Thursday, January 17, 2008

IBM and Kapow on how enterprises exploit application mashups and lightweight data access

Listen to the podcast. Read a full transcript. Sponsor: Kapow Technologies.

The choices among enterprise application development and deployment technologies has never been greater. But what's truly different about today's applications is that line of business people can have a greater impact than ever on how technology supports their productive work.

By exploiting mashups, situational applications, Web 2.0 techniques and lightweight data access, new breeds of Web-based applications and services are being cobbled together fast, cheap, and without undue drain on IT staffs and developers. Tools and online services both are being used to combine external web services like maps and weather with internal data feeds and services to add whole new dimensions of business intelligence and workflow automation, often in a few days, often without waiting in line in order to get IT's attention.

And while many of these mashups happen outside of IT's purview, more IT leaders see these innovative means as a productivity boon that can't be denied, and which may even save them time and resources while improving IT's image in the bargain. The trick is to manage the people and new processes without killing off the innovation.

To help weed through the agony and ecstasy of Enterprise 2.0 application development and deployment in the enterprise, I recently chatted with Rod Smith, Vice President of Internet Emerging Technologies at IBM, and Stefan Andreasen, the Founder and CTO of Kapow Technologies.

Here are some excerpts:
In times of innovation you get some definite chaos coming through, but IT and line of businesses see this as a big opportunity. ... The methodology here is very different from the development methodology we’ve been brought up to do. It’s much more collaborative, if you’re line of business, and it’s much more than a set of specifications.

This current wave is really driven by line of business getting IT in their own hands. They’ve started using it, and that’s created the chaos, but chaos is created because there is a need. The best thing that’s happening now is acknowledging that line-of-business people need to do their own thing. We need to give them the tools, environments and infrastructure so they can do it in a controlled way -- in an acceptable, secured way.

... As we opened up this content [we found] that this isn't just about IT managing or controlling it. It’s really a partnership now. ... The line of business wants to be involved when information is available and published. That’s a very different blending of responsibility than we've seen before on this.

There is a lot of information that’s out there, both on the public Web and on the private Web, which is really meant to be human-readable information. You can just think about something as simple as going to U.S. Geological Service and looking at fault lines of earthquakes and there isn't any programmatic API to access this data.

This kind of data might be very important. If I am building a factory in an earthquake area, I don’t want to buy a lot that is right on the top of a fault line. So I can turn this data into a standard API, and then use that as part of my intelligence to find the best property for my new factory.

It’s just not internal information they want. It's external information, and we really are empowering these content developers now. The types of applications that people are putting together are much more like dashboards of information, both internally and externally over the Internet, that businesses use to really drive their business. Before, the access costs were high.

Now the access costs are continuing to drop very low, and people do say, "Let’s go ahead and publish this information, so it can be consumed and remixed by business partners and others,” rather than thinking about just a set of APIs at a low level, like we did in the past with Java.

If you want to have automatic access to data or content, you need to be able to access it in a standard way. What is happening now with Web Oriented Architecture (WOA) is that we're focusing on a few standard formats like RESTful services and on feeds like RSS and Atom.

So first you need to be able to access your data that way. This is exactly what we do. Our customers turn data they work with in an application into these standard APIs and feeds, so they can work with them in an automated way. ... With the explosion of information out there, there's a realization that having the right data at the right time is getting more and more important. There is a huge need for getting access in an automated way.

The more forward-thinking people in IT departments realize that the faster they can put together publishable data content, they can get a deeper understanding in a very short time about what their customers want. They can then go back and decide the best way to open up that data. Is it through syndication feeds, XML, or programmatic API?

Before, IT had to guess usage and how many folks might be touching it, and then build it once and make it scalable. ... We've seen a huge flip now. Work is commensurate with some results that come quickly. Now we will see more collaboration coming from IT on information and partnerships.

What is interesting about it is, if you think about what I just described -- where we mashed in some data with AccuWeather -- if that had been an old SOA project of nine or 18 months, that would have been a significant investment for us, and would have been hard to justify. Now, if that takes a couple of weeks and hours to do -- even if it fails or doesn’t hit the right spot -- it was a great tool for learning what the other requirements were, and other things that we try as a business.

