Steps to Run Perftest in HCX

Products

VMware Cloud on AWS VMware HCX

Issue/Introduction

Introduction to Perftest in HCX Environments

This article explains the usage of 'perftest' to determine available bandwidth within HCX tunnels. The process involves conducting bidirectional tests for each perftest instance. In a typical setup, there are two Gateways: one located in the customer's datacenter (referred to as OnPrem) and another in the Cloud environment.

Perftest measures throughput in both directions

From OnPrem to Cloud
From Cloud to OnPrem

The perftest throughput mechanism operates by having the sender attempt to transmit data as rapidly as the link can sustain. Consequently, each test typically shows a higher "sender" rate compared to the "receiver" rate. The "receiver" rate should be considered as the one-way throughput result.

For result interpretation

OnPrem to Cloud throughput represents the upload speed from the customer's perspective.
Cloud to OnPrem throughput represents the download speed from the customer's perspective.

Key Points

This article outlines the step-by-step procedure for running 'perftest' in an HCX environment.

'Perftest' provides measurements of available bandwidth for HCX tunnels.

Environment

HCX

Cause

Several scenarios may necessitate running a perftest in an HCX (Hybrid Cloud Extension) environment

Initial Setup Validation

After deploying HCX and establishing tunnels between on-premises and cloud environments.
To verify that the network performance meets expected baselines and service-level agreements (SLAs).

Migration Planning

Before initiating large-scale VM migrations to ensure sufficient bandwidth.
To estimate migration times and plan maintenance windows accurately.

Performance Troubleshooting

When users report slow application performance in migrated VMs.
If replication or vMotion operations are taking longer than expected.

Network Changes

After modifying network configurations, upgrading hardware, or changing service providers.
To confirm that changes haven't negatively impacted HCX tunnel performance.

Capacity Planning

When considering expanding cloud operations or increasing workload migrations.
To determine if current network capacity can handle projected growth.

Regular Health Checks

As part of scheduled maintenance and monitoring routines.
To track performance trends over time and proactively identify degradation.

Multi-site Comparisons

When operating multiple HCX-connected sites to compare performance between different locations.
To optimize traffic routing or resource allocation across sites.

Vendor SLA Verification

To independently verify if service providers are meeting promised network performance metrics.

Application-Specific Requirements

Before deploying bandwidth-intensive applications across hybrid environments.
To ensure specific applications will perform adequately when accessed across HCX tunnels.

Disaster Recovery Testing

During DR drills to confirm that failover and failback operations can occur within acceptable timeframes.

Running a perftest helps administrators gather concrete data on HCX tunnel performance, enabling informed decision-making and ensuring optimal hybrid cloud operations.

Resolution

Using perftest in HCX

The `perftest` command in VMware HCX allows testing to be performed between HCX appliances. This guide will walk you through the process of accessing the HCX Command Line Interface (CLI) and running different perftest commands.

Prerequisites

Access to the VMware HCX Manager (HCM)
Root access to the HCX Manager

Steps

Access the HCX Manager

Connect to the HCX Manager using SSH or the console.
Log in as the root user.

Enter the HCX CLI (CCLI)

Once logged in as root, enter the following command to access the HCX CLI:

ccli

The prompt will appear

ccli>

List Deployed Appliances

To see a list of all deployed HCX appliances, use the following command:

list

This will display a table with information about each appliance, including their IDs, names, and IP addresses.

Select an Appliance

Choose the appliance to run the perftest on. Note its ID from the list.

Use the `go` command followed by the appliance ID to select it:

go <appliance_id>

For example

go 0

The prompt should change to indicate control of the selected appliance similar to

[admin@hcx:SM_7719b17e_b6bac65e-IX-I1]

Now that an appliance is selected perftest can be ran. Here are some examples:

Basic site-to-site performance test

perftest site

Run a site-to-site test with multiple flows

perftest site -P <num_of_flows>

Replace `<num_of_flows>` with the desired number of parallel flows.

Test IPsec tunnel performance

perftest ipsec

Test uplink performance

perftest uplink

Check reachability between appliances

perftest reachability

Run WAN optimization performance test

perftest wanopt

Run all performance tests

perftest all

Interpreting Results

After running a perftest command, there will be a detailed output including:

Test duration
Data transfer amounts
Bandwidth measurements

Pay attention to the "Bandwidth" column in the results, as it shows the achieved throughput for each direction of the test.

Tips

Run tests during off-peak hours to minimize impact on production traffic.
Compare results from different test types (site, ipsec, uplink) to identify potential bottlenecks.
Use the `-P` option with caution, as running too many parallel flows might overload the network or appliances.

Troubleshooting

If a test fails to start, check the connectivity between the appliances.
Ensure that the necessary firewall rules are in place to allow the perftest traffic.
Review the HCX logs for any error messages if tests consistently fail.

Remember to use these tests responsibly and in accordance with organizational policies and network management practices.