FACILITATING ELASTIC ALLOCATION OF ORGANIZATION-SPECIFIC QUEUE RESOURCES IN AN ON-DEMAND SERVICES ENVIRONMENT

• US 20160117195A1
• Filed: 10/28/2014
• Published: 04/28/2016
• Est. Priority Date: 10/28/2014
• Status: Active Grant

• Current Assignee: Salesforce.com, Inc.
• Original Assignee: Salesforce.com, Inc.
• Inventors: WANG, XIAODAN
• Granted Patent: US 10,776,373 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 16/24549 Run-time optimisation
• CPC Class Codes: G06F 16/2471 Distributed queries
• CPC Class Codes: G06F 9/5072 Grid computing
• CPC Class Codes: G06F 9/5083 Techniques for rebalancing the load in a distributed system

Abstract

In accordance with embodiments, there are provided mechanisms and methods for facilitating elastic allocation of tenant-specific queue resources in an on-demand services environment in a multi-tenant environment according to one embodiment. In one embodiment and by way of example, a method includes allocating resources to a plurality of tenants, identifying, in runtime, one or more offending tenants of the plurality of tenants and one or more victim tenants of the plurality of tenants. The one or more offending tenants consume above their allocated share of the resources within a message type, and the one or more victim tenants consume below their allocated share of the resources or none of the resources within the message type. The method may further include isolating, in runtime, the offending tenants and the victim tenants, and routing, in runtime, each tenant of the offending tenants and the victim tenants to a queue dedicated to the tenant and the message type.

Claims

1. A database system-implemented method comprising:

allocating, by and incorporating into the database system, resources to a plurality of tenants;

identifying, in runtime, by the database system, one or more offending tenants of the plurality of tenants and one or more victim tenants of the plurality of tenants, wherein the one or more offending tenants consume above their allocated share of the resources within a message type, and the one or more victim tenants consume below their allocated share of the resources or none of the resources within the message type;

isolating, in runtime, by the database system, the offending tenants and the victim tenants; and

routing, in runtime, by the database system, each tenant of the offending tenants and the victim tenants to a queue dedicated to the tenant and the message type.

2. The method of claim 1, further comprising fetching, in runtime, by the database system, a plurality of queues associated with a plurality of tenants, wherein each queue is associated with at least a tenant and a message type of a plurality of message types, wherein identifying comprises determining, in runtime, by the database system, at least one of a predetermined latency threshold allocated to each message type, and a predetermined consumption capacity threshold of thread resources allocated to each tenant or message type.

3. The method of claim 2, further comprising:

dynamically splitting, in runtime, by the database system, a split group of tenants of the plurality of tenants, wherein one or more tenants of the split group of tenants exceed at least one of the latency threshold and the consumption capacity threshold; and

dedicating one or more queues to the one or more tenants of the split group.

4. The method of claim 1, further comprising:

evaluating, by the database system, the plurality of tenants for the latency threshold, wherein evaluating includes selecting a first group of tenants from the plurality of tenants for exceeding the latency threshold; and

evaluating, by the database system, the first group tenants for consumption of thread time relating to one or more message types, wherein evaluating comprises selecting top tenants from the first group of tenants consuming a highest amount of thread time for the one or more message types compared to other tenants of the plurality of tenants.

5. The method of claim 4, further comprising evaluating, by the database system, the top tenants for the consumption capacity, wherein evaluating further comprises filtering the top group for the split group of tenants exceeding the consumption capacity threshold.

6. The method of claim 2, further comprising dynamically coalescing, in runtime, by the database system, a coalesce group of tenants of the plurality of tenants, wherein one or more tenants of the coalesce group of tenants fall short of at least one of the latency threshold and the consumption capacity threshold, wherein one or more queues are dedicated to the one or more tenants of the coalesce group.

7. The method of claim 6, further comprising:

aggregating, by the database system, a thread time consumed by of the plurality of tenants and the plurality of message types of the plurality of message types; and

evaluating, by the database system, the plurality of the tenants and the plurality of message types, wherein evaluating includes marking the coalesce group of tenants from the plurality of tenants if a portion of the aggregated thread time associated with the one or more tenants of the coalesce group of tenants is less than the consumption capacity of the threshold.

8. The method of claim 1, further comprising merging, by the database system, one or more merge candidate tenants of the split group of tenants and the coalesce group of tenants if one or more queues associated with the one or more merge candidate tenants are selected to be reclaimed.

9. A system comprising:

a processor and a memory to execute instructions at the system; and

a mechanism to;

allocate resources to a plurality of tenants;

identify, in runtime, one or more offending tenants of the plurality of tenants and one or more victim tenants of the plurality of tenants, wherein the one or more offending tenants consume above their allocated share of the resources within a message type, and the one or more victim tenants consume below their allocated share of the resources or none of the resources within the message type;

isolate, in runtime, the offending tenants and the victim tenants; and

route, in runtime, each tenant of the offending tenants and the victim tenants to a queue dedicated to the tenant and the message type.