That’s what a lot of this Web 2.0 and mashups are about -- new avenues for communication, where you can be engaged and you can look at information and how you can put things together. And it has the right costs associated with it -- inexpensive. If I were going to sum up a lot of Web 2.0 and mashups, the magnitude of drop in “customization cost” is phenomenal.

What’s fun about this, and I think Stefan will agree, is that when I go to a customer, I don’t take PowerPoint charts anymore. I look on their website and I see if they have some syndication feeds or some REST interfaces or something. Then I look around and I see if I can create a mashup of their material with other material that hadn’t been built with before. That’s compelling.

People look and they start to get excited because, as you just said, they see business patterns in that. "If you could do that, could you grab this other information from so-and-so?" It’s almost like a jam session at that point, where people come up with ideas.
Listen to the podcast. Read a full transcript. Sponsor: Kapow Technologies.

Wednesday, January 16, 2008

Sun refuses to give up on software acquisitions, buys MySQL for $1 billion

We knew that Sun has been lusting after a real software business in addition to Solaris. We knew that Sun "shares" -- that it digs open source, including Solaris and Java. And we knew that Sun had a love-hate relationship with Oracle and a hate-hate relationship with IBM and Microsoft.

So toss this all in a big pot, put on simmer and you get a logical -- if not three years too late -- stew: Sun Microsystems intends to buy MySQL AB and its very popular open source database. The announcement comes today with a hefty price tag of $1 billion.

The MySQL purchase by Sun makes more sense than any other acquisition they have done since they botched NetDynamics 10 years ago. This could be what saves Sun.

Sun can make a lot of mischief with this one, by taking some significant oxygen out of its competitors' core database revenues. Sun can package MySQL with its other software (and sell some hardware and storage, to boot), with the effect that the database can drive the sales of operating systems, middleware and perhaps even tools. Used to be the other way around, eh? Fellow blogger Larry Dignan sees synergies, too. And Tony Baer has some good points.

Who could this hurt if Sun executes well? IBM, Oracle, Microsoft, Sybase, Red Hat, Ingres. It could hurt Microsoft and SQL Server the most. Sun could hasten the tipping point for the commercial relational database to go commodity, like Linux did to operating systems like Unix/Solaris. Sun could far better attract developers to a data services fabric efficiency than with its tools-middleware-Solaris stack alone. As we recently saw, with Microsoft buying Fast Search & Transfer, the lifecycle of data and content is where software productivity begins and ends.

Sun will need to do this right, which has its risks given Sun's record with large software acquisitions. And Sun won't get a lot of help ecology-wise, from any large vendors. This puts Sun on a solo track, which it seems to prefer anyway. I wonder if the global SIs other than IBM will grok this?

Yes, it makes a lot of sense, which makes the timing so frustrating. I for one -- and I was surely not alone -- told very high-up folks at Sun to buy and seduce MySQL three years ago. (I also told them to merge with SAP, but that's another blog.) When Sun went and renamed it's SunONE stack to the Java what's-it-all, I warned them it would piss off the community. It did. I also told them Oracle was kicking their shins in. It did. I said: "Oracle has Linux, and you have MySQL." Oh, well.

[Now, Oracle has BEA, which pretty much dissolves any common market goals that Oracle and Sun once had as leaders of the anti-Microsoft coalition. The BEA acquisition by Oracle was a given, hastened no doubt to the close by the gathering gloom of of a U.S. economic recession.]

I'm glad the Sun-MySQL logic still holds, but Oracle has already done the damage with Linux, we saw how that Unix-to-Linux transition put Sun on its knees, and on the defensive. And we know that Sun has only been able to get one leg up since then, albeit refraining from falling over completely. Now, with BEA, Oracle with its Linux and other open source strengths -- not to mention those business apps -- will seek to choke out the last light from Sun, and focus on IBM on the top end, and Microsoft on the lower end. As Larry Ellison said, there will be room for only a handful of mega-vendors -- and we cannot be assured yet that Sun will meaningfully be one of them (or perhaps instead the next Unisys).

Indeed, the timing may still have some gold lining .... err, silver lining. Sun has had to pay big-time for MySQL (a lot more than if they had taken a large position in the AB two years ago). And what do they get for the cool $1 billion? Installed base, really. Sun says MySQL has millions of global deployments including Facebook, Google, Nokia, Baidu and China Mobile.

