Gennaro "Jerry" Cuomo is an IBM Fellow and one of the founding fathers of IBM WebSphere Software. Jerry is a breakthrough innovator of solutions in the areas of web application servers, Java, TCP/IP, real-time collaboration software, and high-performance transactional systems. Jerry is currently the Chief Technology Officer of WebSphere, where his prime charter is to “cultivate the future of WebSphere”.

Tree Hugging at Energy Camp

Last week, I waited on line 50 minutes to fill up my car with gas because of a mini-panic brought on by Hurricane Ike. This week, gas prices in N.C. are over $4 US. Needless to say conserving ENERGY is on my mind and it's time to take action.

On Tuesday, I co-hosted Energy Camp at Interop 2008 in NYC. My co-host was James Govenor, from Redmonk and we used his popular Unconference format again (we also did an Unconference at IMPACT 2009).

I delivered the Energy Camp keynote, on Tuesday morning. As many of you know, I like to show short home-made videos on the theme of my talks. During this pitch, I've decided to garnish it with some of IBM's "green-themed" TV commercials, which I think are hilarious. E.g., Green Data Center Man.

Setting the stage, I opened with some facts on how electricity use in the Data Center is of control. As of 2006, the electricity use attributable to the nation’s servers and data centers is estimated at about 61 billion kilowatt-hours (kWh), or 1.5 percent of total U.S. electricity consumption. The yearly bill had double since 2000, costing $4.5B and equivalent to the electricity consumed by 5.8 million average U.S. households and is similar to the amount of electricity used by the entire U.S transportation manufacturing industry, which includes the manufacture of automobiles, aircraft, trucks, and ships.
(US Census Bureau 2006, US DOE 2005/2007)

I had a few key messages:

• Re-Think IT - My main message for the keynote was you can save energy incrementally by targeting select technologies in your data center. Hardware (servers) are usually the first place people look to saving energy. In many cases this brings (low) double digit savings (which can be significant depending on the size of your data center.) However, your biggest return will come when you work across the stack and have your hardware, software and business processes work together to save energy. The higher you go up the stack, the more visibility and insight exists (into the inefficiencies) that if optimized, can lead to multiple X factors in energy savings. I gave two examples energy saving by working higher up the technology stack. The first I called Energy Savings via Rainmaking, the other Energy Savings via Smart SOA.
  
  
• Energy Saving via Rainmaking The idea here is to have your middleware and hardware work together together to more efficiently save energy by apply cleaver application optimization techniques. There are two significant control points, that provide the "dials" which allow you to "tune in" your energy savings. The first aspect has to do with controlling the shape of the work as it flows into your data center. We are working on "application aware" edge servers that are guided by higher-level operational and business policies to ensure that the important work flows more fluidly into your systems. The next facet has to do with controlling the shape of the applications within your data center (or shared resource pool/cloud). This capability ensures that the compute cloud is working on supporting the application with the highest business impact, and gracefully degrades the services for work that has less importants. But applying these techniques, you can do more in your data center with significantly less hardware and software - leading to impressive energy and cost savings.

  I also ran through some cool animations of our WebSphere Virtual Enterprise product to drive some of these points home.

• Saving Energy with Smart SOA The idea here is to get the business user involved to more efficiently direct the flow of business processes in your enterprise. The first step in making this happen is to close the communication gap between IT and Business - by having the business user use tools like a graphical business modeler and monitor to electronically express their business processes. By sharing business processes in electronic form, the business user can more efficiently work though "what if" simulations looking for opportunity to cut out paper use, or find more efficient routes to market. Once the business process is deployed, IT, using a Service Oriented Architecture, can map the business process to a set of loosely coupled services, giving flexibility to change and react efficiently to changing business times - before to much IT is wasted on the more inefficient approach. Lastly, by monitoring the business process the business user can use empirical data to better optimize and reduce waste within the processes - leading to a continue refinement of their processes and a potentially significant saving on the energy use by people and systems.

  I also ran through some compelling animations of our WebSphere Business Modeler and Monitor products to illustrate these points.

I don't consider myself a Tree-hugger. However, I know that just a little technology will go a long way to paint your data centers a little greener. Let's spread the word and get with it - we will also save some green along the way.

