infoTECH Feature

October 15, 2008

SOA and Web Services on the Move

Voltaire said, “Before we converse, we should define our terms” — actually, he was quoting John Locke. Be that as it may, the terms SOA (Service Oriented Architecture) and Web Services / WOA (Web-Oriented Architecture) represent different concepts, yet can complement each other. SOA is an architectural style that brings flexibility to application development. It’s the grandchild of distributed computing and modular programming — it breaks and “de-siloes” application functionality into individual units or services which can be reused and can pass data among each other over a network as they participate in business processes. WOA is probably easier and less expensive to implement than SOA but at the moment is less scalable and more focused on the user interface at the endpoints.
 
Cutting Out the Middleware
 
IONA Technologies (News - Alert) is an expert at delivering a distributed SOA infrastructure. By offering solutions that are fully distributed, organizations can service-enable systems without having to create yet another IT stack and without disrupting or having to expand existing middleware. IONA solutions allow businesses to transform IT operations so new services can be created and modified without affecting existing middleware or disrupting other systems. IONA solutions service-enable systems with lightweight plug-ins at the endpoint, rather than deploying integration technology in a centralized server.
 
Eliminating centralized servers drastically reduces hardware, software, development, testing, and maintenance costs, and increases overall corporate agility.
 
Eric Newcomer, CTO at IONA Technologies, said, “There’s a lot of hype and confusion, but there’s no doubt that the Web side of things is really gaining a lot of traction. However, there is a difficulty, especially for established companies, in adopting a ‘pure Web’ architecture. If you’re somebody like Google, eBay (News - Alert) or Amazon, which grew up in the Web era, then you’ve already designed your IT systems to handle the Web and you’re working fine with HTTP, XML and HTML. But if you’re an old-line telecom company with an established IT department that’s developed many systems on mainframes and UNIX, you have a very different orientation.”
 
“In particular, you’re thinking about how to control all of those resources inside the ‘glass house’ which was the main idea for a long time in the industry. Telcos thus end up with a very different approach to IT, because in the old days they didn’t know the Web was coming, and they didn’t have the idea that you could develop a worldwide presence of your IT department by sharing it on the Internet. Therefore there’s a real mismatch in architectures and IT environments between companies that grew up during the Web era and those whose IT departments predate the Web era,” explained Newcomer.
 
“I recently heard a keynote speaker talking about EasyJet in the U.K. and Europe, and how much business they do on the Web, and how good it is for them,” continued Newcomer. “He said that established telecom companies would love to do more business on the Web, but they really can’t because of what he called their ‘antediluvian IT systems’. So there is a real issue out there — how do you get your old-style IT departments to work better with the Web? There’s no doubt that the Web architecture is gaining traction, but for companies that didn’t grow up during the Web era, they have a big challenge.”
 
“Moreover, some people will try to set web-oriented architectures [WOA] in opposition to SOA, but that’s really an apples-to-oranges comparison,” said Newcomer. “An SOA is really all about a way of doing things. It’s a style of design, a blueprint, an approach to IT. It’s not a sole technology. As far as WOA, it pretty clearly encompasses everything from HTTP and how you do big scalable Web servers. People will take the view, ‘Okay, my SOA is implemented using old-style stuff, such as JavaEE, RPC or .NET, that predate the modern Web in terms of their design center, architecture and installed base.”
 
Newcomer added, “After all, the JavaEE and .NET environments were created as entire environments unto themselves, and people would develop all of their apps within the chosen environment. Now, people promoting WOA try to position their technologies in opposition to the older ones. But SOA is really not about technology, even though some people will argue that, particularly those who promote Web architectures by contrasting them with other things. In fact, we’ve had customers doing SOA for eight or nine years now — perhaps the most famous is Credit Suisse in Zurich. They have about 1,000 services they’ve built up over the past eight years. Zurich Insurance took their cue from them and have a large SOA, similarly built.”
 
“In 2007 we had put together a focus group of eight customers,” says Newcomer. “Half of them did SOA using our products, the other half didn’t do SOA at all. Those who did SOA said the main lesson they learned was that SOA is more about culture than it is about technology. That points back to SOA being about how you do something rather than what it is. The web architecture, however, is about what the Web is.”
 