There's more, though. The next vendor turf battles are moving up yet another abstraction. Remember the cloud thing? Sun in sense pioneered the commercialization of utility computing, only to have Amazon come out strong (and added a database service in the cloud late last year). IBM has cloud lust. Google and Microsoft, too. Sun's acquisition of MySQL could also help it become a larger vendor to the other cloud builders, ie telecos, while seeding the Sun cloud to better rain down data services for its own users and developers.

And that begs the question of an Oracle-BEA cloud. Perhaps a partnership with Google on that one, eh? Then we have the ultimate mega-vendor/provider triumvirate: Apple-Google-Oracle. It's what Microsoft would be if it broke itself up properly and got the anti-trust folks off of their backs (not to mention a reduction in internal dysfunction). And that leaves loose change in the form of Sun, IBM, Amazon, eBay, and the dark horses of the telecos. Sun ought to seduce the telcos, sure, and they know it. Problem is the telecos don't yet.

Google may end up being the cloud king-maker here, playing Oracle and Sun off of one another. Playing coy with IBM, too. Who will partner with Amazon? Fun times.

Surely if Sun can produce a full-service cloud built on Solaris-Intel-Sparc that includes low-energy-use virtualized runtimes, complementary tools, and integrated database -- and price it to win -- well, the cloud wars are on. Sun might hang on for yet another day or two.

Tuesday, January 15, 2008

MuleSource takes aim at SOA governance, launches subscription-based ESB

MuleSource, a provider of open-source service-oriented architecture (SOA) infrastructure software, has jumped into the SOA governance pool with the community release today of Mule Galaxy 1.0.

Galaxy, an open-source platform with integrated registry and repository, allows users to store and manage an increasing number of SOA artifacts and can be used in conjunction with the Mule enterprise service bus (ESB) or as a standalone product. It was also designed with federation in mind, being pluggable to other registries.

In other news today, Mule also announced a subscription-only version of its ESB, as well as a beta version of Mule Saturn, an activity monitoring tool for business processes and workflow.

The subscription ESB smacks of "Mule" It will be interesting to see how well this does in terms of uptake. Integration as a service seems to be gaining traction. We're also told this "ESB in the cloud" supports IBM CICS, which is interesting ... are we approaching transactional mashups en masse?

As enterprises use SOA to expand their consumption of services from both inside and outside the business, governance becomes an all-important issue for control. Galaxy provides such registry and repository features as lifecycle, dependency, and artifact management -- along with querying and indexing.

A RESTful HTTP Atom Pub interface facilities integration with such frameworks as Mule, Apache CXF, and WCF. Galaxy also provides out-of-the-box support for various artifact types, including Mule, WSDLs, and custom artifacts.

Galaxy can be downloaded now, and a fully tested enterprise edition will be available in Q2 for Mule Enterprise subscribers.

On the ESB front, Mule has taken aim at the Fortune 2000 customer base with the introduction of Mule 1.5 Enterprise Edition, a subscription-only commercial enterprise packaging of the Mule ESB integration platform. Prior to this announcement, the ESB had been available only in the community edition.

It's sort of funny, as commercial providers offer open source versions of their products, we also see open source providers handing up commercial versions. I guess that means everyone needs one of each? Perhaps the versions (ala Fedora to RHEL) are becoming alike, in that it takes a subscription of some sort to get the real goods and use them.

Take the traffic when you can, I've always said. Mule's popularity was in evidence in November, when the company announced that community downloads had surpassed one million.

The new enterprise offering is available for a single annual fee and encompasses news features, including:

  • Support for Apache CXF Web Services Framework
  • Patch management and provisioning via MuleHQ
  • Streaming of large data objects through Mule without being read into memory
  • Nested routers to decouple service implementations from service interfaces
  • Support for multiple models
  • Diagnostic feedback for customer support

More information is available from the MuleSource site.

For users looking for a business-activity monitoring tool, MuleSource has released a beta version of Mule Saturn 1.0, which is designed to complement an SOA infrastructure by providing detailed logging and reporting on every transaction that flows through the Mule ESB.

Saturn allows staff to drill down on transaction details and set message-level breakpoints for deep log analytics, allowing for continuous custom improvement. Key features include:

  • Business user view into workflow and state
  • Process visualization
  • Search on transaction, date, various ID
  • Reporting on service-level agreements

Saturn is available immediately to MuleSource subscribers.

