Functionality, or simply NGSIKafkaSink is a sink designed to persist NGSI-like context data events within a Apache Kafka deployment. Usually, such a context data is notified by a Orion Context Broker instance, but could be any other system speaking the NGSI language.

Independently of the data generator, NGSI context data is always transformed into internal NGSIEvent objects at Cygnus sources. In the end, the information within these events must be mapped into specific Kafka data structures at the Cygnus sinks.

Next sections will explain this in detail.


Mapping NGSI events to NGSIEvent objects

Notified NGSI events (containing context data) are transformed into NGSIEvent objects (for each context element a NGSIEvent is created; such an event is a mix of certain headers and a ContextElement object), independently of the NGSI data generator or the final backend where it is persisted.

This is done at the cygnus-ngsi Http listeners (in Flume jergon, sources) thanks to NGSIRestHandler. Once translated, the data (now, as NGSIEvent objects) is put into the internal channels for future consumption (see next section).


Mapping NGSIEvents to Kafka data structures

Apache Kafka organizes the data in topics (a category or feed name to which messages are published). Such organization is exploited by NGSIKafkaSink each time a NGSIEvent is going to be persisted.


Topics naming conventions

A Kafka topic is created (number of partitions 1) if not yet existing depending on the configured data model:

  • Data model by service (data_model=dm-by-service). As the data model name denotes, the notified FIWARE service (or the configured one as default in NGSIRestHandler) is used as the name of the topic. This allows the data about all the NGSI entities belonging to the same service is stored in this unique topic.
  • Data model by service path (data_model=dm-by-service-path). As the data model name denotes, the notified FIWARE service path (or the configured one as default in NGSIRestHandler) is used as the name of the topic. This allows the data about all the NGSI entities belonging to the same service path is stored in this unique topic. The only constraint regarding this data model is the FIWARE service path cannot be the root one (/).
  • Data model by entity (data_model=dm-by-entity). For each entity, the notified/default FIWARE service path is concatenated to the notified entity ID and type in order to compose the topic name. If the FIWARE service path is the root one (/) then only the entity ID and type are concatenated.
  • Data model by attribute (data_model=dm-by-attribute). For each entity's attribute, the notified/default FIWARE service path is concatenated to the notified entity ID and type and to the notified attribute name in order to compose the topic name. If the FIWARE service path is the root one (/) then only the entity ID and type and the attribute name and type are concatenated.

It must be said there is no known character set accepted and/or forbidden for Kafka. Anyway, certaing encoding is applied.

The following table summarizes the topic name composition:

FIWARE service path dm-by-service dm-by-service-path dm-by-entity dm-by-attribute
/ <svc> <svc>xffffx002f <svc>xffffx002fxffff<entityId>xffff<entityType> <svc>xffffx002fxffff<entityId>xffff<entityType>xffff<attrName>
/<svcPath> <svc> <svc>xffffx002f<svcPath> <svc>xffffx002f<svcPath>xffff<entityId>xffff<entityType> <svc>xffffx002f<svcPath>xffff<entityId>xffff<entityType>xffff<attrName>

Please observe the concatenation of entity ID and type is already given in the notified_entities/grouped_entities header values (depending on using or not the grouping rules, see the Configuration section for more details) within the NGSIEvent.



NGSIEvents structure is stringified as a Json object containing an array of headers and another object containing the Json data as it is notified by the NGSI-like source.




Assuming the following NGSIEvent is created from a notified NGSI context data (the code below is an object representation, not any real data format):



Topic names

The topic names will be, depending on the configured data model, the following ones:

FIWARE service path dm-by-service dm-by-service-path dm-by-entity dm-by-attribute
/ vehicles vehiclesxffffx002f vehiclesxffffx002fxffffcar1_car vehiclesxffffx002fxffffcar1xffffcarxffffspeed
/4wheels vehicles vehiclesxffffx002f4wheels vehiclesxffffx002f4wheelsxffffcar1xffffcar vehiclesxffffx002f4wheelsxffffcar1xffffcarxffffspeed



Let's assume a topic name vehiclesxffffx002f4wheelsxffffcar1xffffcarxffffspeed (data model by attribute, non-root service path). The data stored within this topic would be:

$ bin/ --zookeeper localhost:2181 --topic vehiclesxffffx002f4wheelsxffffcar1xffffcarxffffspeed --from-beginning
$ bin/ --zookeeper localhost:2181 --topic vehicles_4wheels_car1_car_oil_level --from-beginning


Administration guide


NGSIKafkaSink is configured through the following parameters:

Parameter Mandatory Default value Comments
type yes N/A Must be
channel yes N/A
enable_grouping no false true or false. Check this link for more details.
enable_name_mappings no false true or false. Check this link for more details.
enable_lowercase no false true or false.
data_model no dm-by-entity dm-by-service, dm-by-service-path, dm-by-entity or dm-by-attribute.
broker_list no localhost:9092 Comma-separated list of Kafka brokers (a broker is defined as host:port).
zookeeper_endpoint no localhost:2181 Zookeeper endpoint needed to create Kafka topics, in the form of host:port.
partitions no 1 Number of partitions for a topic.
replication_factor no 1 For a topic with replication factor N, Kafka will tolerate N-1 server failures without losing any messages committed to the log. Replication factor must be less than or equal to the number of brokers created.
batch_size no 1 Number of events accumulated before persistence.
batch_timeout no 30 Number of seconds the batch will be building before it is persisted as it is.
batch_ttl no 10 Number of retries when a batch cannot be persisted. Use 0 for no retries, -1 for infinite retries. Please, consider an infinite TTL (even a very large one) may consume all the sink's channel capacity very quickly.
batch_retry_intervals no 5000 Comma-separated list of intervals (in miliseconds) at which the retries regarding not persisted batches will be done. First retry will be done as many miliseconds after as the first value, then the second retry will be done as many miliseconds after as second value, and so on. If the batch_ttl is greater than the number of intervals, the last interval is repeated.

