Prevention of head of line blocking in a multi-rate switched Fibre Channel loop attached system

• US 7,586,850 B2
• Filed: 02/23/2005
• Issued: 09/08/2009
• Est. Priority Date: 02/23/2005
• Status: Expired due to Fees

• Current Assignee: Avago Technologies International Sales Pte Limited (Broadcom, Inc.)
• Original Assignee: Emulex Design & Manufacturing Corp. (Broadcom, Inc.)
• Inventors: Goodwin, William P., Doherty, Terrence R., Mies, Carl Joseph, Warren, Bruce Gregory
• Primary Examiner(s): Orgad; Edan
• Assistant Examiner(s): Weidner; Timothy J
• Application Number:
  US11/064,550
• Publication Number:
  US 20060187837A1
• Time in Patent Office: 1,658 Days
• Field of Search: None
• US Class Current: 370/236
• CPC Class Codes: H04L 49/357 Fibre channel switches
• CPC Class Codes: H04L 49/508 Head of Line Blocking Avoidance

Abstract

A method is disclosed for maintaining a table of recent accesses for each port for use in predicting whether a request for data from a source device is likely to be sent to a high speed or low speed destination device. The table of recent accesses lists every source device attached to that port and the speed of the destination device with the most recent access to each source device. When an OPN primitive is received at the source port, the source device is identified and used with the table of recent accesses to predict whether the destination device is likely to be high speed or low speed, and ultimately whether to send data from the source device or reject the request.

Claims

1. A method for reducing head of line blocking in a first multi-port frame-based buffered storage switch having transmit buffers and receive buffers associated with each port when a high speed destination device attached to a first destination port and a low speed destination device attached to a second destination port both need to receive data from a high speed source device communicatively coupled to one or more source ports, the method comprising:

maintaining a table of recent accesses at said ports;

receiving a request to transfer data at the one or more source ports;

determining a speed of a presumed destination device by reference to said table;

if the presumed destination device is a low speed destination device, transferring data to the receive buffers of a source port only after determining that the data can be subsequently transferred to the transmit buffers of the second destination port attached to the low speed destination device without impediment; and

if the presumed destination device is a high speed destination device, transferring data to the receive buffers of a source port only as long as the receive buffers have space to store the data.

2. The method as recited in claim 1, the method further comprising:

determining the speed of the presumed destination device by

receiving an OPN primitive containing the request,

extracting a source device AL_PA from the OPN primitive, and

using the source device AL_PA to index said table of recent accesses and determine the speed of the presumed destination device;

if the data is destined for the low speed destination device, transferring data to the receive buffers of the source port only as long as the receive buffers have space to store the data and the transmit buffers of the second destination port attached to the low speed destination device have space to store the data from the receive buffers; and

if the data is destined for the high speed destination device, transferring data to the receive buffers of the source port as long as the receive buffers have space to store the data.

3. The method as recited in claim 2, further comprising maintaining a separate table of recent accesses for each port in the multi-port frame-based buffered storage switch.

4. The method as recited in claim 3, wherein each table of recent accesses contains a source AL_PA for every device attached to the port associated with that table of recent accesses, the speed of the destination device with the most recent access to the device represented by the source AL_PA.

5. The method as recited in claim 4, further comprising:

updating the table of recent accesses associated with a particular port each time a frame is received.

6. The method as recited in claim 2, further comprising:

if the data is destined for the low speed destination device but the receive buffers do not have space to store the data or the transmit buffers do not have space to store the data from the receive buffers, rejecting the request; and

if the data is destined for the high speed destination device but the receive buffers do not have space to store the data, rejecting the request.

7. The method as recited in claim 6, further comprising:

if the request has been rejected a predetermined consecutive number of times without any transfer of data to the low speed device, transferring data to the receive buffers of one of the one or more source ports only as long as the receive buffers have space to store the data.

