Welcome!

Microservices Expo Authors: Pat Romanski, Elizabeth White, Liz McMillan, Karthick Viswanathan, Andy Thurai

Related Topics: Microservices Expo

Microservices Expo: Article

Horses for Courses: Services, Objects, and Loose Coupling - Integration without compromise

Horses for Courses: Services, Objects, and Loose Coupling - Integration without compromise

Object-oriented technologies are used today in the design and development processes for many computer systems; it is a proven paradigm and has made possible the development of large and complex software systems. Enabling platforms and tools for building and consuming Web services will not be an exception. However, how a service is implemented using objects and the way in which it interacts with other services via message exchanges require very different approaches.

Today, most tools represent Web services to application developers as objects. Such an approach carries the danger of underperformance and fragile applications with too tight a level of coupling, which loses the benefits of service orientation. This article shows you how objects and services should be integrated to build loosely coupled, performant, and reusable services without compromising either paradigm.

Anatomy of a Web Service
To frame this discussion, we must first reach a common understanding of what exactly a Web service is. Our deliberately simple view of a service is as an entity that exchanges messages (usually with other services), with well-defined boundaries. A Web service shares those characteristics, but adds the constraint that the messages exchanged between Web services are in fact SOAP messages carried over some suitable transport protocol (which may include application protocols like HTTP).

Those messages may contain out-of-band data for quality-of-service purposes, and are produced and consumed by a piece of middleware, usually called a SOAP server, SOAP stack, or SOAP message processor. It is this middleware that performs the translation of SOAP messages into a form more suitable for processing at the application level, which it typically does by converting the SOAP XML message into application objects. This canonical Web service is shown in Figure 1, and will form the basis for later discussion on how objects and services interact. Please note that Figure 1 is only a conceptual view of a canonical Web service and there is no suggestion of a one-to-one association between a Web service and a particular host. Indeed, an entire infrastructure with computational, data, human, etc., resources could make a single Web service.

An Object-Oriented System
Before we explore the interaction between objects and services, it is useful to recall how we would implement a purely object-based system. Consider a simple purchase-order system that is used by a (potentially remote) purchase-order application, composed from a number of purchase-order objects.

In Figure 2, we see that the purchasing application holds a reference to (potentially remote) purchase order objects. Through those references, the application can obtain information about the purchase order, such as the order number, the date, the cost of the order, and the items requested.

While this approach is appealing because of its simplicity, it suffers from a number of drawbacks if we try to scale it to Internet-level computing. First, while the interface of the purchase order object is defined at the right level for a closed system (either a stand-alone application or an intranet-based distributed object environment), it is far too fine grained to use efficiently over a wide area network. The network overhead of calling a method like getPurchase OrderNo() outside of a single domain is prohibitively high.

Second, we see that the level of coupling with this approach is very high. If, for example, the purchasing application and the purchase order object physically belong in separate organizations (which is the normal case in CORBA, where object references are propagated around an application and the underlying infrastructure handles location transparency), then the organization hosting the purchase order object can legitimately delete that object but, as a consequence, breaks the other organization's purchasing application, which relied on a reference to that object. Of course, technologies for maintaining referential integrity have been devised (e.g., distributed reference counting and garbage collection, leasing, etc.), but in most cases they don't scale very well either.

Finally, and most crucially, this approach assumes that there is a type system shared by the purchasing application and the purchase order object. If two application domains are unable to share a type system, for instance if one is built on .NET and the other on J2EE, then direct invocation of objects from one domain into the other is all but precluded, which means that specialist adapters then have to be deployed to solve the problem piecemeal.

While middleware like CORBA or DCOM can hide the differences between individual language and platform technologies and allow, for example, a C++ purchasing application to invoke a Java purchase order object, it does so because it uses its own metalevel type system, and does not solve the problem of sharing types but moves it to the metalevel. It also throws up the barrier that if there is no IDL mapping to the language that a specific component is written in, then there is no way that component can be used in the application.

Don Box of Microsoft sums this up very neatly when he notes that "shared abstractions are expensive." Box convincingly argues that for a distributed object approach to have a hope of working across enterprise boundaries, all parties involved in the construction of the application have to agree on the set of shared abstractions. As the number of parties involved increases, there is a combined explosion of different systems, different people, and different corporate politics to navigate, and eventually the rewards for deploying such a system are outweighed simply by the effort required to deploy it.

An Object-like System Using SOAP and WSDL
If there are problems in scaling object-based systems, it seems an attractive idea to use technologies like SOAP and WSDL to alleviate that problem. One such common approach is shown in Figure 3.

