有特点的流处理引擎NiFi

前面写了flink的文章，其实流处理不止有flink、storm、spark streaming，说实话这些其实都是比较传统的流处理框架。今天介绍一个大家不一定用得很多，但是却很有特点的东西，NiFi

NiFi的来源

Apache NiFi项目，它是一种实时数据流处理 系统，在去年由美国安全局（NSA）开源并进入Apache社区，NiFi初始的项目名称是Niagarafiles。当NiFi项目开源之后，一些早先在NSA的开发者们创立了初创公司Onyara，Onyara随之继续NiFi项目的开发并提供相关的支持。Hortonworks公司收购了Onyara并将其开发者整合到自己的团队中,形成HDF（Hortonworks Data Flow）平台。

NiFi的特点

下面是官方的一些关键能力介绍，可以认真看看：

Apache NiFi supports powerful and scalable directed graphs of data routing, transformation, and system mediation logic. Some of the high-level capabilities and objectives of Apache NiFi include:

• Web-based user interface
• Seamless experience between design, control, feedback, and monitoring
• Highly configurable
• Loss tolerant vs guaranteed delivery
• Low latency vs high throughput
• Dynamic prioritization
• Flow can be modified at runtime
• Back pressure
• Data Provenance
• Track dataflow from beginning to end
• Designed for extension
• Build your own processors and more
• Enables rapid development and effective testing
• Secure
• SSL, SSH, HTTPS, encrypted content, etc...
• Multi-tenant authorization and internal authorization/policy management

总结来说，做为一个流处理引擎，NiFi的核心差异化能力主要有两点：

丰富的算子

整合了大量数据源的处理能力，详细的可以登录nifi官网（https://nifi.apache.org/docs.html）详细看各个算子的能力，下面列一列算子，让大家有个感觉，，还是相当丰富的。

