Common Object Request Broker Architecture

From Wikipedia, the free encyclopedia

The Common Object Request Broker Architecture (CORBA) is a standard defined by the Object Management Group (OMG) that enables software components written in multiple computer languages and running on multiple computers to interoperate.

Contents

[edit] General overview

CORBA “wraps” program code into a bundle containing information about the capabilities of the code inside and how to call it. The resulting wrapped objects can then be called from other programs (or CORBA objects) across a network.

CORBA uses an interface definition language (IDL) to specify the interfaces that objects will present to the world. CORBA then specifies a “mapping” from IDL to a specific implementation language like C++ or Java. Standard mappings exist for Ada, C, C++, Lisp, Smalltalk, Java, COBOL, PL/I and Python. There are also non-standard mappings for Perl, Visual Basic, and Tcl implemented by ORBs written for those languages.

This diagram illustrates how the generated code is used within the CORBA infrastructure:

This picture does not reflect all typically used possibilities. Normally the server side has the Portable Object Adapter that redirects calls either to the local servants or (to balance the load) to the other servers. Also, both server and client parts frequently have interceptors that are described below.

In addition to providing users with a language and a platform-neutral remote procedure call specification, CORBA defines commonly needed services such as transactions and security.

[edit] Key Features

[edit] Objects by Value (OBV)

Apart from remote objects, the CORBA and RMI-IIOP define the concept of the OBV. The code inside the methods of these objects is executed locally by default. If the OBV has been received from the remote side, the needed code must be either a priori known for both sides or dynamically downloaded from the sender. To make this possible, the record, defining OBV, contains the Code Base that is a space separated list of URLs from where this code should be downloaded. The OBV can also have the remote methods.

The OBV's may have fields that are transferred when the OBV is transferred. These fields can be OBV's themselves, forming lists, trees or arbitrary graphs. The OBV's have a class hierarchy, including multiple inheritance and abstract classes.

[edit] CORBA Component Model (CCM)

CORBA Component Model (CCM) is an addition to the family of CORBA definitions. It was introduced with CORBA 3, and it describes standard application framework for CORBA components. It is an extension of "language independent Enterprise Java Beans (EJB)". It provides an abstraction of entities that can provide and accept services through well-defined named interfaces called ports.

The CCM has a component container, where software components can be deployed. The container offers a set of services that the components can use. These services include (but are not limited to) notification, authentication, persistence and transaction management. These are the most-used services any distributed system requires, and, by moving the implementation of these services from the software components to the component container, the complexity of the components is dramatically reduced.

[edit] External links

[edit] Portable interceptors

Portable interceptors are the "hooks", used by CORBA and RMI-IIOP to mediate the most important functions of the CORBA system. CORBA standard defines the following types of interceptors:

  1. IOR interceptors mediate the creation of the new references to the remote objects, presented by the current server.
  2. Client interceptors usually mediate the remote method calls on the client (caller) side. If the object Servant (CORBA) exists on the same server where the method is invoked, they also mediate the local calls.
  3. Server interceptors mediate the handling of the remote method calls on the server (handler) side.

The interceptors can attach the specific information to the messages being sent and IORs being created. This information can be later read by the corresponding interceptor on the remote side. Interceptors can also throw forwarding exceptions, redirecting request to another target.

[edit] General InterORB Protocol (GIOP)

The GIOP is an abstract protocol by which Object request brokers (ORBs) communicate. Standards associated with the protocol are maintained by the Object Management Group (OMG.). The GIOP architecture provides several concrete protocols:

  1. Internet InterORB Protocol (IIOP) - The Internet Inter-Orb Protocol, is a protocol for communication between CORBA ORBs that has been published by the Object Management Group. IIOP is an implementation of the GIOP for use over an internet, and provides a mapping between GIOP messages and the TCP/IP layer.
  2. SSL InterORB Protocol (SSLIOP) - SSLIOP is IIOP over SSL, providing encryption and authentication.
  3. HyperText InterORB Protocol (HTIOP) - HTIOP is IIOP over HTTP, providing transparent proxy bypassing.
  4. and many more...

[edit] Data Distribution Service (DDS)

The Object Management Group (OMG) has a related standard known as the Data Distribution Service (DDS) standard. DDS is a publish-subscribe data distribution model, in contrast to the CORBA remotely-invoked object model.

[edit] VMCID (Vendor Minor Codeset ID)

Each standard CORBA exception includes a minor code to designate the subcategory of the exception. Minor exception codes are of type unsigned long and consist of a 20-bit “Vendor Minor Codeset ID” (VMCID), which occupies the high order 20 bits, and the minor code which occupies the low order 12 bits.

