《Boomi：2023年探索数据集成模式白皮书（英文版）（17页）.pdf》由会员分享，可在线阅读，更多相关《Boomi：2023年探索数据集成模式白皮书（英文版）（17页）.pdf（17页珍藏版）》请在三个皮匠报告上搜索。

1、Exploring Data Integration PatternsLearn about the four key data integration patterns and the methods for putting them into action.With the Boomi AtomSphere Platform,you can quickly build and easily manage all your integrations.2eBook|Integration Patterns3 Introduction4 Integration Patterns8 Using P

2、atterns to Solve Integration Challenges13 Accelerate Your Business With a Unified Integration PlatformTable of Contents3eBook|Integration PatternsTheres no question that in todays business world,you need a solid technology foundation to succeed.New SaaS applications hit the market every day.And esta

3、blished enterprises are migrating to a hybrid IT environment to leverage these best-of-breed cloud applications.Even for businesses taking a cloud-first approach,IT complexity can be baffling.And almost inevitably,this results in data silos that block customers,employees and partners from getting th

4、e data they need when they need it.To make information,interactions and innovation flow faster inside and outside your organization,you need integration.But integration is full of challenges.For anyone in technology,the phrase“disparate systems”will ring a few bells.Different data formats.Different

5、languages.Multiple operating platforms.Decentralized systems or centralized systems that arent flexible enough to support a decentralized working model.To this,add an ever-increasing volume of data and rapid technology innovations.When dealing with these challenges,enterprise architects frequently c

6、ome across the same problems,which fall into a handful of common integration patterns.In almost any integration project,one or more of these patterns will be the way you want to build your application and data integrations.In this ebook“Exploring Data Integration Patterns,”well cover the top four in

7、tegration patterns and the methods for putting them into action.And well explain how you can quickly build and easily manage any of those integrations with the Boomi AtomSphere Platform.Introduction4eBook|Integration PatternsSection 1:Integration Patterns5eBook|Integration PatternsThere are four com

8、mon data management techniques that enterprise architects use to address integration challenges.Each has advantages and drawbacks.Its important to understand all of them to know the best one or combination that will provide the most effective support for a given data integration project.1.Transforma

9、tion2.Aggregation3.Routing/Mediation4.OrchestrationA handy mnemonic for this is“taro,”like the starchy root vegetable.1.TransformationDifferent systems and applications have different data types and different data formats.For one system to accept data from another,changes need to be made.Data transf

10、ormation is the process of converting data or information from one format to another.For example,if you want to run analysis of structured and unstructured data,you need to join these as a unified dataset.Another kind of transformation might be to convert data from a CSV file to XML,so it can be ope

11、ned with the right tools.But data transformation goes beyond simple compatibility conversions.The transformation process can also include data cleansing,such as deduplication of records,and data enrichment,like appending data through aggregation.Data Integration ProcessesXMLSource/TargetSource/Targe

12、tTranslatorJSON6eBook|Integration Patterns2.AggregationAggregation is important for eliminating data and business silos.It brings together the data you need to create a comprehensive view.Say you have five,six,ten or more applications that need to talk to one another.With aggregation,you can extract

13、 and process data from multiple systems,such as compiling customer data from Salesforce,Workday and SAP,and presenting a single view to your customer service reps.Then you can deposit it in a single data repository like a data warehouse or data lake.But since each aggregation is specific to each req

14、uest,additional lines of code are required.Building the back-end logic for each new request quickly becomes unwieldy.3.Routing/MediationData routing is a critical part of building an integration process.When a request arrives,you need to understand where it should go where to route it.You also need

15、to know what data is needed.For example,you may have a support ticketing system where you want to trigger certain behaviors for certain customers.If a customer has paid extra for platinum-level support,you want to ensure its service calls are given priority and routed to senior support staff.To ensu

16、re the priority customers support ticket gets to the right people in a timely manner,you need rules or logic(mediation)that use metadata to appropriately route that request.DATA INTEGRATION PROCESSESSourceSourceTargetProcessingLogicSourceProcessingLogicTargetTargetTarget7eBook|Integration Patterns4.