My Virtual Knee and video interview on InfoQ

Virtualized knee to assist in surgery

I seriously injured my knee during a Hapkido class in July. I am now recovering from surgery, which took place two weeks ago. The surgery involved repairing a fracture of my knee cap (which required drilling holes in my knee), replacing my ACL (with an allograft, using tissue from a cadaveric donor), and repairing tears in my meniscus and patellar. Recovery has been slow, but steady.

Being my first surgery, I was really impressed with how technology was used during the medical care process. An MRI provided a 3D image of my knee, which was the basis for creating virtual models of what the repair process would entail. I have to say, working on "what if" scenarios on my virtual knee did not hurt at all! I wish I can say the same for the real surgery. My virtual knee helped me (an engineer) clearly understand my (medical) options. This is a pretty powerful use of virtualization.

As you all know, virtualization is a very hot topic. Whether we are talking about virtualizing "my knee" or a customers data center, the value in this technique seems to transcend across multiple fields of use.

Right before my knee operation, Floyd Marinescu, from InfoQ, did a video interview of me on the topic of virtualization and cloud computing, titled:

Floyd asked some good questions and it was a lot of fun providing the answers.

Questions included:

• What is virtualization and why does it matter...
• What is cloud computing and how does it fit into vitualization
• What is IBM doing in the cloud computing area?
• Explain Rainmaking?
• Explain the concepts of Atomic and Molecular virtualization...
• Who does WebSphere Virtual Enterprise fit in?

Whether it's your knee or your data center (or both) that is causing you problems, virtulization technology has a way of easing the pain.  If the pain really persists, take 10mgs of Oxycodone, as prescribed by your doctor, every 4-6 hours as needed for pain... at least that's what is helping me get though my knee pain right now :-)

Top Performances of the Dark Knight

Heath Ledger - Top Joker Performance

I was one of a few hundred folks to stay up Thursday night to watch the opening of the new Batman movie, "The Dark Knight". I was really impressed with Heath Ledger's Joker performance. Ledger set a new standard with his meticulous attention to details, from his slightly off-centered walk, to his creepy fixations with his facial scare. The performance was so remarkable that, next January, he will likely earn a posthumous Oscar nomination.

The Joker role is a villain benchmark of sorts. The Joker character has appeared in numerous Batman related shows; portrayed by Cesar Romero in the 1960s Batman television series; Jack Nicholson in the 1989 film Batman, (Nicholson's Joker ranks #45 in the American Film Institute's list of the top 50 film villains); voice actor Mark Hamill in the 1990s Batman: The Animated Series television series (also very creepy); and Heath’s recent rendition. The baseline Joker character (white face, red lip paint, exaggerated grin, and wild outbursts of laughter) is visible within all players; each brings their own unique attributes. In a way, each rendition builds from the others and enhances the overall art and gives the consumer a better product.

Whether you’re comparing portraits of the Joker, or the performance of Application Servers, Performance Benchmarking has always been a very interesting topic to me. A benchmark is used to assess the relative performance of systems, normally by running a number of standard tests and trials against them. In fact, I was one of the original authors of the Trade (IBM internal JEE) benchmark, which over the years has been enhanced and evolved (into a SPEC standard) by the some of our finest performance engineers (Stan Cox, John Stecher, Chris Blythe, and Matt Hogstrom). 

When there is critical mass of interest around a particular topic, benchmarks attempt to level the playing field around a set of standardized tests and provide the consumer with information on capability, performance and sometimes even price. Benchmarks have served the Application Server market very well over the years (as they have the Database and CPU evaluation process). As dominant workload patterns emerge in the industry, organizations like TPC and SPEC have done a solid job of fostering the creation of benchmark specifications and reference implementations. SPECJAppServer (which is a loose derivative of our very own Trade2 benchmark), has been helping customers compare the nuances of the Java-based Application Servers from IBM, Oracle (BEA) and Sun. I’ve consistently been impressed with the coop-petition that takes place during the creation of a benchmark. Benchmarking also create a “leapfrog effect”, where one vendor’s leading results being published, motivates the next vendor to really sharpen up their server to leapfrog the results of the other. In the earlier days of JEE benchmarking, we played some pretty intense leapfrog games with BEA in particular – in the end benefiting the customer with major advances in capability and price/performance.