Minor codes for the standard exceptions are prefaced by the VMCID assigned to OMG, defined as the unsigned long constant CORBA::OMGVMCID, which has the VMCID allocated to OMG occupying the high order 20 bits. The minor exception codes associated with the standard exceptions that are found in Table 3-13 on page 3-58 are or-ed with OMGVMCID to get the minor code value that is returned in the ex_body structure (see Section 3.17.1, “Standard Exception Definitions,” on page 3-52 and Section 3.17.2, “Standard Minor Exception Codes,” on page 3-58).

Within a vendor assigned space, the assignment of values to minor codes is left to the vendor. Vendors may request allocation of VMCIDs by sending email to tagrequest@omg.org.

The VMCID 0 and 0xfffff are reserved for experimental use. The VMCID OMGVMCID (Section 3.17.1, “Standard Exception Definitions,” on page 3-52) and 1 through 0xf are reserved for OMG use.

The Common Object Request Broker: Architecture and Specification (CORBA 2.3)

[edit] Corba Location

CorbaLoc means Corba Location and refers to a stringified object reference for a Corba object that looks similar to a URL.

All CORBA products must support two OMG-defined URLs: "corbaloc:" and "corbaname:". The purpose of these is to provide a human readable/editable way to specify a location where an IOR can be obtained.

An example of corbaloc is shown below:

corbaloc::160.45.110.41:38693/StandardNS/NameServer-POA/_root

A CORBA product may optionally support the "http:", "ftp:" and "file:" formats. The semantics of these is that they provide details of how to download a stringified IOR (or, recursively, download another URL that will eventually provide a stringified IOR).

[edit] External links

[edit] Problems and criticism

While CORBA promises to deliver much in the way code is written and software constructed, it has gathered a lot of criticism during its long history, and failed repeatedly to deliver much of its promised features, sometimes even the most fundamental ones.[1][citation needed]

Some of its failure can be tracked down to the implementations and the process in which CORBA was created as a standard, but some reflect more fundamental problems in the design, and others yet stem from the very principles on which CORBA was built.

These problems, combined on various levels, have led to a significant decline in CORBA use, and most importantly, adoption in new projects and areas, and caused the once "prodigial child" technology to degrade into an all but abandoned niche.[2]

It can be stated however, that many of the problems which CORBA has suffered from are now appearing in newer technologies where there is a community deciding the standards.[verification needed]

[3]

[edit] Fundamental flaws

One of the more touted flaws of CORBA has been its 'everything to everyone' nature. It is true that CORBA is a general purpose distributed object computing, as applicable to Telco as it is to Financials. CORBA detractors claim that it lacks a concrete goal and problem to solve however.[verification needed] Because of this, CORBA only has a very high-level goal of "making the world a better place", but no concrete direction to go in. This has caused the standard to include technologies as different as an IPC layer, an object-oriented type system, a programming languages interoperability layer, a component system, a distributed computing platform, and more.[verification needed]

Another disputed concept in CORBA is the notion of "location transparency", ie. pretending that local (that is, residing in the same address space and accessible with a simple function call) and remote (residing anywhere else, from different processes on the same machine, to different continents) objects are really the same. This notion is fundamentally false[4][citation needed], and can only be resolved in two ways: either by making all local accesses as complex as the most complex remote scenario possible, thus contradicting the very goal of simplification by unification, or by making remote objects unable to handle the failures that will occur in distributed environment, and making the resulting system probably non-robust.

[edit] Design and process deficiencies

In addition to the problems with the lack of well-defined goals, CORBA is often cited as a classic example of design by committee. Because there's no process in place to arbitrate between conflicting proposals, and no provisions for deciding on hierarchy of problems to attend, the standard was often created by taking a union of all features in all proposals, with no regard to coherency of the resulting creation.[2] The net result is a specification that's incredibly complex, prohibitively expensive (if at all possible) to implement in its entirety, and often with ambiguous definitions.

To further worsen the matters, the committee that standardises CORBA is composed largely of vendors who would later implement the proposals. Because standards and easy interoperability increase competition and ease customers' movement between alternative implementations, software vendors are fundamentally disinterested in having a comprehensive standard. This leads to much of political fighting within the committee, and frequently release of standard revisions that are impossible to use without proprietary extensions.[1]

[edit] Problems with implementations

Through its history, CORBA has been plagued by shortcomings of its implementations. Oftentimes, there were no implementations available to match the published specification,[2] and the existing ones were incomplete or inadequate. Because of the lack of requirement to provide a reference implementation, members are free to propose features which were never tested for usefulness or implementability. Implementations are further hindered by the general tendency of the standard to be verbose, and the common practice of setting compromise as a sum of all submitted proposals, which often creates APIs that are incoherent, quirky or even impossible to use, even if the individual proposals were perfectly reasonable.[citation needed]

An example of failure to specify useful interfaces could be the BOA (Basic Object Adapter). BOA is one class of object adapters, which allow to tie the calls made by clients to abstract CORBA interfaces with actual implementations of the objects. OAs are thus necessary to actually make any objects accessible via CORBA.

