ULP) parameter values from said header portion of said at least one packet;

c. storing in memory accessible by software, said ULP parameters values extracted from header portion of said at least one packet; and

d. directly placing packet data received in a payload portion of said at least one packet;

said placement based on said stored ULP parameter values.

14. An article of manufacture comprising a computer usable medium, as per claim 13, wherein said ULP is either of:

the iWARP protocol suite or Internet Small Computer System Interface (iSCSI).

15. An article of manufacture comprising a computer usable medium, as per claim 13, wherein said packet data is placed in a memory location specified by at least one of said stored ULP parameter values.

16. An article of manufacture comprising a computer usable medium, as per claim 13, wherein said processing step comprises scheduling interrupts on boundaries of at least one packet.

17. An article of manufacture comprising a computer usable medium, as per claim 14, wherein said packet data is directly placed if said stored ULP parameter values satisfy conditions necessary for direct data placement.

18. An article of manufacture comprising a computer usable medium, as per claim 14, wherein said logic implemented in hardware performs functions comprising:

verifying cyclic redundancy check (CRC) for RDMA messages, marker removal for RDMA messages, and interrupt-scheduling on RDMA message boundaries, if said ULP is the iWARP protocol suite; and

interrupt-scheduling on iSCSI Protocol Data Unit (PDU) boundaries, if said ULP is iSCSI.

19. An article of manufacture comprising a computer usable medium, as per claim 17, wherein said conditions are comprised of:

determining an RDMA message is in sequence, determining that an RDMA message is properly aligned, checking that inbound RDMA write is enabled for an incoming RDMA write message, and checking that inbound RDMA read is enabled for an incoming RDMA read response message;

if said ULP is the iWARP protocol suite;

else if said ULP is iSCSI, said conditions are comprised of;

determining that iSCSI PDUs are received in order, determining correctness of iSCSI PDU header and data digests, and determining that TCP segments in said at least one packet are received in order.

20. An article of manufacture comprising a computer usable medium, as per claim 18, wherein said logic implemented in hardware performs functions further comprising:

segmenting said payload portion of at least one packet, generating checksums for said header portion of said at least one packet, performing header and data digests if said ULP is iSCSI, and inserting markers in said payload portion of at least one packet and generating CRCs if said ULP is the iWARP protocol suite.