Simple SOA Says…
 
Boomi provides integration solutions for Small and Medium-Sized Businesses (SMBs) that can’t afford the time and expense of implementing complex integration solutions often found in larger enterprises. Their Visual Integration Technology, the result of seven years of development, allows companies to integrate applications, trading partners and data in one solution that can be deployed on premise or on demand. Using a visual process designer, users drag and drop integration components from a palette onto a “canvas” to build anything from very simple to very sophisticated integration solutions. Integration configurations are built automatically and in many cases, no further coding is required.
 
Rick Nucci, CTO of Boomi, said, “We’re a SaaS (News - Alert) (Software-as-a-Service) application provider and we’re focused specifically in the area of integration. We live and breathe SOA and Web Services every day. Most SaaS applications are providing some Web Services layer to be able to integrate with their data and make it available to other applications. We provide an integrations service where users, from their web browser, build, deploy and manage integrations between SaaS applications as well as SaaS applications to on-premise applications residing behind the firewall. Essentially the way that we do it, in the context of SOA and Web Services, is to put what we call a ‘connector’ on top of a company’s API, be it SOA-based or relating to behind-the-firewall apps that use APIs predating Web Services technologies. Our connector abstracts away the technical details of the APIs so that users don’t have to contemplate disparate implementations of the various Web Services standards, disparate authentication models, or how to connect to a particular Web Service. Instead, we ‘bake’ that into the connector so the user can focus on using a wizard and the connector to browse the objects that the API supports and then pick what they want to integrate. It could be lead data in Salesforce.com or purchase order documents in Intact or NetSuite.”
 
“Our typical user thus not a developer,” said Nucci. “They don’t have a lot of technical resources and they don’t understand APIs from a technical standpoint, and they don’t have developers to write code. They’re generally ‘orchestrating’ consumable Web Services, and they’re visually discovering them using our product, and then they’re integrating them to automate exchanges of information between them. So we’re purpose-built for that kind of user.”
 
“The way we price our platform is by the number of applications being connected up,” added Nucci. “So when people are evaluating Boomi they count up the number of applications they want to integrate, and then buy those connections from us. These range in price from $65 to $495 a month. It’s very disruptive pricing. For example, if you were integrating something such as Salesforce.com to Great Plains software, that’s two standard connections at $135 a month each, or $270 a month total, with unlimited volume. We can do that because we’re SaaS-based. We have a very low cost infrastructure, thanks to our multi-tenancy capabilities, than our product-based competitors.”
 
Fast, Faster, Fastest
 
Businesses that increasingly create, sell and deliver large digital content files over the Internet are often hampered by slow download times, resulting in poor end-user experience. FastSoft’s centralized approach fixes the leading Internet transmission protocol, TCP, to speed downloads by 3 to 7 or even 30 times and to deliver content anywhere in the world with no additional technology at the endpoints. By using FastSoft’s E Series, companies can now deliver high-quality video, to enhance SaaS and cloud computing offerings, and to speed up other Web-based applications.
 
Dan Henderson, vice president of marketing at FastSoft, said, “We fix TCP, and enable Internet-based platforms and services either within a company or even a SaaS-type model, with a focus on video. We were launched 2006 out of CalTech’s Networking Lab, where they developed the next generation of TCP called FastTCP. The original $7 million in funding from 2000 to 2005 came from DARPA, Cisco (News - Alert) and some other big entities. The technology was used to break world records on off-the-shelf hardware for TCP/IP data transfer, and the IETF made an argument to standardize it. CalTech and the founders decided not to standardize FastTCP, since that would take too long and be too political. Instead, they launched the company, FastSoft.”
 
“We ‘fix’ TCP so that enterprises and service providers can use the Internet for distributing and monetizing video and other digital assets or content, anywhere in the world with no investment in remote technology,” continued Henderson. “We’re an asymmetrical acceleration product. TCP/IP wasn’t originally designed for video or for centralized distribution, which means in today’s world that web-based services are not able to run well over long distances, because the traffic runs into congestion, and video and monetizing video, which is so important, is difficult because of the low quality of the viewing experience, when watched over typical TCP connections. The good news is that TCP has scaled incredibly well, but it has reached its breaking point.”
 
“By some accounts only 10 percent of digital media is currently online,” said Henderson. “But we’re not just talking about video like YouTube. In fact, we’re referring to long-form content that can be delivered through IP networks to your TV, to set-top boxes, even in workflow around the world. Huge files need to be delivered with quality of experience and for workflow. Obviously cloud computing, centralized computing and storage, centralized delivery, and SaaS offerings are all part of the landscape. It’s all coming down to centralized architectures and Internet-based delivery platforms, and of course the companies that are actually monetizing content online.”
 