Processors

• AttributeRollingWindow 1.3.0
• AttributesToJSON 1.3.0
• Base64EncodeContent 1.3.0
• CaptureChangeMySQL 1.3.0
• CompareFuzzyHash 1.3.0
• CompressContent 1.3.0
• ConnectWebSocket 1.3.0
• ConsumeAMQP 1.3.0
• ConsumeEWS 1.3.0
• ConsumeIMAP 1.3.0
• ConsumeJMS 1.3.0
• ConsumeKafka 1.3.0
• ConsumeKafka_0_10 1.3.0
• ConsumeKafkaRecord_0_10 1.3.0
• ConsumeMQTT 1.3.0
• ConsumePOP3 1.3.0
• ConsumeWindowsEventLog 1.3.0
• ControlRate 1.3.0
• ConvertAvroSchema 1.3.0
• ConvertAvroToJSON 1.3.0
• ConvertAvroToORC 1.3.0
• ConvertCharacterSet 1.3.0
• ConvertCSVToAvro 1.3.0
• ConvertExcelToCSVProcessor 1.3.0
• ConvertJSONToAvro 1.3.0
• ConvertJSONToSQL 1.3.0
• ConvertRecord 1.3.0
• CreateHadoopSequenceFile 1.3.0
• DebugFlow 1.3.0
• DeleteDynamoDB 1.3.0
• DeleteGCSObject 1.3.0
• DeleteHDFS 1.3.0
• DeleteS3Object 1.3.0
• DeleteSQS 1.3.0
• DetectDuplicate 1.3.0
• DistributeLoad 1.3.0
• DuplicateFlowFile 1.3.0
• EncryptContent 1.3.0
• EnforceOrder 1.3.0
• EvaluateJsonPath 1.3.0
• EvaluateXPath 1.3.0
• EvaluateXQuery 1.3.0
• ExecuteFlumeSink 1.3.0
• ExecuteFlumeSource 1.3.0
• ExecuteProcess 1.3.0
• ExecuteScript 1.3.0
• ExecuteSQL 1.3.0
• ExecuteStreamCommand 1.3.0
• ExtractAvroMetadata 1.3.0
• ExtractCCDAAttributes 1.3.0
• ExtractEmailAttachments 1.3.0
• ExtractEmailHeaders 1.3.0
• ExtractGrok 1.3.0
• ExtractHL7Attributes 1.3.0
• ExtractImageMetadata 1.3.0
• ExtractMediaMetadata 1.3.0
• ExtractText 1.3.0
• ExtractTNEFAttachments 1.3.0
• FetchAzureBlobStorage 1.3.0
• FetchDistributedMapCache 1.3.0
• FetchElasticsearch 1.3.0
• FetchElasticsearch5 1.3.0
• FetchElasticsearchHttp 1.3.0
• FetchFile 1.3.0
• FetchFTP 1.3.0
• FetchGCSObject 1.3.0
• FetchHBaseRow 1.3.0
• FetchHDFS 1.3.0
• FetchParquet 1.3.0
• FetchS3Object 1.3.0
• FetchSFTP 1.3.0
• FuzzyHashContent 1.3.0
• GenerateFlowFile 1.3.0
• GenerateTableFetch 1.3.0
• GeoEnrichIP 1.3.0
• GetAzureEventHub 1.3.0
• GetCouchbaseKey 1.3.0
• GetDynamoDB 1.3.0
• GetFile 1.3.0
• GetFTP 1.3.0
• GetHBase 1.3.0
• GetHDFS 1.3.0
• GetHDFSEvents 1.3.0
• GetHDFSSequenceFile 1.3.0
• GetHTMLElement 1.3.0
• GetHTTP 1.3.0
• GetIgniteCache 1.3.0
• GetJMSQueue 1.3.0
• GetJMSTopic 1.3.0
• GetKafka 1.3.0
• GetMongo 1.3.0
• GetSFTP 1.3.0
• GetSNMP 1.3.0
• GetSolr 1.3.0
• GetSplunk 1.3.0
• GetSQS 1.3.0
• GetTCP 1.3.0
• GetTwitter 1.3.0
• HandleHttpRequest 1.3.0
• HandleHttpResponse 1.3.0
• HashAttribute 1.3.0
• HashContent 1.3.0
• IdentifyMimeType 1.3.0
• InferAvroSchema 1.3.0
• InvokeHTTP 1.3.0
• InvokeScriptedProcessor 1.3.0
• ISPEnrichIP 1.3.0
• JoltTransformJSON 1.3.0
• ListAzureBlobStorage 1.3.0
• ListDatabaseTables 1.3.0
• ListenBeats 1.3.0
• ListenHTTP 1.3.0
• ListenLumberjack 1.3.0
• ListenRELP 1.3.0
• ListenSMTP 1.3.0
• ListenSyslog 1.3.0
• ListenTCP 1.3.0
• ListenUDP 1.3.0
• ListenWebSocket 1.3.0
• ListFile 1.3.0
• ListFTP 1.3.0
• ListGCSBucket 1.3.0
• ListHDFS 1.3.0
• ListS3 1.3.0
• ListSFTP 1.3.0
• LogAttribute 1.3.0
• LogMessage 1.3.0
• LookupAttribute 1.3.0
• LookupRecord 1.3.0
• MergeContent 1.3.0
• ModifyBytes 1.3.0
• ModifyHTMLElement 1.3.0
• MonitorActivity 1.3.0
• Notify 1.3.0
• ParseCEF 1.3.0
• ParseEvtx 1.3.0
• ParseSyslog 1.3.0
• PartitionRecord 1.3.0
• PostHTTP 1.3.0
• PublishAMQP 1.3.0
• PublishJMS 1.3.0
• PublishKafka 1.3.0
• PublishKafka_0_10 1.3.0
• PublishKafkaRecord_0_10 1.3.0
• PublishMQTT 1.3.0
• PutAzureBlobStorage 1.3.0
• PutAzureEventHub 1.3.0
• PutCassandraQL 1.3.0
• PutCloudWatchMetric 1.3.0
• PutCouchbaseKey 1.3.0
• PutDatabaseRecord 1.3.0
• PutDistributedMapCache 1.3.0
• PutDynamoDB 1.3.0
• PutElasticsearch 1.3.0
• PutElasticsearch5 1.3.0
• PutElasticsearchHttp 1.3.0
• PutElasticsearchHttpRecord 1.3.0
• PutEmail 1.3.0
• PutFile 1.3.0
• PutFTP 1.3.0
• PutGCSObject 1.3.0
• PutHBaseCell 1.3.0
• PutHBaseJSON 1.3.0
• PutHDFS 1.3.0
• PutHiveQL 1.3.0
• PutHiveStreaming 1.3.0
• PutHTMLElement 1.3.0
• PutIgniteCache 1.3.0
• PutJMS 1.3.0
• PutKafka 1.3.0
• PutKinesisFirehose 1.3.0
• PutKinesisStream 1.3.0
• PutLambda 1.3.0
• PutMongo 1.3.0
• PutParquet 1.3.0
• PutRiemann 1.3.0
• PutS3Object 1.3.0
• PutSFTP 1.3.0
• PutSlack 1.3.0
• PutSNS 1.3.0
• PutSolrContentStream 1.3.0
• PutSplunk 1.3.0
• PutSQL 1.3.0
• PutSQS 1.3.0
• PutSyslog 1.3.0
• PutTCP 1.3.0
• PutUDP 1.3.0
• PutWebSocket 1.3.0
• QueryCassandra 1.3.0
• QueryDatabaseTable 1.3.0
• QueryDNS 1.3.0
• QueryElasticsearchHttp 1.3.0
• QueryRecord 1.3.0
• QueryWhois 1.3.0
• ReplaceText 1.3.0
• ReplaceTextWithMapping 1.3.0
• ResizeImage 1.3.0
• RouteHL7 1.3.0
• RouteOnAttribute 1.3.0
• RouteOnContent 1.3.0
• RouteText 1.3.0
• ScanAttribute 1.3.0
• ScanContent 1.3.0
• ScrollElasticsearchHttp 1.3.0
• SegmentContent 1.3.0
• SelectHiveQL 1.3.0
• SetSNMP 1.3.0
• SplitAvro 1.3.0
• SplitContent 1.3.0
• SplitJson 1.3.0
• SplitRecord 1.3.0
• SplitText 1.3.0
• SplitXml 1.3.0
• SpringContextProcessor 1.3.0
• StoreInKiteDataset 1.3.0
• TailFile 1.3.0
• TransformXml 1.3.0
• UnpackContent 1.3.0
• UpdateAttribute 1.3.0
• UpdateCounter 1.3.0
• UpdateRecord 1.3.0
• ValidateCsv 1.3.0
• ValidateXml 1.3.0
• Wait 1.3.0
• YandexTranslate 1.3.0

