LOAD-BALANCING INPUT/OUTPUT REQUESTS IN CLUSTERED STORAGE SYSTEMS

• US 20160234297A1
• Filed: 02/11/2015
• Published: 08/11/2016
• Est. Priority Date: 02/11/2015
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Ambach, Christian, Dietz, Mathias, Wolafka, Rainer
• Granted Patent: US 9,888,064 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: H04L 67/1008 based on parameters of servers, e.g. available memory or workload monitoring of computer activity G06F11/30
• CPC Class Codes: H04L 67/1029 using data related to the state of servers by a load balancer
• CPC Class Codes: H04L 67/141 Setup of application sessions admission control or resource allocation in data switching networks H04L47/70
• CPC Class Codes: H04L 69/14 Multichannel or multilink protocols

Abstract

A computer-implemented method for load-balancing client input/output (I/O) requests in a clustered storage system may include receiving a request by a first node of a clustered storage system from a client to initiate a session between the client and the first node. The request may specify a multi-channel communication session. In response to receiving the request, an Internet Protocol (IP) address of the first and at least a second node may be transmitted to the client. The multi-channel communication session may be established between the client and the first and second nodes in which the client communicates with the first node using a first communication channel and communicates with the second node using a second communication channel. The first node may transmit to the second node session data determined at the first node. The second node may transmit to the first node session data determined at the second node.

Claims

1. A computer-implemented method for load-balancing client input/output (I/O) requests in a clustered storage system comprising:

receiving a request by a first node of a clustered storage system from a client to initiate a session between the client and the first node, the clustered storage system having a plurality of nodes and the request specifying a multi-channel communication session;

in response to receiving the request, transmitting an internet protocol (IP) address of the first node and an IP address of at least a second node to the client by the first node;

establishing the multi-channel communication session between the client and the first and second nodes in which the client communicates with the first node using a first communication channel and communicates with the second node using a second communication channel; and

transmitting to the second node from the first node session data determined at the first node and transmitting to the first node from the second node session data determined at the second node.

2. The method of claim 1, further comprising:

determining a first I/O request limit for the clustered storage system;

determining a second I/O request limit for the first node and determining a third I/O request limit for the second node; and

in response to the determining of the first, second, and third I/O request limits, granting the client a first quantity of I/O requests for the client to transmit to the first node and granting the client a second quantity of I/O requests for the client to transmit to the second node, wherein the first quantity of I/O requests do not surpass the first or second I/O request limits and the second quantity of I/O requests do not surpass the first or third I/O request limits.

3. The method of claim 2, wherein the first quantity of I/O requests is changed to a third quantity of I/O requests, and the second quantity of I/O requests is changed to a fourth quantity of I/O requests.

4. The method of claim 2, further comprising adjusting the second I/O request limit and the third I/O request limit based on respective I/O request loads of each of the first and second nodes, and based on a health state of each of the first and second nodes.

5. The method of claim 2, wherein the first and the second quantity of I/O requests are available for the client for a first amount of time, and after the first amount of time no I/O requests are available for the client.

6. The method of claim 2, wherein the determining of the second and the third I/O request limits include determining a hardware capability value for a plurality of hardware components of each of the first and second nodes, wherein a lowest hardware capability value for the plurality of hardware components of the first node is the second I/O request limit and a lowest hardware capability value for the plurality of hardware components of the second node is the third I/O request limit.

7. The method of claim 2, wherein the determining of the first I/O request limit for the clustered storage system includes determining the clustered storage system'"'"'s storage bandwidth and dividing the bandwidth by an average I/O request size for each I/O request performed by a plurality of clients to the plurality of nodes of the clustered storage system in a particular time interval.

8. The method of claim 2, wherein the determining of the first I/O request limit for the clustered storage system includes determining a quantity of I/O requests per second that the clustered storage system executes in a particular time interval.

9. A system for load-balancing client I/O requests in a clustered storage system, comprising:

a first node of a clustered storage system having plurality of nodes, the first node configured to receive a request from a client to initiate a session between the client and the first node, the request specifying a multi-channel communication session;

at least a second node, wherein the first node is further configured to transmit an internet protocol (IP) address of the first node and an IP address of the second node to the client in response to receiving the request;

wherein the first and second nodes are further configured to establish the multi-channel communication session between the client and the first and second nodes in which the client communicates with the first node using a first communication channel and communicates with the second node using a second communication channel; and

wherein the first node is further configured to transmit to the second node session data determined at the first node and the second node is further configured to transmit to the first node session data determined at the second node.

