Dividing the operation and balancing the load between several machines will increase the performance of the entire solution.
Why to use load balancer?
Performance optimisation
Faster server response
Optimal distribution of requests to individual servers speeds up their processing and thus the response of the entire infrastructure.
Automatic failover
If one machine fails, the traffic will automatically switch to other servers, whether it is a web server, DB cluster or kubernetes.
We offer a load balancer for every service that has a public IP address. Its purchase pays off at least two services, between which the load is to be balanced. We also deploy load balancing to customers with third-party infrastructure or in our own data centre.
HAproxy Load Balancer and its features
In order to balance the performance between the compute-nodes, open-source load balancer, which runs on the cloud, is used, so that its high availability is facilitated.
Reduction of provisioning costs
“Events” based architecture (so-called event-driven) minimises RAM and CPU . The single-buffering function skips the process of copying data between read and write operations, allowing HAProxy to process hundreds of requests without delay.
Performance function distribution
Operational distribution between servers thus might be configure globally or based on the nodes. If you need to perform maintenance on one of the servers, you can redirect traffic to another server. Therefore there is no outage during the restart.
Advanced HTTP analysis
HAProxy can sort (parse) traffic according to HTTP headers and perform load balancing on that basis. For example, requests from individual clients are always sent to the same server, which results in processing faster.
Monitoring panel
The server monitoring panel alerts you to the status of front-end and back-end nodes, enabling the detection of errors and bugs.
Geo Load balancer
For customers with high requirements for availability, we offer the so-called geo load balancig. These are two software load balancers in separate locations (Prague and Brno). Is one of the sites fails, traffic is automatically redirected to the other.