With the emergence of SOA, a new breed of workload patterns have emerged.  SOA is the next clear candidate for an industry benchmark. In fact, my colleague and friend, Andrew Spyker, just recently blogged on the progress of such a benchmark. A SOA benchmark seems a little trickier because SOA is not as precisely defined as JEE. However, I would think such a benchmark would address areas like using web services standards, intermediary interaction patterns (i.e., interactions found in an ESB; transforming protocols and payloads) and business process flow management. I would hope the TPC evolves this new SOA benchmark in these directions.

As an Industry, I would like to see us rally around this benchmark. SOA is important to our customers. As vendors of SOA technology, we need the “leapfrog effect” to kick in, where each performance builds from the others and enhances the overall art and gives the consumer a better product. 

And like before, I am sure that WebSphere will emerge as the Dark Knight, steady, reliable, performant with many tricks up its sleeve. And I’m also sure there will be many Jokers out there looking to steal the "Academy Award" from the Dark Knight.

DataPower-lution

IBM introduced the Type 285 electric bookkeeping and accounting machine in 1933.

Purpose-Built Systems

Ryan Betts, from the WebSphere DataPower team, turned me on to a paper, written by Jim Barton- CTO and co-founder of Tivo, called Tivo-lution. The paper was inspiring as it confirmed the motivations and aspirations that I've had ever since I led IBM's acquisition of DataPower in 2005. In the paper, Barton describes the challenges in making complex systems usable and how "purpose built" computer systems are one answer to the challenge.

We in IBM are all over this!   Heck. The name of our company implies we get it.  International "BUSINESS MACHINES".   As this photo illustrates, IBM has a long history in building purposed machines, like this 1933 Type  285 - Electric bookkeeping and accounting machine.  I can imagine this puppy being delivered to an accountant, plugging it in, and away they go.  They didn't have to worry about hard drive capacity, operating system levels, compatibility between middleware vendors, or application functionality.   It just did the job.   I can also imagine it followed the 80/20 rule.   It probably didn't do 100% of what all accountants needed. But it probably did 80% of what all accountants needed, very well.  And you just dealt with the remaining 20%, or learned to live without it.

"Business Machines, Again".  This is my inspiration.   Tivo gets it.   IBM gets it... And our customers are starting to really buy into it in a big way.   It's all about time-to-value and total cost of ownership/operation.   And appliance like our WebSphere DataPower XI50, deliver on these attributes.

At the extreme, purpose built systems, like a Tivo DVR and a XI-50, are built from the ground up for their purpose.   While they might use off the shelf parts, like an embedded Linux OS, it is important that all part are "right sized" for the job.   Right-sizing source code, in a hardware appliance is more like firmware (with strong affinity to the underlying hardware) than it is software.   As such, the Tivo-lution paper describes the need to own every line of source code to ensure the highest level of integration and quality:

The Tivo team even modified the GNU C++ compiler to elimiate the use of exceptions (which generate a lot of code that is seldom used) in favor of rigid checking of return code usage in the firmware.   DataPower similarly contains a custom XML compiler that generates both standard executable code, for its general purpose CPUs, as well as custom code for the (XG4) XML coprocessor card. 

A physical appliance has the unparalleled benefit of being hardened for security.  Jim talks about this in his Tivo paper.

Rich Salz, who has led the Security work for DataPower and now is our chief programmer for DP, has taught me the meaning of "tamper proof" appliances (or more precisely "tamper evident".  Like the 1982 Tylenol scare, we can't stop you from opening the box, but we can protect you, if someone does).   In fact, the physical security characteristics make the DP XS-40, one of the only technologies some of our most stringent customers will ever consider putting in their network Demilitarized Zone (DMZ).    If a DP box is compromised and opened, it basically stops working.   An encrypted flash drive makes any configuration data, including security keys, very difficult to exploit.   "DP is like the roach motel, private keys go in, but never come out", is the way Rich describes one of the tamper proof qualities of the XS-40. 

But the truth is, DP is not a DVR.  DP is a middleware appliance.  Middleware is a tricky thing to make an appliance out of.  Middleware is enabling technology and by its nature is not specific to any application or vendor.   The Tivo appliance is a very specific application (TV/guide) which makes it somewhat easier to constrain.

