Document updated on Jan 14, 2022
Service Discovery
Service discovery (sd
) is an attribute in the backend
section that enables KrakenD to detect and locate services automatically on your enterprise network.
The chosen service discovery strategy determines how to retrieve (statically or dynamically) the final list of IPs, hostnames, or services pointing to your backends. If your host list is dynamic, you can use an external service discovery provider and let KrakenD interact with it to get the hosts. If your host list is static (it doesn’t change) or you use a service name or an external balancer, you can use static
resolution and directly use the values provided under host[]
.
KrakenD must be in a network where can reach any declared hosts. When there is more than one host, KrakenD load balances the connections to the hosts in the list.
The possible values for sd
are:
static
: For static resolutiondns
: For DNS SRV resolution
See below.
Static resolution
The static
resolution is the default service discovery strategy. It implies that you write directly in the configuration the protocol plus the service name, hosts or IPs you want to connect to.
The static
resolution uses a list of hosts to load balance (in a Round Robin fashion) all servers in the list, and you should expect more or less an equivalent number of hits on each backend. However, if you use a Kubernetes service, then it load-balances itself so that you will need one entry only.
You don’t need to declare anything other than the host
list to use static resolution. However, you can also add the "sd": "static"
property in the backend configuration. It is the default value when sd
is not declared). Example:
{
"backend": [
{
"url_pattern": "/some-url",
"sd": "static",
"host": [
"http://my-service-01.api.com:9000",
"http://my-service-02.api.com:9000"
]
}
]
}
DNS SRV Service Discovery (Kubernetes/Consul)
The DNS SRV
(RFC) is a market standard used by systems such as Kubernetes, Mesos, Haproxy, Nginx plus, AWS ECS, Linkerd, and more. An SRV entry is a custom DNS record used to establish connections between services. When KrakenD needs to know the location of a specific service, it will search for a related SRV record.
The format of the SRV
record is as follows:
_service._proto.name. ttl IN SRV priority weight port target
Example. A service running on port 8000
with maximum priority (0
) and a weight 5
):
_api._tcp.example.com. 86400 IN SRV 0 5 8000 foo.example.com.
To integrate Consul, Kubernetes or any other DNS SRV
compatible system as the Service Discovery, you only need to set two keys:
"sd": "dns"
: To set service discovery = DNS SRV"host": []
: The list of all the names providing the resolution
Add these keys in the backend
section of your configuration. If there is another host
key in the root level of the configuration, you don’t need to declare it here if the value is the same.
For instance:
{
"backend": [
{
"url_pattern": "/foo",
"sd": "dns",
"host": [
"api-catalog.service.consul.srv"
],
"disable_host_sanitize": true
}
]
}
Priority and weight importance on balancing
The SRV
record provides the hostname, port, priority, and weight that KrakenD uses to balance. KrakenD reads these values every 30 secondsand generates an internal balancing list.
The balancing list honors the distribution described in the SRV
records. Nevertheless, KrakenD will use only the records with the lower priority. So, for instance, if you have 5 servers with priority 0
and another with priority 2
, the latter won’t be included in the balancing.
As per the weights, KrakenD distributes the traffic in the proportion they represent. To be memory and space-efficient, KrakenD compacts and normalizes the final list of weights if needed. It’s essential to be aware that KrakenD will remove servers with a weight orders of magnitude inferior (under 1% of the total representation) from the final list as they are negligible.
Some examples on the space optimization and removal of neglectable items:
SRV
passes the weights of 3 servers with values[100 500 1000]
and KrakenD builds a list[1 5 10]
SRV
passes[25 10000 1000]
and KrakenD compacts it as[0 10 1]
. The server with a weight of25
is removed as it is vastly inferior to the rest (0,2% weight).SRV
passes[25 1000 10000 65535]
becomes[0 1 13 85]
. Again,25
drops. The rest are converted to a lower value with the same distribution proportion.