Skip to navigationSkip to main contentSkip to footerScaleway DocsSparklesIconAsk our AI
SparklesIconAsk our AI
ProfileOutlineIcon Log inSign up

Load Balancer limitations

This page lists limitations that apply to Scaleway Load Balancers.

You can monitor resource usage in real time from the Cockpit Load Balancer dashboard. Scaleway also provides managed alerts in Cockpit to notify you when a limit is approaching. Refer to the dedicated documentation for more information.

Performance limitations by size

The following limits vary depending on the Load Balancer (LB) size. You can also find them in the console when creating a new Load Balancer.

LB sizeBandwidthMax simultaneous connections
LB-S200 Mbps20,000
LB-M500 Mbps50,000
LB-L1 Gbps160,000
LB-XL4 Gbps3,000,000

Maximum simultaneous connections

If the number of concurrent connections on a frontend, or on the Load Balancer as a whole, reaches the limit for its size, the Load Balancer rejects new connections.

AlertCircleIcon
Important

The maximum simultaneous connections value is a theoretical upper bound. The actual limit depends on factors such as the rate of incoming connections, and the traffic volume processed per connection. As these metrics increase, CPU usage increases accordingly. CPU saturation can become the binding constraint before the connection limit is reached.

Hardware and network limitations

CPU usage

CPU consumption varies depending on the traffic pattern:

  • Load Balancers with a high connection rate (requests per second) consume more CPU than Load Balancers with the same number of active connections at a lower rate.
  • HTTP frontends and backends (especially with SSL encryption/decryption) consume more CPU than basic TCP Load Balancers.

You can monitor CPU usage in the Cockpit Load Balancer dashboard.

Memory usage

Memory consumption depends on the Load Balancer workload. In most cases, CPU saturation occurs before memory saturation.

You can monitor memory usage in the Cockpit Load Balancer dashboard.

Maximum connections per backend server

Each Load Balancer has a limited number of local ports available to establish connections to backend servers. A Load Balancer cannot maintain more than 50,000 TCP sessions for a given server and port combination.

IP addresses

  • A public Load Balancer must have an IPv4 address, it cannot have only an IPv6 address. However, it is possible to allocate a public IPv6 address in addition to an IPv4 address. IPv6 is also supported between your Load Balancer and backend servers.
  • Each Load Balancer supports only one public IPv4 and one IPv6 address at a time.

For more information about restrictions with Load Balancer public IP addresses, see the dedicated documentation.

InformationOutlineIcon
Note

If you reach this limit, add more servers to the relevant backend to distribute the connections.

Configuration limitations

The following limitations apply to the configuration of each Load Balancer.

ResourceLimit
Certificates200
Frontends300
Backends300
ACLs300
Still need help?

Create a support ticketArrowRightIcon
SearchIcon
No Results