10. The system of claim 9, further comprising:

a buffer credit coordinator configured to determine a first I/O request limit for the clustered storage system;

wherein the buffer credit coordinator is further configured to determine a second I/O request limit for the first node and determine a third I/O request limit for the second node; and

wherein the buffer credit coordinator is further configured to, in response to the determining of the first, second, and third I/O request limits, grant the client a first quantity of I/O requests for the client to transmit to the first node and to grant the client a second quantity of I/O requests for the client to transmit to the second node, wherein the first quantity of I/O requests do not surpass the first or second I/O request limits and the second quantity of I/O requests do not surpass the first or third I/O request limits.

11. The system of claim 10, wherein the buffer credit coordinator is further configured to change the first quantity of I/O requests to a third quantity of I/O requests, and change the second quantity of I/O requests to a fourth quantity of I/O requests.

12. The system of claim 10, wherein the buffer credit coordinator is further configured to adjust the second I/O request limit and the third I/O request limit based on respective I/O request loads of each of the first and second nodes, and based on a health state of each of the first and second nodes.

13. The system of claim 10, wherein the buffer credit coordinator determines the second and the third I/O request limits by determining a hardware capability value for a plurality of hardware components of each of the first and second nodes, wherein a lowest hardware capability value for the plurality of hardware components of the first node is the second I/O request limit and a lowest hardware capability value for the plurality of hardware components of the second node is the third I/O request limit.

14. The system of claim 10, wherein the buffer credit coordinator determines the first I/O request limit for the clustered storage system by determining a quantity of I/O requests per second that the clustered storage system executes in a particular time interval.

15. A computer program product for load-balancing client I/O requests in a clustered storage system, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a clustered storage system to cause the clustered storage system to:

receive a request by a first node of the clustered storage system from a client to initiate a session between the client and the first node, the clustered storage system having a plurality of nodes and the request specifying a multi-channel communication session;

transmit, in response to the receiving of the request, an internet protocol (IP) address of the first node and an IP address of at least a second node to the client by the first node;

establish the multi-channel communication session between the client and the first and second nodes in which the client communicates with the first node using a first communication channel and communicates with the second node using a second communication channel; and

transmit to the second node from the first node session data determined at the first node and transmit to the first node from the second node session data determined at the second node.

16. The computer program product of claim 15, wherein the program instructions executable by the clustered storage system further cause the clustered storage system to:

determine a first I/O request limit for the clustered storage system;

determine a second I/O request limit for the first node and determine a third I/O request limit for the second node; and

grant the client, in response to the determining of the first, second, and third I/O request limits, a first quantity of I/O requests for the client to transmit to the first node and grant the client a second quantity of I/O requests for the client to transmit to the second node, wherein the first quantity of I/O requests do not surpass the first or second I/O request limits and the second quantity of I/O requests do not surpass the first or third I/O request limits.

17. The computer program product of claim 16, wherein the program instructions executable by the clustered storage system further cause the clustered storage system to adjust the second I/O request limit and the third I/O request limit based on respective I/O request loads of each of the first and second nodes, and based on a health state of each of the first and second nodes.

18. The computer program product of claim 16, wherein the program instructions executable by the clustered storage system further cause the clustered storage system to make the first and the second quantity of I/O requests available for the client for a first amount of time, and after the first amount of time no I/O requests are available for the client.

19. The computer program product of claim 16, wherein the program instructions executable by the clustered storage system to cause the clustered storage system to determine the second and the third I/O request limits includes determining a hardware capability value for a plurality of hardware components of each of the first and second nodes, wherein a lowest hardware capability value for the plurality of hardware components of the first node is the second I/O request limit and a lowest hardware capability value for the plurality of hardware components of the second node is the third I/O request limit.

20. The computer program product of claim 16, wherein the program instructions executable by the clustered storage system to cause the clustered storage system to determine the first I/O request limit for the clustered storage system includes determining the clustered storage system'"'"'s storage bandwidth and dividing the bandwidth by an average I/O request size for each I/O request performed by a plurality of clients to the plurality of nodes of the clustered storage system in a particular time interval.