17、OrchestrationBuilding a complex response composed of data from multiple sources and systems requires data orchestration.Data orchestration facilitates and tracks data flows and data consumption from disparate sources,within a strong data governance framework.This goes beyond simple aggregation.It in

18、volves creating an automated process or workflow to fulfill a specific requirement.Data orchestration works a bit like a conductor but for automating and streamlining processes involving multiple systems.Lets take invoice creation as an example.When a sales order comes through,there are a few proces

19、ses,all of which probably happen within separate systems:The invoice needs to be generated.The details of the order need to be filled in.Warehouse systems need to be updated.The invoice needs to be sent to the purchaser.And later,the order needs to be fulfilled,payment applied to the invoice,a recei

20、pt sent,etc.Orchestration makes sure all these processes happen in the right order,at the right time.DATA INTEGRATION PROCESSESTargetTargetNotificationSourceTargetProcessingLogic8eBook|Integration PatternsSection 2:Using Patterns to Solve Integration Challenges9eBook|Integration PatternsIts a given

21、that you need to integrate your data.But you still have choices about how to make that integration happen.You must decide on your approach to data mapping.Next,you need to aggregate the data.And then transform it.With a clear understanding of the four types of integration patterns,lets look at the b

22、est ways to put them to use.1.Point-to-PointThe most common integration process is point-to-point simply connecting two applications as quickly as possible.Lets say you need to send an update from Salesforce to NetSuite.With a point-to-point connection,you take the data from one place,in this case S

23、alesforce Sales Cloud,and put it into another a NetSuite ERP.Its that straightforward.You can use point-to-point connections for either real-time or batch integration,and they can be bidirectional.These integrations provide a quick and easy way to replace manual data entry processes and speed workfl

24、ows.But its a tactical,rather than a strategic,approach.Very few(if any)companies have only two applications to connect.With the volume and complexity of data,even small businesses have to be good at managing their data and creating unified views of their business.Because point-to-point builds each

25、connection separately,its not scalable.When you have more than two endpoints,point-to-point fails.If you try to connect multiple point-to-point integrations,you wind up with“spaghetti code.”Every time something changes in one application,you have to adjust the code in every integration individually.

26、This complex,tangled pile of code is difficult and time-consuming to maintain.And it slows down your business.Data Integration Patterns in ActionTargetSourceProcessingLogicTranslator10eBook|Integration Patterns2.Publish/SubscribePublish/subscribe messaging,commonly referred to as“pub/sub,”is perhaps

27、 the most well-known integration model.It uses asynchronous service-to-service communication,where senders publish messages categorized by specific topics.Data is stored in a queue,and each receiving application consumes that data at its own pace.Queuing mechanisms can help govern subscriptions if n

28、eeded.A simple example of this is an RSS feed(“rich site summary”or“really simple syndication”).Say youre interested in getting any news related to Game of Thrones.You can subscribe to an RSS feed that regularly checks with the servers in its list,pulls down any new information it finds on the show,

29、and presents it to you.Of course,you can also use pub/sub in a business setting.Take the customer support ticketing example we used earlier to describe the routing/mediation pattern.The account manager can subscribe to any tickets involving a premier customer.Or a product manager may subscribe to to

30、pics related to his product roadmap.Pub/sub is easy to scale,secure and reliable.The architecture can support multiple back-ends that need access to the same data.Instead of 10 different integrations,you need only one.As opposed to a point-to-point integration,where you have two applications communi

31、cating directly with each other,pub/sub has a message bus in the middle.A message bus combines a common data model and command set with a messaging infrastructure to allow different systems to communicate through a shared set of interfaces.Subscribers can choose the application they wish to use to r

32、eceive the data,and they can choose the frequency with which the application checks the servers for new information.Many of the messaging buses in use also provide guaranteed delivery,so a subscriber doesnt have to worry about missing an update.But there are drawbacks.Pub/sub is coupled to a protoco

33、l,meaning that subscribers must use the same protocol as the sender.And if the architecture wasnt set up properly,management becomes difficult.For example,you would need to reconcile different queues with different permission settings.Furthermore,while pub/sub is much more scalable than point-to-poi