Henderson elaborates, “There are three broad categories of companies that need our product: First, there are Internet content providers, the companies that are selling content online in some way, shape or form, and this can be media, business information, and so forth. Second are the content delivery networks themselves. Third are the managed hosting providers. These are the market drivers in which the problem of TCP can be found. We’ve seen that both enterprises and service providers want to centralize, for many reasons. Enterprises want to share their digital content and assets with their partners and sometimes the world at large. It’s much easier to manage and control your digital assets when they’re in a centralized form. What enterprises and service providers want and can’t achieve is to have their digital content available all over the world quickly and effectively, but without having to build regional datacenters.”
 
“Our FastTCP replaces TCP,” said Henderson. “No technology is needed on the receiving side, only at the server side. The workings of our device are transparent both to the server/sending side and the receiver. It’s transparent to the serving side because it’s implemented as a Layer 2 bridging device, so it looks like a wire-in/wire-out device. We can be transparent to both sides because our FastTCP is a native TCP variant that’s 100 percent standards-compliant, so the receiving TCP stack sees it as simply TCP. We’ve just announced the E1000, which is a 1 gigabit per second model. It’s a 1U high server. It’s doesn’t use compression or caching. It is what we call ‘mile agnostic’. Because the protocol is at Layer 4, it doesn’t ‘see’ the first, middle, or last mile. It just sees the path from the sending TCP stack to the receiving TCP stack, and optimizes that path. We price the device based on capacity. The models come in 10 Mbps, 50 Mbps and 100 Mbps and now our 1 Gbps products. We’re working on models with even higher throughput.”
Speeding Up XML, Too
 
XML, the Extensible Markup Language, is a general-purpose specification for creating custom markup languages, but is actually considered to be an extensible language since users can define their own elements. It helps information systems share structured data, particularly via the Internet. Businesses today are exploiting XML and Web Services to do rapid application integration internally across the enterprise and externally with trading partners, suppliers and end-customers. However, XML often results in bloated message sizes, further complicated by adding SOAP protocols to the mix, as well as security tokens and encryption. This all results in performance degradation. Improved forms of XML integration based on high performance, such as what’s needed in a full-blown roll out of a SOA, can be achieved with products such as Vordel’s Vordel 5, a set of enterprise XML network infrastructure products for general XML-based integration and SOA.
 
The Vordel XML Gateway (News - Alert) employs the Vordel XML Accelerator (VXA) engine, to accelerate processor-intensive XML processing tasks. VXA maintains high levels of throughput without compromising security, abstraction, or policy enforcement.
 
Vordel’s Hugh Carroll, vice president of marketing, said, “We’re an XML network management company. We provide high performance enterprise-level hardware and software products that enable enterprises confidently to deploy their SOA. Essentially, we’re accelerating, managing and protecting XML applications. By doing that we allow organizations to govern their usage of Web Services and ensure that they deliver on the performance of those services in terms of their internal and external customers. We’ve very much a horizontal infrastructural play. We’re a key infrastructural component of the SOA approach. However, some of the key verticals in which we see our products include governments — U.S., E.U. and Council Governments. In the financial services space we deal with Allianz and Fortis and many Fortune 500 companies. In the telecom space, we have many large clients in North America and Europe using our product.”
 
Mark O’Neill, CTO of Vordel, stated, “XML is perhaps the fundamental technology that’s driving everything around Web Services. It underlies and enables all of the new services deployed out to consumers. Personalization, billing and all of that depends on XML.”
 
“One of the big challenges we’ve seen from our telecoms customers centers on performance and latency when services are deployed out to consumers,” said O’Neill. “Any type of latency is very apparent to them. One challenge of XML is that it’s verbose, slow to parse, the documents are just bigger and the messaging is bigger than some competing technologies. Our appliances use what’s called XML acceleration. They quickly parse XML, validate it and transform it, faster than what can be done in an application server. That reduces the latency and it means that people can use XML without their services suffering bad performance. That’s a key driver that brings people to XML appliance vendors such as Vordel.”
 
Service Delivery Frameworks for Everybody
 
Recently, I spoke with Martin Creaner, president of the TM Forum (TMF) and Lee Himbeault, senior strategy manager, TELUS Partner Solutions, and Business Committee Chair of the IPsphere program. The TM Forum is an industry association focused on transforming business processes, operations and systems for managing and monetizing on-line information, communications and entertainment services.
 