A configuration example could be:

cygnus-ngsi.sinks = kafka-sink
cygnus-ngsi.channels = kafka-channel
cygnus-ngsi.sinks.kafka-sink.type = = kafka-channel
cygnus-ngsi.sinks.kafka-sink.enable_grouping = false
cygnus-ngsi.sinks.kafka-sink.enable_lowercase = false
cygnus-ngsi.sinks.kafka-sink.enable_name_mappings = false
cygnus-ngsi.sinks.kafka-sink.data_model = dm-by-entity
cygnus-ngsi.sinks.kafka-sink.broker_list = localhost:9092
cygnus-ngsi.sinks.kafka-sink.zookeeper_endpoint = localhost:2181
cygnus-ngsi.sinks.kafka-sink.partitions = 5
cygnus-ngsi.sinks.kafka-sink.replication_factor = 1
cygnus-ngsi.sinks.kafka-sink.batch_size = 100
cygnus-ngsi.sinks.kafka-sink.batch_timeout = 30
cygnus-ngsi.sinks.kafka-sink.batch_ttl = 10
cygnus-ngsi.sinks.kafka-sink.batch_retry_intervals = 5000


Use cases

Use NGSIKafkaSink if you want to integrate OrionContextBroker with a Kafka-based consumer, as a Storm real-time application.


Important notes

About batching

As explained in the programmers guide, NGSIKafkaSink extends NGSISink, which provides a built-in mechanism for collecting events from the internal Flume channel. This mechanism allows extending classes have only to deal with the persistence details of such a batch of events in the final backend.

What is important regarding the batch mechanism is it largely increases the performance of the sink, because the number of writes is dramatically reduced. Let's see an example, let's assume a batch of 100 NGSIEvents. In the best case, all these events regard to the same entity, which means all the data within them will be persisted in the same Kafka topic. If processing the events one by one, we would need 100 writes to Kafka; nevertheless, in this example only one write is required. Obviously, not all the events will always regard to the same unique entity, and many entities may be involved within a batch. But that's not a problem, since several sub-batches of events are created within a batch, one sub-batch per final destination Kafka topic. In the worst case, the whole 100 entities will be about 100 different entities (100 different Kafka topics), but that will not be the usual scenario. Thus, assuming a realistic number of 10-15 sub-batches per batch, we are replacing the 100 writes of the event by event approach with only 10-15 writes.

The batch mechanism adds an accumulation timeout to prevent the sink stays in an eternal state of batch building when no new data arrives. If such a timeout is reached, then the batch is persisted as it is.

Regarding the retries of not persisted batches, a couple of parameters is used. On the one hand, a Time-To-Live (TTL) is used, specifing the number of retries Cygnus will do before definitely dropping the event. On the other hand, a list of retry intervals can be configured. Such a list defines the first retry interval, then se second retry interval, and so on; if the TTL is greater than the length of the list, then the last retry interval is repeated as many times as necessary.

By default, NGSIKafkaSink has a configured batch size and batch accumulation timeout of 1 and 30 seconds, respectively. Nevertheless, as explained above, it is highly recommended to increase at least the batch size for performance purposes. Which are the optimal values? The size of the batch it is closely related to the transaction size of the channel the events are got from (it has no sense the first one is greater then the second one), and it depends on the number of estimated sub-batches as well. The accumulation timeout will depend on how often you want to see new data in the final storage. A deeper discussion on the batches of events and their appropriate sizing may be found in the performance document.


About the encoding

Cygnus applies this specific encoding tailored to Kafka data structures:

  • Alphanumeric characters are not encoded.
  • Numeric characters are not encoded.
  • Underscore character, _, is not encoded.
  • Hyphen character, -, is not encoded.
  • Dot character, ., is not encoded.
  • Equals character, =, is encoded as xffff.
  • All other characters, including the slash in the FIWARE service paths, are encoded as a x character followed by the Unicode of the character.
  • User defined strings composed of a x character and a Unicode are encoded as xx followed by the Unicode.
  • xffff is used as concatenator character.


Programmers guide

NGSIKafkaSink class

As any other NGSI-like sink, NGSIKafkaSink extends the base NGSISink. The methods that are extended are:

void persistBatch(Batch batch) throws Exception;

A Batch contains a set of NGSIEvent objects, which are the result of parsing the notified context data events. Data within the batch is classified by destination, and in the end, a destination specifies the Kafka topic where the data is going to be persisted. Thus, each destination is iterated in order to compose a per-destination data string to be persisted thanks to the KafkaProducer.

public void start();

KafkaProducer is created. This must be done at the start() method and not in the constructor since the invoking sequence is NGSIKafkaSink() (contructor), configure() and start().

public void configure(Context);

A complete configuration as the one described above is read from the given Context instance.