ltstudiooo - Fotolia
Server applications must exchange data with end users and storage subsystems, so effective server virtualization...
must also include solid network performance. Inefficient configurations and latency can conspire against workloads and impair the user experience. Fortunately, hypervisors like VMware ESXi provide ample tools to measure network latency along with techniques for increasing efficiency.
Free point tools like Iperf or Iometer are often recommended for ad-hoc network performance testing. The goal is to test the maximum bandwidth from strategic points in the environment. For example, test bandwidth at the suspect virtual machine, test bandwidth on a system outside of the virtualized host server, test bandwidth on a system outside of the virtualized host on the same VLAN, and then test bandwidth on two VMs on the same virtualized server and vSwitch.
After you measure network latency, compare results and look for the presence of possible network bottlenecks. For example, poor network performance on only one VM may suggest resource allocation problems -- CPU and memory -- to that troubled workload. By comparison, network issues afflicting every workload on a specific server -- while other servers perform normally -- may suggest possible network configuration issues or trouble with the server's network interface cards or software optimization issues like improper traffic shaping or incompatible jumbo frame configurations. And network performance issues with related server groups may suggest network problems such as a fault in a common shared switch. There are certainly other possible causes to consider, but using tools to evaluate the problem's scope is a standard troubleshooting tactic that can help narrow the cause.
If you do choose to use tools like Iperf to measure network latency, VMware suggests changing the TCP window size to 64 KB at the client and server sides. Smaller TCP sizes may cause excess handshakes resulting in unexpectedly low network performance results. Refer to the instructions for your specific tool in order to define the optimum TCP window size or other test settings that are most appropriate for your particular environment.
Dig Deeper on Virtual machine monitoring, troubleshooting and alerting
Related Q&A from Stephen J. Bigelow
Learn how load balancing in the cloud differs from a traditional network traffic distribution, and explore services available from AWS, Google and ... Continue Reading
Access management is critical to securing the cloud. Understand the differences between AWS IAM roles and users to properly restrict access to AWS ... Continue Reading
Containers have rapidly come into focus as a popular option for deploying applications, but they have limitations and are fundamentally different ... Continue Reading