However, the first version (1.0) of the spec did not specify any mechanism for such implementation, effectively making it impossible to export objects for use by clients. Version 1.1, which was the first in wider circulation, attempted to fix that by introducing a Basic Object Adapter interface.[5] As the "basic" in the name suggests, this interface was only meant as a minimal interface for simple cases. However, even as such, it had a series of elementary shortcomings, which made it impossible to use in the specified form.[6] This meant each vendor had to extend BOA to make it useful at all, and in the result made it impossible to write server code which wasn't tied to given implementation from the very beginning. This flaw in the specification was fixed only by the addition of POA (Portable Object Adapter) in CORBA 2.2, which was released full six years later, and even then was still not matched by any working implementations of the feature. Working implementations of POA appeared in the year 2000 – two of which were Orbix 2000 and Orbacus 4. These are two products which are still being deployed in 2006.

CORBA was also traditionally very hard to get a working implementation of. Implementations were complex, slow, incompatible and incomplete, yet very expensive. This was changed somewhat by availability of free implementations later, but as of 2006, there's virtually no freely available implementation of the 3.0 version of the standard[7], despite its release back in July 2002.

[edit] CORBA successes

CORBA has achieved consensus and usage which is why there are so many battle hardened CORBA systems working today. Even though CORBA was designed by committee it has proven resilient to other technologies such as DCOM which are proprietary. Companies that can co-operate on a standard prove that they are mature enough to be able to co-operate for the benefit of customers.

Technically speaking, there is much benefit associated with a group-derived standard as well. A resounding example is the bundling of the static stub/skeleton approach with the dynamic DII/DSI approach. Originally submitted as two separate specifications for CORBA, there were arguments for and against each paradigm. In the end no clear-cut decision could be made in favor of one or the other and so both were included in the CORBA specification. It was only years later during the maturity of the various implementations that people realized this was a brilliant decision. Proponents of the stub/skeleton approach soon realized that some CORBA services were much easier implemented using DII/DSI and vice-versa.

[edit] Wide industry usage

CORBA is used in verticals such as telecommunication, aerospace, finance, military/defense, manufacturing, and automotive – all of which have the highest levels of security concerns, performance and scalability.

CORBA is today finding new markets in the embedded world particularly in software defined radio http://www.sdrforum.org/. where CORBA has been selected as a component of the overall SCA – software communications architecture.

[edit] Use in subsequent technologies

CORBA has made its way into subsequent technologies. For example, J2EE’s architecture relies on CORBA to work: Application servers rely on CORBA’s IIOP for RMI invocations, and require the Object Reference Template specification. The Java Transaction Service is intimately tied in with CORBA’s Object Transaction Service (OTS). In fact, a CORBA ORB ships as part of every copy of the JDK.

It has also been noted by many users of CORBA that they have had a SOA successfully running in their business for many years through the use of their CORBA implementations. CORBA has been called a first generation SOA.

[edit] Success of CORBA implementations

Much of the success of CORBA has been due to the implementations. There are very successful CORBA implementations in both the commercial and open-source markets. IONA for example has over 4500 CORBA deployments across various industry verticals. PrismTech is another successful CORBA vendor to sectors including: defense, aerospace, telecoms, finance, manufacturing and software defined radio with recent new announcements including supplying CORBA products to the U.S. Navy’s DDG 1000 Zumwalt Class Destroyer program. Borland is also a very successful CORBA vendor with its VisiBroker product powering many customers. In the open-source market, TAO (and to a lesser extent, MICO and JacORB) provide a successful and mature CORBA offering with thousands of users. Object Computing Inc. who provides commercial TAO support has posted a great presentation on how CORBA is still delivering. All of the above vendors have many examples of successful CORBA implementations on their websites. The Object Management Group also maintains a site with successful CORBA implementations.

[edit] Notes

  1. ^ a b Trouble with CORBA An historical essay, outlining the same problems that eventually led to CORBA's demise
  2. ^ a b c Rise and Fall of CORBA, a comprehensive analysis of CORBA failures, by one of its former architects, Michi Henning
  3. ^ Response to the Rise and Fall of CORBA, the other side of the story
  4. ^ A Note on Distributed Computing A classic paper, explaining in detail why local and remote context are fundamentally different, and the impossibility of unifying them into a single access model.
  5. ^ History of CORBA from the OMG site.
  6. ^ [1] A discussion of BOA interface and its problems.
  7. ^ [2] A listing of various, freely available CORBA implementations

[edit] CORBA implementations

[edit] OMG trademarks

CORBA, IIOP and OMG are the registered marks of the Object Management Group and should be used with care. Differently, the GIOP is not a registered OMG trademark. Hence in some cases may be more appropriate just to say that the application uses or implements the GIOP-based architecture. The text inside the CORBA specification itself ensures that no person shall be deemed just for using or implementing this specification.

[edit] See also

Wikibooks
Wikibooks Programming has more about this subject:

[edit] References

[edit] External links