Infrastructure / Application Availability / Progress Kemp LoadMaster
Update a server without anyone noticing.
Progress Kemp LoadMaster, formerly Kemp.A load balancer isn’t primarily designed to handle traffic spikes. In a company with 200 employees, its purpose is to ensure that a server restart, update, or outage goes unnoticed from the user’s desk. That’s its actual use—and it’s the one product descriptions mention the least.
Three Situations Where Not Having a Load Balancer Can Cost You
None of these three represents a peak in traffic. These are routine operating conditions, and they occur every week.
The evening when the patches need to be installed
Without a load balancer, updating an application server means shutting it down, which interrupts service. The update is postponed until Saturday, then until the following Saturday, and the server eventually falls a year behind.
What can't be stopped doesn't get updated anymore.
The morning a server stops responding
When a single server hosts the service, a failure results in a total and immediate outage. The time it takes to restore service means no email, no intranet, and no remote access for everyone at the same time.
A single server is a recipe for disaster.
The day a certificate expires
An expired certificate cuts off access without warning and without displaying an explicit error message. When they’re spread across six servers and three different consoles, there’s always one missing—and it’s the one that goes down on a Friday.
A forgotten certificate is the most common problem.
Four ways to install it, same function
You should choose the format based on what you're already using, not on advertised performance. All four offer the same features; what differs is where the installation is hosted and how it's billed.
Virtual LoadMaster
A virtual machine on your existing platform— VMware, Hyper-V, KVM, or Xen. This is the most common format, and the one that can be set up the fastest when the platform is already in place.
Hardware LoadMaster
Dedicated hardware designed to be installed in a rack. This approach is warranted when the required throughput exceeds what a virtual machine can comfortably handle, or when the service must not depend on any virtualization platform.
Cloud LoadMaster
The same setup in Microsoft Azure or Amazon Web Services, billed on a pay-as-you-go basis or with your own license. Useful as soon as some of your applications have moved off-premises.
LoadMaster Ingress Controller
The entry point for external traffic to a Kubernetes cluster, which is automatically configured as services are created and destroyed.
Microsoft applications, along with their configuration templates
That’s what sets this product apart in our industry: the vendor publishes deployment guides and ready-to-import configuration templates for the applications we install most often. Configuring a service is no longer a matter of guesswork—it’s just a matter of loading it.
Exchange Server
Distribution of web traffic, mobile connections, and email traffic across multiple servers, with status monitoring for each server. When a server is taken offline for maintenance, it stops receiving connections without anyone losing their session.
Published Configuration TemplateSharePoint Server
Load balancing across front-end servers, centralized encryption management, and user session maintenance. This also makes it possible to add a front-end server without interruption when usage increases.
Published Configuration TemplateRemote Desktop Services
User distribution across session servers and a user’s reconnection to their own open session. For a service used around the clock, this is the difference between seamless maintenance and an announced evening outage.
Published Configuration TemplateAD FS
Distribution of the federated authentication service across multiple servers, along with the health checks specific to that service. When authentication fails, everything else fails with it: this is the service that most justifies having a second machine.
Published Configuration TemplateDecisions regarding email and external hosting must be made first
Load balancing for an email server only makes sense if that server is to remain on your premises. This issue must be decided before sizing the system, and we devote two full pages to it.
What stays on your premises, and what goesHow We Do It
There are four steps, and the second one determines the budget. It takes place before the quote is prepared, because it’s this step that determines which license you need—and sometimes that none is necessary.
What Must Never Stop
A list of applications whose downtime is noticeable, along with the amount of downtime your business can tolerate for each one. This is a management decision, not a technical setting.
Measuring Actual Traffic
We measure actual traffic during peak hours rather than estimating it. The results are often surprising: an intranet with 200 users consumes far less bandwidth than one might imagine.
This measure prevents you from buying a license that's two sizes too big, and helps you determine when a license-free version is sufficient.
Installation and Failure Testing
Setup, importing configuration templates, then real-world testing: we shut down a server while someone is working, and we watch what happens on the screen.
A high-availability system that has never been tested is not a high-availability system.
Certificate Tracking
Certificates are consolidated and monitored in one place, with an alert sent before they expire. This is the part that everyone puts off, and the one that causes the most avoidable disruptions.
What You Need to Know Before Making a Decision
Four questions that come up in every project, along with our answers— even when they don’t go our way.
Is unlicensed publishing enough?
Yes, in certain situations, and we do mention that. The vendor offers a license-free version, called Free LoadMaster, which is limited to 20 megabits per second at Level 7. For an internal intranet or a test environment, that’s often enough.
Its two limitations are almost always decisive: no automatic failover to a secondary device, and application filtering limited to the engine, without commercial filter packages. As soon as the service needs to survive a failure of the load balancer itself, this solution is no longer suitable.
Has the name Kemp disappeared?
No. Progress Kemp acquired Kemp in November 2021, and the product is now called LoadMaster. The Kemp name has been retained; the product line has not been merged into anything else.
Some license numbers, however, have been removed: if you are using a model purchased before 2025, we first check until when it will receive updates.
Do you need a second set of equipment?
A single load balancer becomes the sole point on which everything else depends. Doubling it eliminates this dependency and costs one additional license.
The answer lies in a single line from the Stage 01 report: how long does the activity last without the relevant app? If the answer is “half a day,” a single device is more than sufficient.
Who manages it after that?
A central console consolidates installations, measurements, and certificates. It makes a big difference, especially when there are multiple pieces of equipment or multiple sites.
Your teams can handle this, and we'll train them. We can also manage the operations, with availability schedules specified in the contract.
The questions that come with it
An app's availability is never determined by just one factor. Here are the three issues that most often arise at the same time.
Exchange, and what becomes of him
The first application for which a load balancer is installed—and the one you need to decide first whether it will remain on your premises.
AppetizersWhat comes in through the network and email
Application-level filtering by a load balancer does not replace a network firewall or message monitoring: these are three separate functions.
The FarmMaking sure all of this lasts over time
Updates, certificate expiration dates, regular failure tests. What happens to a system when no one takes care of it.
Let's start by taking measurements.
The actual throughput of your applications during peak hours, and the amount of downtime each can tolerate. These two figures determine almost everything else— including the answer that “the unlicensed version is enough for you.”
Three offices in French-speaking Switzerland: Renens, Sion , and Châtel-Saint-Denis. What we offer is the opportunity to meet the engineers who will do the work.
Shutting Down a Server While Someone Is Working
Not outside of business hours, not in a test environment. On the live service, with a user sitting in front of their screen, and watching what they see.
That's the only way to know how well a high-availability system performs, and we do it together with you.