According to his post, surveys of CIOs reveal that integrating on-premise and cloud-based applications, as well as cloud-to-cloud integration, are big concerns. This post pays attention to application integration and data integration, the first two of the four cloud integration dimensions Seroter focuses on.
Often, one application depends on information provided by another. And as Mr. Seroter points out, application integration “is about connecting business applications at a functional level.” Data is shared, of course, but one application triggers activity in another by making a request or sending a real-time business event to it. If the application on the receiving end doesn’t respond, action grinds to a halt.
With cloud-based applications, such as Salesforce integration application, architects probably prefer to avoid synchronous remote procedure calls because of latency issues. Scalability also comes into play as more and more users access an application. “Callbacks,” as Mr. Seroter notes, are a possible solution, but asynchronous messaging, where data is sent between systems, is more scalable.
You’ll want to choose application integration products with REST, rather than SOAP web service endpoints in mind. You can consider either cloud-friendly traditional platforms or those designed specifically for the cloud. If you don’t require an integration server, you may want to investigate cloud-based queues like Amazon SQS. If you need more powerful functionality, then you’ll need to look at a solution with a messaging engine and cloud endpoint adapters, such as Salesforce integration or another CRM integration connector.
If you’re synchronizing, transforming, and transporting large batches of data from one repository to another, then you’re involved in data integration. During these ETL processes, latency is not as important as having sufficient bandwidth. You’ll want to consider cross connect architecture to maximize bandwidth available.
Security is important for all cloud integration tools. You may want to store data within your firewall, rather than on a public, cloud-based server. And you’ll need to consider the nature of your database. Many data integration tools are relational database oriented, but many cloud databases are not, and their structure won’t be familiar to your staff.
Just as with application integration, cloud data integration tools providers either offer adapters to cloud endpoints or solutions specifically for the cloud. The best of the former category will continuously expand their roster of adapter offerings. What is most important in either situation is that local server agents connect data repositories to your network so that you do not need to expose internal data bases to a public cloud solution.
Paying attention to these integration dimensions will go a long way toward ensuring that your organization will enjoy much-touted benefits of cloud computing and make the best use of your current resources.