The TMF had just announced that they had joined forces with the IPsphere Forum (IPSF) so that TMF could broaden the scope of its work and initiatives.
 
“Historically, the TM Forum is the industry association for telecom software, particularly all software related to Operational Support Systems and Billing Support Systems (OSS/BSS),” Craner begins. “It was founded back in the late 1980s and it has evolved and grown since then. It was a forum for big operators and service provider to get together. It now consists of a wide range of operators, software vendors, systems integrators, consultancy businesses, and so forth. There are now about 708 member companies from 75 countries.”
 
Himbeault adds, “IPsphere is a younger organization formed in 2005 and was set up to look at the challenge of how new syndicated services are going to be delivered across very complex value chains, where multiple different parts of each service are provided by a range of different operators across the world. The IPsphere Forum has about 27 members, many are also TM Forum member companies.”
 
“During the last 12 months we’ve had a lot of discussion between the two organizations,” said Creaner. “Because the TM Forum since around 2006 has kicked off a program looking at the whole area of Service Delivery Frameworks (SDFs), which is really the next generation of service delivery software and structures being put in place by operators in fixed, mobile and cable, so they can offer advanced new services to their customer base. Our program attempted to identify the confusing points about that. As we delved into it, we became more and more aware of the IPsphere Forum, which had been tackling the same problems, and perhaps coming at it from a different perspective — initially more from a network transport perspective, but increasingly moving into the service software domain and addressing essentially the same problems: How do you syndicate services; how do you create frameworks to allow sophisticated service creation with a service provider environment?”
 
“So, as part of our SDF activity in the TM Forum, we also kicked off discussion with the IPsphere Forum and they attended many of our meetings,” said Creaner. “They became more involved and in February 2007 we discussed bringing the two organizations together. The advantage of this is two-fold: First, it creates more of a critical mass, all focused on trying to resolve the problem. That’s good for both organizations. The TM Forum can bring some companies to the table to which IPsphere would never have access, and IPsphere can bring in some additional companies and also bring to bear several years of expertise. Second, it keeps our work from diverging, which is always a concern with your dealing with standards initiatives. So we all felt it was a good idea to bring both organizations together.”
 
“IPsphere’s work on advanced service delivery impacts on a range of TM Forum work, including the TM Forum program on Service Delivery Networks,” added Creaner.
 
“IPSF’s work will soon be integrated with the TMF’s ongoing Service Delivery Framework. Together, we will provide a more encompassing set of best practices, standards and guidelines to drive the development and roll-out of IP-based information, communications and entertainment services,” said Himbeault. “Once we get beyond initial demonstrations of up-to-date specifications and field trials, we will be looking at how we can integrate the IPsphere activities much more tightly, not just into our service delivery framework activities, but also into our programs that examine the fundamental business processes of the telcoms and communications industry and the common information models, and so forth.”
 
The TMF and its partners realize that Fixed-Mobile Convergence (FMC) means more than “any service, on any network, to any device”. Personalized information and entertainment will require all-encompassing, truly standardized approaches for managing end-user devices (e.g, mobile handsets, TV, PC, games consoles, etc.).
 
Additionally, the TM Forum has announced that it has been working with the U.K. Government (among others) to establish a new industry program aimed at serving the procurement needs of large-scale end-user organizations. These buyers purchase IT and communications technology on a scale equivalent to many service providers, so many of the Forum’s standards and guidelines and best practices are directly applicable to them.
 
“The needs of large-scale service providers are becoming indistinguishable from the needs of large-scale users such as governments, defense industries, aerospace and financial services industries,” said Creaner. “The needs for an integrated procurement approach are rising with the advent of convergent approaches such as SOA, all IP networking, and advanced SDPs. Our new program is expected to focus on ensuring that clear procurement guides are developed to assist large-scale buyers in their procurements and market education.”
 
Companies mentioned in this article:
 
Boomi
www.boomi.com
 
FastSoft
www.fastSoft.com
 
IONA
www.iona.com
 
TeleManagement Forum & IPSphere
www.tmforum.org
 
Vordel
www.vordel.com
 

Richard Grigonis is Executive Editor of TMC�s IP Communications Group. To read more of Richard�s articles, please visit his columnist page.

Edited by Michelle Robart
FOLLOW US

Subscribe to InfoTECH Spotlight eNews

InfoTECH Spotlight eNews delivers the latest news impacting technology in the IT industry each week. Sign up to receive FREE breaking news today!
FREE eNewsletter

infoTECH Whitepapers