Hence, the challenge (and the art) in building a middleware appliance involves providing the right amount of constraint, without rendering the appliance useless.   For example,  DP does not run Java code (which is the primary means of customizing much of the WebSphere portfolio), instead it uses XML as the primary mode of behavior customization.   So, at some level, DP is not programmed, but instead it is configured.   Now, for those who have used XML (and its cousin XSLT), you know that it's more than configuration; however, it is a constraint over Java programming, which has unbounded levels of customize-ability.   Adolfo Rodriguez, who leads the integration of DP into the WebSphere family, has been bridging this gap (of special to general purpose) very effectively.  He and this team recently added features to DP to allow it to seamlessly connect to IBM mainframe software (e.g., IMS and DB2) as well as adding capabilities to manage a collection of appliances as if they were one.

We are blessed in IBM to have a very healthy general purpose software business.   Our WebSphere/Java-based middleware is the poster child for general purpose middleware (i.e., write once, run almost everywhere).    However, there is a place for business machines - purposed built devices that focus on providing the 80 part of the 80/20 rule.    We are heading down this path in a big BLUE way.

But there is more...

What about Virtual Appliances, you ask?

I recently was asked this question at a conference.  My knee jerk response was that the term "virtual appliances" was an oxymoron.   You can't have something be virtual and be an appliance at the same time. (e.g., a virtual toaster, producing virtual toast ?)  However, I've been working with my team on some ideas in this space and have begun to chill out a bit on the oxymoron thought.

You see,  solving for the 80% case (of 80/20) is a good thing.   Making more of the system "invisible" to the end-user by constraining the problem space is a great technique that can be applied at multiple levels.    My friend, Frank Kenny, from Gartner, shows a picture of a spectrum that I think makes sense.  At one end of the spectrum (the non-appliance end) you have CDs (install and do what you will), at the other end of the spectrum (the appliance end) you have purpose built- physical appliances.   As you progress from CD to hardened psychical appliance, you can constrain and add more purpose along the way- in increments.   Perhaps you can say a virtual appliance is some point on this spectrum half way between CD and Toaster.  Again,  I am starting to see the merit here; especially in enabling users to quickly get going with a complex software platform (development and test).    By making choices and pre-configuring and hardening the environment for specific task, and executing specific application topologies and patterns, we can help jump-start the process of application development and test in a significant way.    We are heading down this path in a big way as well... more blogging on this topic later.

So...

Business Machines, both physical and virtual, have their place.  As we continue to make strong strides towards introducing technologies that make interacting with your middleware systems as fluid as interacting with your television set.  Yeah, we have a lot of work to do, but we have a vision and we are making strong strides towards making it a reality.

WAS.NEXT; It's got some spring in its feet.

"Jerry kicking back, no worries about Spring or his feet"

The folks from SpringSource caused a buzz last week with an announcement of their Spring Application Platform beta. An article on InfoQ, broke the story and my Inbox lit up like a Christmas tree. Subject lines cited – “SpringSource is declaring war on WebSphere.” 

Really? I don’t quite see it that way. Let me start by saying this… I’m a fan of Spring, and I think that a foundation of OSGi and technologies like Spring framework are the future of Java-based application serving. The days of one-size fits all servers are evolving towards purpose built servers that specialize in specific types of workloads (e.g., Web).

Java EE is a standards based “profile” that has served our customers well for the past 10 years, and will continue to serve us. This is especially true when you consider the shear number of JEE applications that our customers have deployed over these years. But there are other profiles, for sure. For example, purposing servers as an Enterprise Service Bus, or a server that executes BPEL processes, or simply processes Web requests - this is where we are heading. These server “profiles” are purpose build and right-sized for the task at hand. 

In case you haven’t noticed, the WebSphere team, including Alan Little, Billy Newport and Ian Robinson, has been all over this (and have been recently blogging about this subject as well). WebSphere Application Server (WAS) v6.1 shipped (~2 years ago) an Equinox-based OSGi core and bundles for key internal components. V7 (open beta now available) goes even further in making the move to a “schizophrenic ” application platform (i.e., an application server supporting multiple application personalities), adding a service for explicitly installing these purposed profiles. The WAS team has also been focusing on integration with the Spring Framework, which has been certified with WAS since V6. 

