Microsoft's Azure Traffic Manager operates at the DNS layer to quickly and efficiently direct incoming DNS requests based on the routing method of your choice.
N/A
HAProxy Community Edition
Score 9.3 out of 10
N/A
HAProxy Community Edition is a free, open source reverse-proxy offering high availability, load balancing, and proxying for TCP and HTTP-based applications. It is presented as suited for very high traffic web sites.
$0
Kemp LoadMaster
Score 8.9 out of 10
N/A
LoadMaster from Kemp Technologies in New York is an application delivery controller.
Setup is easier and more straight forward on the Kemp. While I am a technical person and can setup HAProxy for the client's needs, it would be difficult for the client to manage the configuration themselves.
Azure Traffic Manager is a great product, if you have multiple sites hosting similar services (Primary and DR), and you want to ensure that users are directed to the DR in case of a primary datacenter failure, [Azure] Traffic Manager does this very nicely. If you have a service hosted across multiple regions/datacenters and you want to balance the inbound load between the regions, [Azure] Traffic Manager does this very well, of course such scenario would require a database replication or something like Cosmos-DB in the backend [Azure Traffic Manager] is also well suited for inbound traffic with multiple IPs, you can fail-over traffic from one inbound IP to another based on its availability, or if you have multiple internet connections that you want to balance the load across, it does this pretty nicely too.
It prevents a single server failure from being a downtime event by adding redundancy to every layer of your architecture. A load balancer facilitates redundancy for the backend layer (web/app servers), but for a true high availability setup, you need to have redundant load balancers as well. So it is well suited for all production related servers and less suited for individual servers that do not require redundancy.
Loadmaster is very powerful and flexible load balancer. Variety of options allows to create a complex network of rules and routes. During our website rollout Loadmaster allowed us to run multiple generations of the website simultaneously and seamlessly by doing the content switching on the fly. Powerful API allows easy integration into any development lifecycle.
Where the LoadBalancer excels is the multiple levels at which you can load balance servers. We currently use layer 7 LB, but others are available as well.
I particularly like the ability for the LB to know when servers are down, and if all are offline then you can create a redirect to a static HTML page or some other destination that is more informative.
The ease in which a service can be created and deployed using already pre-canned templates makes it a very convenient setup process.
Traffic View is a great feature, but doesn't work very well, sometimes it gets stuck and stops loading traffic view data
Automatic probing for endpoints sometimes gets stuck too, I would recommend a technique to test the endpoint in real time from Azure Portal
Traffic View heatmap is buggy and doesn't point correctly to locations
Traffic View portal doesn't show source countries (Shows coordinates) it would be much more helpful to have coordinates auto-translated to geolocations/countries
A few, rare times each year, HAProxy CPU utilization spikes to 100% and server has to be rebooted - this may be related to HAProxy OR it could be an external factor causing this.
It is very easy to use. I was able to find a lot of documents for it on the internet. Very good community support. There are lots of examples available to try. We mostly use a command-line user interface to interact with it. The CLI is also super easy to use and very easy to interact with
Kemp makes it very easy to setup, configure and manage the LoadMaster without needing a lot of help from their engineers. The interface is very easy to understand and intuitive to use. We like how it is not complicated - I can easily have one of my techs login and they can figure out how to setup/configure virtual services for load balancing without needing a manual or tech support.
We haven't used customer support. We mostly used the community version. We build a multi-node HAProxy cluster with HA to the proxy itself using opensource plugins available. With the support available on the internet and the documents available we don't need to use much customer support.
Support has been easy to deal with; I have only need[ed] to contact them a few times during setup. Once its been in place and operational, we have not need[ed] to mess with the system [which] is a huge advantage. I like system[s] that do not break and require constant attention in a production environment.
Amazon Route 53 Traffic Flow does what [Azure] Traffic Manager does, however, in Azure Configuration is separated between Azure DNS Zones (For DNS Zone Management) and [Azure] Traffic Manager for DNS Traffic Management and Load Balancing, Route 53 in a unified product for DNS Traffic Management using Traffic Flow and DNS Zone Management. Route 53 does a great job, however, we found it to be a little bit more complex to setup than [Azure] Traffic Manager, Setting up traffic manager is pretty easy even for the first time, and getting the best out of it is relatively simple.
We chose HA Proxy because it is cheaper than a hardware balancer, it is an open-source solution with a large community behind it and with constant updates. It also allows custom scripts according to needs.HA Proxy is a solution used in many internet sites like GitHub, Reddit, Twitter, and Tuenti.
We chose Kemp LoadMaster because it is 1/10 the price of the competition and MUCH easier to deploy and configure and WORKS. We have had ZERO issues with the product since installation. Their engineers and their sales team have both reached out post-install to check in and see if everything is working as expected.
Significantly lower investment vs competitors. In the case of F5s we have Virtual Editions so we're paying for the hardware to run it on top of the several thousand dollar licenses that are required for each pair and we currently have a pair of F5s per client so there's a huge potential for cost savings there.
Requires our network engineers to learn a new skill or our Systems engineers to take on the responsibility of managing the load balancers. It's not a huge difference either way, but it does impact the way we have done business in the past.
We used Kemp LoadMaster for many projects. For a lot of customers, load balancers were too expensive or too complicated before we introduced Kemp products.
It's not a overly complicated product, so we were able to train many engineers on it and have them get a certification.