10. The system of claim 9, wherein the mechanism is further to fetch, in runtime, by the database system, a plurality of queues associated with a plurality of tenants, wherein each queue is associated with at least a tenant and a message type of a plurality of message types, wherein identifying comprises determining, in runtime, by the database system, at least one of a predetermined latency threshold allocated to each message type, and a predetermined consumption capacity threshold of thread resources allocated to each tenant or message type.

11. The system of claim 10, wherein the mechanism is further to:

dynamically split, in runtime, a split group of tenants of the plurality of tenants, wherein one or more tenants of the split group of tenants exceed at least one of the latency threshold and the consumption capacity threshold; and

dedicate one or more queues to the one or more tenants of the split group.

12. The system of claim 9, wherein the mechanism is further to:

evaluate the plurality of tenants for the latency threshold, wherein evaluating includes selecting a first group of tenants from the plurality of tenants for exceeding the latency threshold; and

evaluate the first group tenants for consumption of thread time relating to one or more message types, wherein evaluating comprises selecting top tenants from the first group of tenants consuming a highest amount of thread time for the one or more message types compared to other tenants of the plurality of tenants.

13. The system of claim 12, wherein the mechanism is further to evaluate the top tenants for the consumption capacity, wherein evaluating further comprises filtering the top group for the split group of tenants exceeding the consumption capacity threshold.

14. The system of claim 10, wherein the mechanism is further to dynamically coalesce, in runtime, a coalesce group of tenants of the plurality of tenants, wherein one or more tenants of the coalesce group of tenants fall short of at least one of the latency threshold and the consumption capacity threshold, wherein one or more queues are dedicated to the one or more tenants of the coalesce group.

15. The system of claim 14, wherein the mechanism is further to:

aggregate a thread time consumed by of the plurality of tenants and the plurality of message types of the plurality of message types; and

evaluate the plurality of the tenants and the plurality of message types, wherein evaluating includes marking the coalesce group of tenants from the plurality of tenants if a portion of the aggregated thread time associated with the one or more tenants of the coalesce group of tenants is less than the consumption capacity of the threshold.

16. The system of claim 9, wherein the mechanism is further to merge one or more merge candidate tenants of the split group of tenants and the coalesce group of tenants if one or more queues associated with the one or more merge candidate tenants are selected to be reclaimed.

17. A machine-readable medium comprising a plurality of instructions which, when executed by a processing device, cause the processing device to perform one or more operations comprising:

allocating resources to a plurality of tenants;

identifying, in runtime, one or more offending tenants of the plurality of tenants and one or more victim tenants of the plurality of tenants, wherein the one or more offending tenants consume above their allocated share of the resources within a message type, and the one or more victim tenants consume below their allocated share of the resources or none of the resources within the message type;

isolating, in runtime, the offending tenants and the victim tenants; and

routing, in runtime, each tenant of the offending tenants and the victim tenants to a queue dedicated to the tenant and the message type.

18. The machine-readable medium of claim 17, wherein the mechanism is further to fetch, in runtime, by the database system, a plurality of queues associated with a plurality of tenants, wherein each queue is associated with at least a tenant and a message type of a plurality of message types, wherein identifying comprises determining, in runtime, by the database system, at least one of a predetermined latency threshold allocated to each message type, and a predetermined consumption capacity threshold of thread resources allocated to each tenant or message type.

19. The machine-readable medium of claim 18, wherein the mechanism is further to:

dynamically split, in runtime, a split group of tenants of the plurality of tenants, wherein one or more tenants of the split group of tenants exceed at least one of the latency threshold and the consumption capacity threshold; and

dedicate one or more queues to the one or more tenants of the split group.

20. The machine-readable medium of claim 17, wherein the mechanism is further to:

evaluate the plurality of tenants for the latency threshold, wherein evaluating includes selecting a first group of tenants from the plurality of tenants for exceeding the latency threshold; and

evaluate the first group tenants for consumption of thread time relating to one or more message types, wherein evaluating comprises selecting top tenants from the first group of tenants consuming a highest amount of thread time for the one or more message types compared to other tenants of the plurality of tenants.

21. The machine-readable medium of claim 20, wherein the mechanism is further to evaluate the top tenants for the consumption capacity, wherein evaluating further comprises filtering the top group for the split group of tenants exceeding the consumption capacity threshold.

22. The machine-readable medium of claim 18, wherein the mechanism is further to dynamically coalesce, in runtime, a coalesce group of tenants of the plurality of tenants, wherein one or more tenants of the coalesce group of tenants fall short of at least one of the latency threshold and the consumption capacity threshold, wherein one or more queues are dedicated to the one or more tenants of the coalesce group.

23. The machine-readable medium of claim 22, wherein the mechanism is further to:

aggregate a thread time consumed by of the plurality of tenants and the plurality of message types of the plurality of message types; and

evaluate the plurality of the tenants and the plurality of message types, wherein evaluating includes marking the coalesce group of tenants from the plurality of tenants if a portion of the aggregated thread time associated with the one or more tenants of the coalesce group of tenants is less than the consumption capacity of the threshold.

24. The machine-readable medium of claim 17, wherein the mechanism is further to merge one or more merge candidate tenants of the split group of tenants and the coalesce group of tenants if one or more queues associated with the one or more merge candidate tenants are selected to be reclaimed.