So, again, this is where WebSphere and the industry were already heading – no new news there.

Now, I do feel it’s a shame that SpringSource went off and did this under a GPL license. The industry would benefit from an Apache-licensed “reference” application platform just like we (customers and vendors alike) benefited from having references like Apache Harmony and the Apache Geronimo server. But that’s okay because I believe this will logically happen in OSGi - where other industry players, including SpringSource, are already working productively. Who knows, perhaps we will even donate a bundle or two from WAS to jumpstart this effort.

This leaves plenty of room for capitalism. (After all, I do have a family to feed). Fit for purpose plug-ins will provide the extended value to our customer. The WebSphere-brand will major in providing “profiles” and quality-of-service (QoS) for those profiles. This is pretty consistent with how we’ve built out the WebSphere (and much of IBM Software Groups’) product-line. Specifically, we have long used WAS as a general-purpose run-time, hosting a wide variety of products and features including Process Server, ESB, Business Monitor, Data Grid, Portal, Commerce and much more. The OSGi+Spring combination will allow us to continue “right sizing” these products and more affectively create our product combinations (E.g., our process server includes our ESB function). This architecture will also help support our well received Feature Packs, that allow new function to be incrementally delivered in-between versions.

Our secret sauce, and the foundation for hundreds of patents and generally cool technology, has been in providing QoS for the WAS foundation, and the applications hosted on our servers; High Availably, Failure capture, Performance, Scale, Clustering, Monitoring, Trace and Systems Management are just of few of these features. Using techniques like “inversion of control”, we hope to also to insert to right qualities for the right task. 

Hence, the trend of evolving towards a platform with a right-sized, multiple personality complexion is not a disorder, nor a pending war between SpringSource and WebSphere –it is just the way the industry is evolving– and perhaps the Java community can pull together like it did before and produce a level field of opportunity. Our customers really appreciated this, the last go around.

Come on then, this is my call to the Industry players; Let’s meet in OSGi and Apache and do it again. Certainly this is where WAS.NEXT is - who else wants to play?

Impact 2008

6 days of peace, love, middleware and 6000 customer

Well, it’s here, IMPACT 2008. The WebSphere team is converging on Las Vegas for 6 days of peace, love and 6000 plus customers for a big dose of SOA and an impressive line up of IBM executive speakers, customer speakers, celebrities and new product announcements. 

This is going to be a very big week for the WebSphere brand. I am really excited about a couple of our announcements. After 10 months of incubating, we will announce our first product that utilizes the technology on projectzero.org. We are also announcing our business event-processing product. I will post an entry that expands on these – and more – later in the week.

A bunch of us are twittering about IMPACT. Dan Jemiolo, from the Project Zero team, wrote this cool widget, using PZ, that collects our recent twitter entries into a single frame.

My Whereabouts

On Monday

Monday Night

On Tuesday

On Wednesday

On Thursday

On Friday

Every spare moment in-between will be filled with customer conversations. Last year’s IMPACT was one of the most productive weeks of the year. The customer conference is extremely energizing. And while it may not be all peace and love, it does inspire me.

Rainmaking

Virtualization Progression - Machines, Servers, Clouds, SaaS

It seems like 'virtualization', ‘cloud computing’ and 'software as a service' are all big industry buzzwords that get tossed around with varying definitions. I am getting more and more questions about how these things play in the IBM and WebSphere space. As I study the attributes associated with these terms, I am starting to get a feel for how they fit together.

One interesting way to think about these topics is as a progression of virtualization - from the machine, to the server, to the cluster, to the application-environment, and then finally to the application itself. As you progress towards virtualized applications and application environments (inward towards SaaS, in the picture above) software producers and consumers become increasingly agile (which implies improving time to value).

I also see ‘appliances’ playing a role in how IBM materializes and delivers these technologies to enterprises. Let me elaborate.

A Progression of Virtualization

Virtualization: I think of virtualization as a foundational technology. Base virtualization technology starts with a focus on the machine. Computers have long virtualized memory, disk, and CPUs, but now we're seeing the industry focus on virtualizing the entire machine and operating system - the way VMWare, Xen, and PowerVM do. These hypervisors are similar to the logical partitions (LPARs) that have been around IBM for a long time in our System z and System p products.

