Performance characterization for datacenters

Patent metadata

Field	Value
Publication number	US-10747577-B2
Application number	US-201816112696-A
Country	US
Kind code	B2
Filing date	Aug 25, 2018
Priority date	Aug 26, 2017
Publication date	Aug 18, 2020
Grant date	Aug 18, 2020

Abstract

Computer system and method for characterizing throughput performance of a datacenter utilize bandwidth information of physical network interfaces in the datacenter and results of benchmark testing on throughput on a single processor core to compute a plurality of throughput constraints that define a throughput capacity region for the datacenter to improve throughput performance of the datacenter.

First claim

What is claimed is:

1. A method for characterizing throughput performance of a datacenter having a plurality of host computers connected to a physical network, the method comprising:
retrieving bandwidth information of physical network interfaces in the datacenter;
executing benchmark testing on the datacenter to determine a processor processable throughput on a single processor core for data traffic between virtual processing instances in the datacenter;
computing a plurality of throughput constraints that define a throughput capacity region for the datacenter using the bandwidth information of the physical network interfaces and the determined processor processable throughput on the single processor core; and
adjusting resource allocation schemes in the datacenter to modify the throughput capacity region for the datacenter to improve the throughput performance of the datacenter.

2. The method of claim 1 , wherein computing the plurality of throughput constraints includes computing at least one constraint that specifies that rates of flows processed by the same processing core on one of transmit side and receive side are limited by a processing capability of that processor core.

3. The method of claim 2 , wherein computing the at least one constraint includes computing an inequality that states that a summation of processor shares by all flows in the datacenter that are processed by a processor core on a transmit or receive side of a hypervisor is equal to or less than 1, for any transmit or receive queue that is mapped to the processor core and any hypervisor.

4. The method of claim 1 , wherein computing the plurality of throughput constraints includes computing at least one constraint that specifies that an aggregate rate of flows that are sent out of or received by a physical network interface cannot exceed the bandwidth of the physical network interface.

5. The method of claim 4 , wherein computing the at least one constraint includes computing an inequality that states that a summation, across all transmit or receive queues that are mapped to processor cores, of rates of all transmitted or received flows that are processed and mapped to a transmit or receive queue, is equal to or less than the bandwidth of a network interface on transmit side or receive side of any hypervisor.

6. The method of claim 1 , wherein executing the benchmark testing on the datacenter includes executing the benchmark testing on the datacenter to determine a best possible transmit-side processor processable throughput on a single processor core for data traffic between virtual processing instances in the datacenter and a best possible receive-side processor processable throughput on a single processor core for data traffic between virtual processing instances in the datacenter.

7. The method of claim 1 , further comprising computing an effective capacity value, wherein the effective capacity is defined to be an expected value of an aggregate system throughput of all flows through the datacenter if the aggregate system throughput is considered to be a function of random flow matrix F with a density function g(F) across the capacity region.

8. The method of claim 1 , further comprising providing suggestions for resource allocation schemes to modify the throughput capacity region of the datacenter.

9. A non-transitory computer-readable storage medium containing program instructions for characterizing throughput performance of a datacenter having a plurality of host computers connected to a physical network, wherein execution of the program instructions by one or more processors of a computer system causes the one or more processors to perform steps comprising:
retrieving bandwidth information of physical network interfaces in the datacenter;
executing benchmark testing on the datacenter to determine a processor processable throughput on a single processor core for data traffic between virtual processing instances in the datacenter;
computing a plurality of throughput constraints that define a throughput capacity region for the datacenter using the bandwidth information of the physical network interfaces and the determined processor processable throughput on the single processor core; and
adjusting resource allocation schemes in the datacenter to modify the throughput capacity region for the datacenter to improve the throughput performance of the datacenter.

10. The computer-readable storage medium of claim 9 , wherein computing the plurality of throughput constraints includes computing at least one constraint that specifies that rates of flows processed by the same processing core on one of transmit side and receive side are limited by a processing capability of that processor core.

11. The computer-readable storage medium of claim 10 , wherein computing the at least one constraint includes computing an inequality that states that a summation of processor shares by all flows in the datacenter that are processed by a processor core on a transmit or receive side of a hypervisor is equal to or less than 1, for any transmit or receive queue that is mapped to the processor core and any hypervisor.

12. The computer-readable storage medium of claim 9 , wherein computing the plurality of throughput constraints includes computing at least one constraint that specifies that an aggregate rate of flows that are sent out of or received by a physical network interface cannot exceed the bandwidth of the physical network interface.

13. The computer-readable storage medium of claim 12 , wherein computing the at least one constraint includes computing an inequality that states that a summation, across all transmit or receive queues that are mapped to processor cores, of rates of all transmitted or received flows that are processed and mapped to a transmit or receive queue, is equal to or less than the bandwidth of a network interface on transmit side or receive side of any hypervisor.

14. The computer-readable storage medium of claim 9 , wherein executing the benchmark testing on the datacenter includes executing the benchmark testing on the datacenter to determine a best possible transmit-side processor processable throughput on a single processor core for data traffic between virtual processing instances in the datacenter and a best possible receive-side processor processable throughput on a single processor core for data traffic between virtual processing instances in the datacenter.

15. The computer-readable storage medium of claim 9 , wherein the steps further comprise computing an effective capacity value, wherein the effective capacity is defined to be an expected value of an aggregate system throughput of all flows through the datacenter if the aggregate system throughput is considered to be a function of random flow matrix F with a density function g(F) across the capacity region.

16. The computer-readable storage medium of claim 9 , wherein the steps further comprise providing suggestions for resource allocation schemes to modify the throughput capacity region of the datacenter.

17. A computer system comprising:
memory; and
a processor configured to:
retrieve bandwidth information of physical network interfaces in a datacenter;
execute benchmark testing on the datacenter to determine a processor processable throughput on a single processor core for data traffic between virtual processing instances in the datacenter;
compute a plurality of throughput constraints that define a throughput capacity region for the datacenter using the bandwidth information of the physical network interfaces and the determined process

Assignees

• Nicira Inc

Inventors

• Wang Dexiang
• Qian Bin
• Yang Jinqiang
• Kankipati Naga S S Kishore
• Pillai Sanal
• Sundararaman Sujatha
• Chandrashekhar Ganesan
• Mehta Rishi

Classifications

• H04L41/0896 Primary

  Bandwidth or capacity management, i.e. automatically increasing or decreasing capacities (flow or congestion control using dynamic resource allocation, e.g. in-call renegotiation, H04L47/76) · CPC title

• G06F9/5011 Primary

  the resources being hardware resources other than CPUs, Servers and Terminals · CPC title

• G06F2209/508

  Monitor · CPC title

• H04L43/0894

  Packet rate · CPC title

• H04L43/0888

  Throughput · CPC title