The example shown in Figure 3 at first appears to solve the issues that we previously identified with integrating multiple enterprises' object-based systems. Certainly, it seems to solve the problem about having shared types, since now all type information is exposed via a WSDL interface that allows any software agent that understands the WSDL to invoke the object behind the service. It also means that specialist adapters can be replaced with commodity SOAP stacks in order to plumb various application components together. The ability to understand XML, XML Schema, and SOAP are the only assumptions being shared.

However, there are significant flaws in this approach. While this approach may be platform neutral because of the XML payloads that we move across the network, applications built this way may not be performant because we have exposed the same fine-grained object interface for consumption, and we still have to bear the cost of translating to and from XML whenever we want to invoke such a method.

Worse, applications built this way are brittle. For example, imagine that the purchase order object is made available at the URL http://example.com/wsj/purchaseOrder/14529 and that the client binds to that URL and begins to invoke methods (via SOAP) on the purchase order object. Given that the purchase order object is a resource that belongs to an enterprise, that enterprise is perfectly within its rights to manage that resource as it sees fit, including to move or delete it. If the URL of the purchase order object changes or disappears, this has the knock-on effect of breaking the purchasing application.

While we have schemes such as that used in OGSI, which address this brittleness by providing a mechanism to re-resolve an endpoint in case a resource is moved, this patch fails to address the fundamental problem with this approach - that an enterprise's private resources (objects) are being exposed to the network. In short, this approach is nothing more than CORBA or DCOM with angle brackets, and while CORBA and DCOM were extremely useful technologies in their specific domain, in this context such an architecture is, at best, ungainly.

Given these drawbacks, we must ask what would drive developers and architects to propose such solutions. While one motivation is the (completely inaccurate) view that Web services are a yet another distributed object system, we feel the primary reason for this model having permeated so far into the developer mindset is tool support.

For instance, the WebMethod attribute in .NET makes it easy to expose an object as a Web service, even though such objects may well have fine-grained interfaces that are wholly unsuitable for exposure as a Web service. On the client side, tools like Axis WSDL2Java present the abstraction of a Web service as an object with methods (which is ironic given that at least some of the time it, lamentably, is).

A Service-Oriented View
So far we have seen the pitfalls of mixing service and object orientation. However, both are extremely valuable techniques and in order to derive maximum benefit from them we simply need to understand where it is best to utilize each, and how the object-service interface should look. To put things into perspective, let's revisit the purchase order system from a service-oriented architecture point of view.

The diagram shown in Figure 4 fixes all of the problems we encountered in examining previous approaches. There is much looser coupling in this system since no back-end resources are exposed across administrative boundaries. Instead, a (logically, though not necessarily physically) single, fixed endpoint is used as the entry point to a process that deals with the creation of a purchase order. When that process is invoked, some business-specific activity is executed and a purchase order document is passed back to the purchasing application. The purchase order service interface is now much more coarse grained as it captures an entire business process rather than individual method calls, and is likely to be performant and scalable.

However, this is only half the story as far as we are concerned. While this architecture is the correct choice for this application, we still need to see how it can be mapped into code in the service implementation. Figure 5 presents the details of the implementation of the purchasing order service and shows in greater detail how a service could be implemented using objects without actually exposing them.

The fundamental issue shown in Figure 5 is that it is the messages exchanged with Web services that should be consumed by applications, and not the Web services themselves. Where Web services are seen as application-level objects, there is the constant danger we'll fall into the trap of treating them as objects.

However, with this approach we use fewer but richer message exchanges, we increase the granularity of a Web service, and reduce the performance penalty. This strategy also ensures that loose coupling is the norm since applications never bind directly to services, only to messages. Since those messages can come from any source as far as the application is concerned, migrating the service provider to another service is not invasive to the application. Finally, since the architecture is concerned only with messages and message exchanges, the barrier to entry for any enterprise is low. There is no need to argue about whether a system should be based on J2EE, .NET, or some other technology since most platforms have Web services support nowadays.

Conclusion
Building object-like systems on top of Web services specifications is a strategy that leads to tightly coupled, brittle, and underperforming systems. While the current crop of tool support implicitly advocates this approach to the unwary, a far better solution is to think of an application in terms of services and the messages that those services exchange. When building code for those service implementations don't be tempted to bind directly to a Web service, but instead bind to the messages that the service produces and consumes. Such an approach promotes loose coupling; your applications will be able to transcend changes in the service they consume and will be much easier to maintain.

