Load Balancing Reference

Kong provides multiple ways of load balancing requests to multiple backend services: a straightforward DNS-based method, and a more dynamic ring-balancer that also allows for service registry without needing a DNS server.

DNS-based load balancing

When using DNS-based load balancing, the registration of the backend services is done outside of Kong, and Kong only receives updates from the DNS server.

Every Service that has been defined with a host containing a hostname (instead of an IP address) will automatically use DNS-based load balancing if the name resolves to multiple IP addresses, provided the hostname does not resolve to an upstream name or a name in your DNS hosts file.

The DNS record ttl setting (time to live) determines how often the information is refreshed. When using a ttl of 0, every request will be resolved using its own DNS query. Obviously this will have a performance penalty, but the latency of updates/changes will be very low.

A records

An A record contains one or more IP addresses. Hence, when a hostname resolves to an A record, each backend service must have its own IP address.

Because there is no weight information, all entries will be treated as equally weighted in the load balancer, and the balancer will do a straight forward round-robin.

SRV records

An SRV record contains weight and port information for all of its IP addresses. A backend service can be identified by a unique combination of IP address and port number. Hence, a single IP address can host multiple instances of the same service on different ports.

Because the weight information is available, each entry will get its own weight in the load balancer and it will perform a weighted round-robin.

Similarly, any given port information will be overridden by the port information from the DNS server. If a Service has attributes host=myhost.com and port=123, and myhost.com resolves to an SRV record with 127.0.0.1:456, then the request will be proxied to http://127.0.0.1:456/somepath, as port 123 will be overridden by 456.

DNS priorities

The DNS resolver will start resolving the following record types in order:

  1. The last successful type previously resolved
  2. SRV record
  3. A record
  4. CNAME record

This order is configurable through the dns_order configuration property.

DNS caveats

  • Whenever the DNS record is refreshed a list is generated to handle the weighting properly. Try to keep the weights as multiples of each other to keep the algorithm performance, e.g., 2 weights of 17 and 31 would result in a structure with 527 entries, whereas weights 16 and 32 (or their smallest relative counterparts 1 and 2) would result in a structure with merely 3 entries, especially with a very small (or even 0) ttl value.

  • DNS is carried over UDP with a default limit of 512 Bytes. If there are many entries to be returned, a DNS Server will respond with partial data and set a truncate flag, indicating there are more entries unsent. DNS clients, including Kong’s, will then make a second request over TCP to retrieve the full list of entries.

  • Some nameservers by default do not respond with the truncate flag, but trim the response to be under 512 byte UDP size.

    • Consul is an example. Consul, in its default configuration, returns up to the first three entries only, and does not set the truncate flag to indicate there are remaining entries unsent. Consul includes an option to enable the truncate flag. Please refer to Consul documentation for more information.
  • If a deployed nameserver does not provide the truncate flag, the pool of upstream instances might be loaded inconsistently. The Kong node is effectively unaware of some of the instances, due to the limited information provided by the nameserver. To mitigate this, use a different nameserver, use IP addresses instead of names, or make sure you use enough Kong nodes to still keep all upstream services in use.

  • When the nameserver returns a 3 name error, then that is a valid response for Kong. If this is unexpected, first validate the correct name is being queried for, and second check your nameserver configuration.

  • The initial pick of an IP address from a DNS record (A or SRV) is not randomized. So when using records with a ttl of 0, the nameserver is expected to randomize the record entries.

Ring-balancer

When using the ring-balancer, the adding and removing of backend services will be handled by Kong, and no DNS updates will be necessary. Kong will act as the service registry. Nodes can be added/deleted with a single HTTP request and will instantly start/stop receiving traffic.

Configuring the ring-balancer is done through the upstream and target entities.

  • target: an IP address or hostname with a port number where a backend service resides, e.g. “192.168.100.12:80”. Each target gets an additional weight to indicate the relative load it gets. IP addresses can be in both IPv4 and IPv6 format.

  • upstream: a ‘virtual hostname’ which can be used in a Route host field, e.g., an upstream named weather.v2.service would get all requests from a Service with host=weather.v2.service.

Upstream

Each upstream gets its own ring-balancer. Each upstream can have many target entries attached to it, and requests proxied to the ‘virtual hostname’ (which can be overwritten before proxying, using upstream’s property host_header) will be load balanced over the targets. A ring-balancer has a maximum predefined number of slots, and based on the target weights the slots get assigned to the targets of the upstream.

Adding and removing targets can be done with a simple HTTP request on the Admin API. This operation is relatively cheap. Changing the upstream itself is more expensive as the balancer will need to be rebuilt when the number of slots change for example.

Within the balancer there are the positions (from 1 up to the value defined in the slots attribute), which are randomly distributed on the ring. The randomness is required to make invoking the ring-balancer cheap at runtime. A simple round-robin over the wheel (the positions) will do to provide a well distributed weighted round-robin over the targets, while also having cheap operations when inserting/deleting targets.