As part of the virtualization of the machine and OS, middleware can be virtualized, allowing for virtual servers. This is where the fun starts. Once we have virtual servers, we have the basic building blocks. These ‘atomic’ pieces can stand alone, but are more interesting when linked with others image. Where atomic virtualization is focused on the art of virtualizing a single server, molecular virtualization is the art of creating a linked collection of virtual servers – e.g., linking a virtualized application server with a virtualized database server. Given our experience with customer facing services, we have a wealth of practical experience making our servers work together. We can “bake” our best-practices knowledge into the creation of these virtual molecules to create highly robust combinations of purposed middleware images. This molecular approach to virtualization has the promise of delivering the highest degree of value to our customers.

Cloud Computing: The progression continues, from virtualizing a machine/server to virtualizing a group or cluster of servers. Cloud Computing builds on Grid Computing, providing a management system that allows resources “in the cloud” (cluster) to be systematically assigned to users, while providing an appropriate environment for their workloads - e.g., JEE, LAMP. Now, clouds (or grids or clusters – pick your favorite name) need not be built atop virtualization technology like VMWare, Xen, or PowerVM technology. However, if I were building a cloud, I would use these technologies.

The cloud is really about the management and dispensing of the machines and the software that runs on these machines. A modern compute cloud allows a consumer to use a web application (or some programmatic means) to request compute capability containing pre-installed software (operating system and middleware). Amazon EC2 and the NC State’s Virtual Computing Lab are good examples of public compute clouds. The cloud can also do more than just dispense single atomic images. The molecular thought also applies to clusters. You can create a virtual cluster, consisting of several linked virtual servers, and then treat the entire molecule as an entity to be managed (created, moved, replicated, etc.). Again, this is where the value really starts to be evident.

One problem that I’ve heard some of our enterprise customers express is a concern with running enterprise applications on a public cloud. The concern is centered around exposing mission critical applications and data outside their corporate firewalls. That’s a legitimate concern. So, we’ll introduce private or enterprise clouds, which are clouds that run behind the corporate firewall.

SaaS: Software as a Service sits atop this stack and leads with a business thought. You don't “buy” software any more; you just pay for the service. Along with the business thought, SaaS environments attempt to virtualize the application. In a SaaS environment, applications and application-environments are not written or deployed from scratch, they are customized. Now many of us (in IBM Software Group) struggle with the term SaaS. You see, it’s really not about “software” as-a-service. It’s about “useful function” as-a-service. For example, the service platform makes customer relationship function available as a hosted service. In a classic SaaS environment, you are given a purposed solution that is fairly complete “out of the box”. The platform is extendable using configuration and plug-ins, hence, value is not far off because you are starting with something that can stand on its own (after you populate it with data and tweak a few .CSS style sheets). I’ve extended the thought of “function-as-a-service” to application-environments as well. In this form of SaaS, the environment (e.g., think of it as a purposed virtual server molecule containing a virtual app server, database server, LDAP server, etc.), is pre-provisioned and configured. All that is left for the user to do is install their application and express (in policy) how they want their application to behave in the virtual environment.

Now, you don’t need virtualization or cloud technologies to create a SaaS environment; however, if I were creating one, I would use them.

Our Strategy: Become Rainmakers

I see IBM (Systems and Technology Group, Tivoli and WebSphere) getting in the business of what I like to call rainmaker technology. Rainmaker technology is software and hardware that work together to help enterprises create clouds ☺. There are many technical facets to being a rainmaker, and we have teams tackling many different aspects of the technology. IBM Blue Cloud, and Tivoli DCM are all examples of IBM technology that enable clouds.

Ruth Willenborg is our lead chemist and is mastering the art of atomic and molecular virtualization. She and her team have made great progress in virtualizing parts of the WebSphere product line, including WebSphere Application Server and collections of our SOA infrastructure (e.g., WebSphere Process Server and WebSphere Portal). It will not be long until we deliver our first product offerings as virtual images.

Brian Martin and our WebSphere XD team is actively working on enabling and managing enterprise clouds, by controlling middleware running on virtual machines as well as real machines. Even a well-virtualized machine isn't exactly like physical hardware, and provisioning and load-balancing software needs to understand the difference. We're also working with our hardware team on finding ways to provision large numbers of application servers on racks of purposed Power–based systems.