Monday, January 14, 2008

WSO2 Web services framework builds bridge between Ruby and enterprise apps

WSO2 has built a bridge between Ruby-based applications and enterprise-class Web services with the introduction of its Web Services Framework for Ruby (WSF/Ruby) 1.0.

WSF/Ruby, an open-source framework for providing and consuming Web services in the Ruby object-oriented programming language, offers support for the WS-* stack, allowing developers to combine Ruby with security and messaging capabilities required for enterprise SOAP -based Web services. Disclosure: WSO2 is a sponsor of BriefingsDirect podcasts.

WSO2 Chairman/CEO Sanjiva Weerawarana explained the bridging capabilities in a pre-release interview with Infoworld:

While Ruby has been popular in the Web 2.0 realm, sometimes it needs to talk to legacy architectures, he said. With the new framework, developers could build a Web application using Ruby and then hook into enterprise infrastructures, such as JMS (Java Message Service) queues. For example, a Web site might be built with Ruby that then needs to link to an order fulfillment system based on an IBM mainframe or minicomputer, Weerawarana said.

With WSF/Ruby, developers can also consume Web services with Representational State Transfer (REST). WSF/Ruby also provides a fully open-source Ruby extension based on Apache Axis2/C, Apache Sandesha2/C, and Apache Rampart/C.

WSF/Ruby features both client and service APIs. The client uses the WSClient class for one-way and two-way service invocation support. The service API for providing Web services used the WSService class with support for one-way and two-way operations. Both APIs incorporate WSMessage class to handle message-level options.

WSF/Ruby 1.0 supports basic Web services standards, including SOAP 1.1 and SOAP 1.2. It also provides interoperability with Microsoft .NET, the Apache Axis2/Java-based WSO2 Web Services Application Server (WSAS), and other J2EE implementations. Key features of WSF/Ruby 1.0 are:

  • Comprehensive support for the WS*- stack, including the SOAP Message Transmission Optimization Mechanism (MTOM), WS-Addressing, WS-Security, WS-SecurityPolicy, and WS-Reliable Messaging.
  • Secure Web services with advanced WS*-Security features, such as encryption and signing of SOAP messages. Users also can send messages with UsernameToken and TimeStamp support.
  • Reliable messaging for Web services and clients.
  • REST support, so a single service can be exposed both as a SOAP-style and as a REST-style service. The client API also supports invoking REST services using HTTP GET and POST methods.
  • Class mapping for services, enabling a user to provide a class and expose the class operations as service operations.
  • Attachments with Web services and clients that allow users to send and receive attachments with SOAP messages in optimized formats and non-optimized formats with MTOM support.
According to WSO2, WSF/Ruby has been tested on Windows XP with Microsoft Visual C++ version 8.0, as well as with Linux GCC 4.1.1.

LogMeIn files for IPO, sets up the market for cloud-as-PC-support continuum

I see that remote PC services start-up LogMeIn is going to conduct an IPO on Nasdaq in the not too distant future, pointing up the vibrancy of the intersection of cloud computing and the personal computer.

And the encouraging growth that LogMeIn has enjoyed shows that cloud, remote maintenance and the long-term health of the PC are all quite mutually compatible, thank you. Microsoft has is right when they chime about "software and services," just as there will be for a long time the need for PCs and the cloud services that they will increasingly rely on.

So congrats to LogMeIn, they are a great bunch of folks. Disclosure: LogMeIn has been a sponsor of BriefingsDirect podcasts. I am sure glad I had that chat about the Web as operating system way back when with Mike and Joe.

This intention of filing seems only the beginning of LogMeIn's next phase. According to the filing, LogMeIn plans to raise up to $86 million from the IPO, but this could change. It may not be that large of a sum, but it shows how Internet firms don't require the capital they used to to grow substantially. And there'a always the possibility of LogMeIn making acquisitions to fill out its services and support portfolio.

Nice thing about the LogMeIn services is that they straddle the consumer, SOHO, SMB and enterprise markets. The services can cut across them all -- adding value while cutting costs on the old way of doing things. Nice recipe these days. More telcos and service providers will need such abilities too.

As I've said, I expect to see more telcos buying software and services vendors in 2008 to expand their offerings beyond the bit-pipe and entertainment content stuff. If you can serve it up on subscription, well then do it broadly and monetize the infrastructure as many ways as possible.