34、nt,the higher the volume of subscriptions,the more cumbersome it becomes to manage.And significant increases in message size can lead to performance degradation as those messages take up greater and greater storage space.Thats why we believe pub/sub is best for connecting applications within an orga

35、nizations or with other applications where some performance degradation is acceptable.DATA INTEGRATION PATTERNS IN ACTIONSourceTargetsMessage Broker/Queue11eBook|Integration Patterns3.APIs/Web ServicesThe rising popularity of cloud-based applications has spurred interest in application programming i

36、nterfaces(APIs)to expose web services.An API is code that allows two software programs to communicate with each another.In the case of a web service,the API spells out the proper way for a developer to write a program requesting services from an operating system or other application via the web.The

37、majority of APIs today use the Representational State Transfer(REST)protocol,an architectural style for distributed hypermedia systems the web as we know it today.RESTful APIs allow users to connect and interact with cloud services,using less bandwidth than Simple Object Access Protocol(SOAP)technol

38、ogy.Theres no standard format in which the data must be processed,and once the API is created,everyone using it can get access to that data.This modern,lightweight way to unlock data abstracts the technical details,making it easy for the end user to consume digital services without worrying about th

39、e underlying technology.For example,without the use of a RESTful API,connecting data between two cloud applications,such as PeopleSoft and SAP CRM,would require you to know the format of the fields,authentication,etc.In a RESTful API,that information is part of the metadata,making it simpler and eas

40、ier to exchange information among cloud applications.The data can be exchanged without any application-level knowledge.Plus,authentication is very strong,providing more secure access than some other methods of communication.As the majority of web services are synchronous,RESTful APIs are best used f

41、or real-time,bidirectional data transfer.They are great for extending connectivity to other applications and third-party providers outside your organization.They are the preferred method of integration for many large SaaS providers.There arent many disadvantages to using RESTful APIs.Nevertheless,li

42、ke any other connector,the higher the volume of data and message size,the more likely you are to confront performance issues.DATA INTEGRATION PATTERNS IN ACTIONClient/UserSource/TargetAPI GatewayProcessingLogic12eBook|Integration Patterns4.Extract,Transform,LoadMost organizations are familiar with e

43、xtract,transform,load(ETL)a proven method for transferring large volumes of data.Its frequently used in data migration projects and for building data warehouses.ETL is really the only way to get large amounts of data in and out of legacy applications.Its also a top option for data transfers with new

44、er applications.ETL can handle vast quantities of structured or unstructured data.ETL extracts data from various sources,which is then cleansed,enriched,transformed and stored(in a data warehouse or data lake).Growing in popularity is a closely related alternative to ETL:extract,load and transform(E

45、LT).In ELT,the extracted data is loaded into the data warehouse and then transformed.ELT pipelines often use cloud-based data warehouses such as Amazon Redshift and Google BigQuery because theyre highly efficient in performing transformations.These methods work nicely for moving large amounts of dat

46、a that require complex rules and data transformations.The automation of processes makes maintenance and traceability much easier versus hand-coding.But because ETL and ELT are batch-driven,they arent a good fit for event-based or real-time integration.Plus,it can easily take months and a squad of da

47、tabase experts to build it and manage changing requirements.DATA INTEGRATION PATTERNS IN ACTIONSourceTargetTranslator13eBook|Integration PatternsSection 3:Accelerate Your Business with a Modern Integration Platform14eBook|Integration PatternsBoomi accelerates your digital enterprise by making it fas

48、ter and easier to unify data,systems,applications,processes and people.With proven connectivity to 1,500+endpoints,Boomis integration cloud lets you quickly and easily connect any combination of applications public or proprietary,on-premise or in the cloud regardless of data format.Boomi can support

49、 any integration pattern or method real-time,event-based,batch,ELT,asynchronous processes and more.The Boomi AdvantageSource/TargetSource/TargetClient/UserSource/TargetProcessingLogicAPIGatewayMessageBroker/QueueTranslator15eBook|Integration PatternsHow Boomi Does It With the Boomi AtomSphere Platfo