References

  • Box, Don. (2003). "Service-Oriented Architecture and Programming (SOAP), Parts 1 and 2." Talks from the MSDN TV archive (Part 1, Part 2).
  • Tuecke, Steve., et al. (2003) "Open Grid Services Infrastructure (OGSI) - Version 1.0". https://forge.gridforum.org/projects/ogsi-wg.
  • More Stories By Jim Webber

    Dr. Jim Webber is a senior researcher from the University of Newcastle
    upon Tyne, currently working in the convergence of Web Services and Grid
    technologies at the University of Sydney, Australia. Jim was previously
    Web Services architect with Arjuna Technologies where he worked on Web
    Services transactioning technology, including being one of the original
    authors of the WS-CAF specification. Prior to Arjuna, Jim was the lead
    developer with Hewlett-Packard on the industry's first Web Services
    Transaction solution. Co-author of "Developing Enterprise Web Services -
    An Architect's Guide," Jim is an active speaker and author in the Web
    Services space. Jim's home on the web is http://jim.webber.name

    Comments (0)

    Share your thoughts on this story.

    Add your comment
    You must be signed in to add a comment. Sign-in | Register

    In accordance with our Comment Policy, we encourage comments that are on topic, relevant and to-the-point. We will remove comments that include profanity, personal attacks, racial slurs, threats of violence, or other inappropriate material that violates our Terms and Conditions, and will block users who make repeated violations. We ask all readers to expect diversity of opinion and to treat one another with dignity and respect.


    Microservices Articles
    DevOps is often described as a combination of technology and culture. Without both, DevOps isn't complete. However, applying the culture to outdated technology is a recipe for disaster; as response times grow and connections between teams are delayed by technology, the culture will die. A Nutanix Enterprise Cloud has many benefits that provide the needed base for a true DevOps paradigm. In their Day 3 Keynote at 20th Cloud Expo, Chris Brown, a Solutions Marketing Manager at Nutanix, and Mark Lav...
    The Software Defined Data Center (SDDC), which enables organizations to seamlessly run in a hybrid cloud model (public + private cloud), is here to stay. IDC estimates that the software-defined networking market will be valued at $3.7 billion by 2016. Security is a key component and benefit of the SDDC, and offers an opportunity to build security 'from the ground up' and weave it into the environment from day one. In his session at 16th Cloud Expo, Reuven Harrison, CTO and Co-Founder of Tufin, ...
    "NetApp's vision is how we help organizations manage data - delivering the right data in the right place, in the right time, to the people who need it, and doing it agnostic to what the platform is," explained Josh Atwell, Developer Advocate for NetApp, in this SYS-CON.tv interview at 20th Cloud Expo, held June 6-8, 2017, at the Javits Center in New York City, NY.
    Gone are the days when application development was the daunting task of the highly skilled developers backed with strong IT skills, low code application development has democratized app development and empowered a new generation of citizen developers. There was a time when app development was in the domain of people with complex coding and technical skills. We called these people by various names like programmers, coders, techies, and they usually worked in a world oblivious of the everyday pri...
    Your homes and cars can be automated and self-serviced. Why can't your storage? From simply asking questions to analyze and troubleshoot your infrastructure, to provisioning storage with snapshots, recovery and replication, your wildest sci-fi dream has come true. In his session at @DevOpsSummit at 20th Cloud Expo, Dan Florea, Director of Product Management at Tintri, provided a ChatOps demo where you can talk to your storage and manage it from anywhere, through Slack and similar services with...
    Kin Lane recently wrote a couple of blogs about why copyrighting an API is not common. I couldn’t agree more that copyrighting APIs is uncommon. First of all, the API definition is just an interface (It is the implementation detail … Continue reading →
    The Jevons Paradox suggests that when technological advances increase efficiency of a resource, it results in an overall increase in consumption. Writing on the increased use of coal as a result of technological improvements, 19th-century economist William Stanley Jevons found that these improvements led to the development of new ways to utilize coal. In his session at 19th Cloud Expo, Mark Thiele, Chief Strategy Officer for Apcera, compared the Jevons Paradox to modern-day enterprise IT, examin...
    DevOps tends to focus on the relationship between Dev and Ops, putting an emphasis on the ops and application infrastructure. But that’s changing with microservices architectures. In her session at DevOps Summit, Lori MacVittie, Evangelist for F5 Networks, will focus on how microservices are changing the underlying architectures needed to scale, secure and deliver applications based on highly distributed (micro) services and why that means an expansion into “the network” for DevOps.
    Many organizations are now looking to DevOps maturity models to gauge their DevOps adoption and compare their maturity to their peers. However, as enterprise organizations rush to adopt DevOps, moving past experimentation to embrace it at scale, they are in danger of falling into the trap that they have fallen into time and time again. Unfortunately, we've seen this movie before, and we know how it ends: badly.
    Kubernetes is a new and revolutionary open-sourced system for managing containers across multiple hosts in a cluster. Ansible is a simple IT automation tool for just about any requirement for reproducible environments. In his session at @DevOpsSummit at 18th Cloud Expo, Patrick Galbraith, a principal engineer at HPE, discussed how to build a fully functional Kubernetes cluster on a number of virtual machines or bare-metal hosts. Also included will be a brief demonstration of running a Galera MyS...