Controller Services

• AvroReader 1.3.0
• AvroRecordSetWriter 1.3.0
• AvroSchemaRegistry 1.3.0
• AWSCredentialsProviderControllerService 1.3.0
• CouchbaseClusterService 1.3.0
• CSVReader 1.3.0
• CSVRecordSetWriter 1.3.0
• DBCPConnectionPool 1.3.0
• DistributedMapCacheClientService 1.3.0
• DistributedMapCacheServer 1.3.0
• DistributedSetCacheClientService 1.3.0
• DistributedSetCacheServer 1.3.0
• FreeFormTextRecordSetWriter 1.3.0
• GCPCredentialsControllerService 1.3.0
• GrokReader 1.3.0
• HBase_1_1_2_ClientMapCacheService 1.3.0
• HBase_1_1_2_ClientService 1.3.0
• HiveConnectionPool 1.3.0
• HortonworksSchemaRegistry 1.3.0
• IPLookupService 1.3.0
• JettyWebSocketClient 1.3.0
• JettyWebSocketServer 1.3.0
• JMSConnectionFactoryProvider 1.3.0
• JsonPathReader 1.3.0
• JsonRecordSetWriter 1.3.0
• JsonTreeReader 1.3.0
• PropertiesFileLookupService 1.3.0
• ScriptedLookupService 1.3.0
• ScriptedReader 1.3.0
• ScriptedRecordSetWriter 1.3.0
• SimpleCsvFileLookupService 1.3.0
• SimpleKeyValueLookupService 1.3.0
• StandardHttpContextMap 1.3.0
• StandardSSLContextService 1.3.0
• XMLFileLookupService 1.3.0

Reporting Tasks

• AmbariReportingTask 1.3.0
• ControllerStatusReportingTask 1.3.0
• DataDogReportingTask 1.3.0
• MonitorDiskUsage 1.3.0
• MonitorMemory 1.3.0
• ScriptedReportingTask 1.3.0
• SiteToSiteBulletinReportingTask 1.3.0
• SiteToSiteProvenanceReportingTask 1.3.0
• SiteToSiteStatusReportingTask 1.3.0
• StandardGangliaReporter 1.3.0

优雅的界面

就是我个人认为非常不错的界面，可以看到非常详细的数据流向。

NiFi在Hortonworks的定位

因为NiFi可以对来自多种数据源的流数据进行处理，Hortonworks认为HDF平台非常适合用于物联网 (IoAT)的数据处理。HDF中的数据流动可以是多个方向，甚至是点对点的，用户可以同收集到的数据流进行交互，这种交互甚至可以延伸到数据源，比如一些传感器或是设备。按照Hortonworks公司的说法，HDF产品是对HDP产品的补充，前者主要处理移动中的数据，而后者基于Hadoop技术，主要负责从静止的数据中获取洞察。可以看一看Hortonworks官方宣传对HDF的定位，已经号称是端到端流数据处理分析。

Hortonworks DataFlow (HDF) provides the only end-to-end platform that collects, curates, analyzes and acts on data in real-time, on-premises or in the cloud, with a drag-and-drop visual interface. HDF is an integrated solution with Apache Nifi/MiNifi, Apache Kafka, Apache Storm and Druid.

上图是概要介绍HDF三大部分，The HDF streaming data analytics platform includes data Flow Management, Stream Processing, and Enterprise Services.Nifi是作为数据管理和接入，可以延伸部署到边缘网关的重要能力。

结语

如果你的项目中也有同样的对多数据源的处理诉求，NiFi是个不错的选择。