50、rm,you can easily address any integration scenario.Along with access to pre-built integration processes and reusable components,Boomi offers a vast library of technology and application connectors.These make it quick and easy for you to build simple to sophisticated integrations in our intuitive dev

51、elopment environment.Heres how Boomi addresses the common integration methods:Point-to-PointBoomi provides technology and application connectors to integrate and decouple applications,regardless of where they reside.Using Boomi also facilitates data mapping with Boomi Suggest.Boomi Suggest offers mi

52、llions of indexed,successful mappings built by other users that automatically recommend data flows for new integrations.Publish/SubscribeWith Boomi,you gain the benefit of format conversion,which a queuing mechanism cannot do on its own.Boomi can transform and send data to a publisher as well as con

53、vert data to read from an internal subscriber.Boomi provides an internal add-on queue for simple use cases.For more complex integrations,Boomi uses a JMS connector to tie to an enterprise queuing mechanism or other queuing technology.APIs/Web ServicesIn an API-driven integration approach,Boomi acts

54、as an enterprise service bus(ESB).We expose data via our API management tool and help applications unlock the target data.With Boomi,you can build modern applications that allow any kind of transformation.Boomi also supports all the security mechanisms for APIs.ETL/ELTBoomi serves as an ETL process.

55、After a certain data volume threshold,Boomi uses the ELT method,which has proven to be the most efficient way to manipulate huge amounts of data.For example,you can use our BigQuery connector(or an HTTP client or app connector)to extract and load the data,but youll need to transform it in BigQuery o

56、r with SQL once its in the destination system.16eBook|Integration PatternsBut integration is only the beginning.The Unified Boomi Platform offers organizations“economies of skill”development and management efficiencies that help companies scale and grow faster.Boomi Master Data Hub unifies data acro

57、ss systems,making it possible for analysts to model,match,synchronize,cleanse and enrich data spanning multiple domains.These comprehensive data management and governance capabilities ensure the highest quality of data while integrating with any combination of SaaS and on-premise applications.And Bo

58、omi Flow provides a low-code development platform to create customer journeys and automate workflows from simple to sophisticated.By abstracting code into visual models,Flow speeds development,helping roll out new applications and digital services faster.And it allows business stakeholders to offer

59、input more effectively while IT maintains control over development and deployment.Boomi API Management provides a unified and scalable cloud-based platform to centrally manage and enrich API interactions through their entire lifecycle.With Boomi,you can manage and control APIs,while rapidly creating

60、 and publishing any endpoint as an API on-premise or in the cloud.If you need to manage a trading partner network,look no further than Boomi B2B/EDI Management.Rapidly onboard trading partners,support traditional EDI standards(as well as newer web services),and monitor all your partner interactions

61、in real time.Harness Economies of Skill to Unify Your Digital Ecosystem17To learn more about how the Boomi AtomSphere Platform can help your digital enterprise move faster and with greater agility,please contact a Boomi integration expert today.Copyright 2021 Boomi,LP.Boomi,the B logo,Atom,AtomSpher

62、e,and Molecule,are trademarks of Boomi,LP or its subsidiaries.Other names or marks may be the trademarks of their respective owners.Learn more about the Boomi integration platform B all customer stories at B on socialModernize Integration Processes to Drive Business TransformationYour business needs

63、 a technology foundation that makes information,interactions and innovations flow faster inside and outside the organization.By modernizing integration processes,you can dramatically reduce the time and complexity required to connect systems and break down data and business silos.There are several o

64、ptions to how you approach data and application integration.Each has pros and cons.But a modern,low-code integration platform makes it easy to adopt any of them.Boomis cloud-native,low-code platform allows you to start quickly,build efficiently and grow confidently.ABOUT BOOMI Boomi instantly connec

65、ts everyone to everything with our cloud-native,unified,open,and intelligent platform.Boomis integration platform as a service(iPaaS)is trusted by customers globally for its speed,ease-of-use,and lower total cost of ownership.As the pioneer in fueling intelligent use of data,Boomis vision is to make it quick and easy for customers and partners to discover,manage,and orchestrate data,while connecting applications,processes,and people for better,faster outcomes.Learn more at .