Detailed information on adding and manipulating upstreams is available in the upstream section of the Admin API reference.

Target

A target is an IP address/hostname with a port that identifies an instance of a backend service. Each upstream can have many targets. Detailed information on adding and manipulating targets is available in the target section of the Admin API reference.

The targets will be automatically cleaned when there are 10x more inactive entries than active ones. Cleaning will involve rebuilding the balancer, and hence is more expensive than just adding a target entry.

A target can also have a hostname instead of an IP address. In that case the name will be resolved and all entries found will individually be added to the ring balancer, e.g., adding api.host.com:123 with weight=100. The name ‘api.host.com’ resolves to an A record with 2 IP addresses. Then both IP addresses will be added as target, each getting weight=100 and port 123. NOTE: the weight is used for the individual entries, not for the whole!

Would it resolve to an SRV record, then also the port and weight fields from the DNS record would be picked up, and would overrule the given port 123 and weight=100.

The balancer will honor the DNS record’s ttl setting, and queries and updates the balancer when it expires.

Exception: When a DNS record has ttl=0, the hostname will be added as a single target, with the specified weight. Upon every proxied request to this target it will query the nameserver again.

Balancing algorithms

The ring-balancer supports the following load balancing algorithms:

  • round-robin
  • consistent-hashing
  • least-connections
  • latency

By default, a ring-balancer uses the round-robin algorithm, which provides a well-distributed weighted round-robin over the targets.

When using the consistent-hashing algorithm, the input for the hash can be either none, consumer, ip, header, or cookie. When set to none, the round-robin scheme will be used, and hashing will be disabled. The consistent-hashing algorithm supports a primary and a fallback hashing attribute; in case the primary fails (e.g., if the primary is set to consumer, but no Consumer is authenticated), the fallback attribute is used.

Supported hashing attributes are:

  • none: Do not use consistent-hashing; use round-robin instead (default).
  • consumer: Use the Consumer ID as the hash input. If no Consumer ID is available, it will fall back on the Credential ID (for example, in case of an external authentication mechanism like LDAP).
  • ip: Use the originating IP address as the hash input. Review the configuration settings for determining the real IP when using this.
  • header: Use a specified header as the hash input. The header name is specified in either hash_on_header or hash_fallback_header, depending on whether header is a primary or fallback attribute, respectively.
  • cookie: Use a specified cookie with a specified path as the hash input. The cookie name is specified in the hash_on_cookie field and the path is specified in the hash_on_cookie_path field. If the specified cookie is not present in the request, it will be set by the response. Hence, the hash_fallback setting is invalid if cookie is the primary hashing mechanism.

The consistent-hashing algorithm is based on Consistent Hashing, which ensures that when the balancer gets modified by a change in its targets (adding, removing, failing, or changing weights), only the minimum number of hashing losses occur. This maximizes upstream cache hits.

The latency algorithm is based on peak EWMA (exponentially weighted moving average), which ensures that the balancer selects the upstream target by lowest latency (upstream_response_time). This latency is not only TCP connect time, but also includes body response time. In the latency algorithm, the latency is the service response latency because it describes the combined score of service load and network latency. This balancer algorithm is suitable for a single type upstream service. If the backend service has different requests with different body types (for example, video data, audio data, or text data), it causes the latency algorithm loss load balancing function. Before using the latency algorithm, make sure that the QPS of your requests is as large as possible because sharing data between multiple workers in Nginx is difficult and this algorithm only stores data on a single worker. So, the more requests there are, the more data Kong will uncover and the more latency balanced it will be.

The ring-balancer also supports the least-connections algorithm, which selects the target with the lowest number of connections, weighted by the Target’s weight attribute.

For more information on the exact settings see the upstream section of the Admin API reference.

Balancing caveats

The ring-balancer is designed to work both with a single node as well as in a cluster. For the weighted-round-robin algorithm there isn’t much difference, but when using the hash based algorithm it is important that all nodes build the exact same ring-balancer to make sure they all work identical. To do this the balancer must be build in a deterministic way.

  • Do not use hostnames in the balancer as the balancers might/will slowly diverge because the DNS ttl has only second precision and renewal is determined by when a name is actually requested. On top of this is the issue with some nameservers not returning all entries, which exacerbates this problem. So when using the hashing approach in a Kong cluster, add target entities only by their IP address, and never by name.

  • When picking your hash input make sure the input has enough variance to get to a well distributed hash. Hashes will be calculated using the CRC-32 digest. So for example, if your system has thousands of users, but only a few consumers, defined per platform (e.g. 3 consumers: Web, iOS and Android) then picking the consumer hash input will not suffice, using the remote IP address by setting the hash to ip would provide more variance in the input and hence a better distribution in the hash output. However, if many clients will be behind the same NAT gateway (e.g. in call center), cookie will provide a better distribution than ip.