8. The method as recited in claim 1, operable when the high speed source device is attached to a second cascaded multi-port frame-based buffered storage switch having transmit buffers and receive buffers associated with each port and communicatively coupled to two source ports on the first buffered storage switch via a high speed trunk and a low speed trunk, the method further comprising:

determining the speed of the presumed destination device by

receiving a frame at the high speed source device,

extracting an actual destination device AL_PA from the frame, and

using the actual destination device AL_PA to index said table of devices and determine a speed of the actual destination device;

if the data is destined for the low speed destination device, transferring data through the second buffered storage switch and the low speed trunk to the receive buffers of the source port connected to the low speed trunk as long as the receive buffers have space to store the data; and

if the data is destined for the high speed destination device, transferring data through the second buffered storage switch and the high speed trunk to the receive buffers of the source port connected to the high speed trunk as long as the receive buffers have space to store the data.

9. A first multi-port frame-based buffered storage switch for reducing head of line blocking, comprising:

one or more first ports couplable to a high speed source device for sending data;

a second port couplable to a high speed destination device for receiving data from the high speed source device;

a third port couplable to a low speed destination device for receiving data from the high speed source device;

a table of recent accesses associated with each of the one or more first ports and containing a list of all devices connected to the respective port and, for each device in the list, a speed of the destination device having a most recent access to that device; and

a state machine associated with each of the one or more first ports for

receiving a request to transfer data,

determining a speed of a presumed destination device by reference to the associated table of recent accesses,

if the presumed destination device is a low speed destination device, transferring data to receive buffers in one of the one or more first ports only after determining that the data can be subsequently transferred to transmit buffers in the third port without impediment, and

if the presumed destination device is a high speed destination device, transferring data to receive buffers in the second port only as long as the receive buffers have space to store the data.

10. The first storage switch as recited in claim 9, further comprising:

wherein the state machine associated with the first port is further configured for

determining a presumed destination device for the received request and a speed of the presumed destination device by

receiving an OPN primitive containing the request,

extracting a source device AL_PA from the OPN primitive, and

using the source device AL_PA to index the associated table of recent accesses and determine a presumed destination device AL_PA for the request and the speed of the presumed destination device,

if the data is destined for the low speed destination device, transferring data to the receive buffers of the first port only as long as the receive buffers have space to store the data and the transmit buffers of the third port have space to store the data from the receive buffers, and

if the data is destined for the high speed destination device, transferring data to the receive buffers of the first port as long as the receive buffers have space to store the data.

11. The first storage switch as recited in claim 10, wherein each table of recent accesses contains a source AL_PA for every device attached to the port associated with that table of recent accesses and a speed of the destination device with the most recent access to the device represented by the source AL_PA.

12. The first storage switch as recited in claim 10, the state machine associated with each of the one or more first ports further configured for:

if the data is destined for the low speed destination device but the receive buffers do not have space to store the data or the transmit buffers do not have space to store the data from the receive buffers, rejecting the request; and

if the data is destined for the high speed destination device but the receive buffers do not have space to store the data, rejecting the request.

13. The first storage switch as recited in claim 10, the state machine associated with each of the one or more first ports further configured for:

if the request has been rejected a predetermined consecutive number of times without any transfer of data to the low speed device, transferring data to the receive buffers of the first port only as long as the receive buffers have space to store the data.

14. The first storage switch as recited in claim 10, the state machine associated with each of the one or more first ports further configured for:

updating the table of recent accesses associated with a particular port each time a frame is received.

15. A storage system comprising the first storage switch as recited in claim 9, further comprising:

a second cascaded multi-port frame-based buffered storage switch having transmit buffers and receive buffers associated with each port, the second storage switch attached to the high speed source device and communicatively coupled to two first ports on the first storage switch via a high speed trunk and a low speed trunk;

a table of devices stored in the first and second storage switches containing a list of all devices connected to the storage system and, for each device in the list, the speed of that device; and

the state machine associated with each of the two first ports further configured for

determining a presumed destination device for the received request and a speed of the presumed destination device by

receiving a frame at the high speed source device,

extracting an actual destination device AL_PA from the frame, and

using the actual destination device AL_PA to index the table of devices and determine a speed of the actual destination device,

if the data is destined for the low speed destination device, transferring data through the second buffered storage switch and the low speed trunk to the receive buffers of the first port connected to the low speed trunk as long as the receive buffers have space to store the data, and

if the data is destined for the high speed destination device, transferring data through the second buffered storage switch and the high speed trunk to the receive buffers of the first port connected to the high speed trunk as long as the receive buffers have space to store the data.

16. A storage area network (SAN) comprising the storage system of claim 15.