Dennis Quan and our Tivoli team are working on a project called Blue Cloud, which is a general-purpose management system that includes cluster management software, virtualization tools Xen and PowerVM, and the open-source Hadoop parallel workload-scheduling software. The suite uses Tivoli software for automatically provisioning hardware to adjust for fluctuating computing demand.

Andy Rindos, from our team is working with our university partners on the Virtual Computing Initiative. This allows universities and their partners to have access to pre-loaded images at any time, in any place. In environments like this, where demand for particular machines is bursty but predictable - having these virtual labs allows for tremendous savings, which allows the function to be extended to places it never was before.

Virtualization is partly about increasing agility and improving time to value. A natural way for us in IBM to deliver that function is with appliances. We are experimenting with ideas around “rainmaking appliances” at several (all?) levels of the virtualization progression. These appliances will give customers options to quickly and easily create clouds and SaaS environments behind their firewalls – where they can better control their applications and data.

In all of these ways we are working with our customers to enable them to make their own clouds – we’re rainmakers after all.

Found a business needle in a haystack of events

Business needle in a haystack of events

Our business event processing (BEP) story is really starting to come together. In past blogs, I've said that BEP is an abstraction atop Complex Event Processing (CEP). The BEP abstraction manifests itself, first and foremost, as a set of visual tools. These tools allow the (non-programmer) business user to "sift" through a potentially massive collection of events and find that important business "needle" in that "haystack" of events.

Jeff Garratt, on my team, recently ran a set of lab experiments to illustrate the power of our BEP capabilities.

THE SCENARIO - For this exercise, we used a options pricing application written by Billy Newport to price options in real-time and a set of business rules to identify volatile options, giving brokers an opportunity to react early to these changing conditions.  This application processes events within two zones - the Raw Event (options pricing) and Actionable Event (identity volatile options) zones.

• The Raw Event zone is in charge of partitioning stocks across the cloud (i.e., processes disbursed across a set of IBM Blade Centers). Each stock has some number of options associated with it. The partitions are running a routine that updates the stock price and triggers "repricing"of the associated options using a Black-Sholes algorithm. If a options prices changes beyond a certain threshold, a business event is triggered to the BEP zone.
• The Actionable Event zone applies business correlation rules to these repricing events", checking the volatility of the options. This involves looking for "bursts" of repricing events for a particular option, within a fixed time period (say 1 second). If the fluctuation of the options prices is greater than a given amount (say .49) than the option is considered volatile. When a volatile option is found, another business event is triggered (perhaps destined for the iPhone of the Broker).

THE ENVIRONMENT - For this exercise we use the following "technology ingredients":

• WebSphere eXtreme Scale (aka ObjectGrid), plays the role of the event haystack (or cloud as I like also like to call it), handling the lower level events in the Raw Event zone.
• WebSphere Business Events - WBE (with AptSoft technology), plays the role of the business event correlator, handling the higher level events in the Actionable zone.
• WBE (with AptSoft), also plays the role of the tool used by the business user to express a "volatile option".
• IBM Blade Center, is the hardware we used for this exercise. We used 2 Blade Centers with 18 HS21s, which gives us 72 core to run the BEP/CEP zones on.

As I said, our BEP solution is anchored in visual tooling. For this example, the BEP rules for option volatility would look like this:

THE RESULTS - We ran the experiment with two basic styles of behavior; non-replicated and replicated. These behaviors, dictate the level of resiliency in the application. Non-replicated, means that if a failure occurs, only the stock partition in question fails. Replicated means that if a failure occurs, the stock partition fails over to another partition in the cluster. Hence, Replication trades off performance for extreme availability. We also report the throughput of the experiments based on zones; Raw and Actionable. The numbers are reported per processor CORE as well as for the entire set of 18 blades (72 core). The most compelling part of these experiments is that the application scales linearly. As you add blades, it scales accordingly. Since we had 18 blades in the lab at the time of the experiment, we used what we had. If we had more, we would get bigger numbers.

Here is a table containing the results:

SUMMARY - This results show that we can process millions of events per second in a scalable event cloud, while continually analyzing the events in the cloud for business activity of interest. We also demonstrated how we can use visual business tools to define rules and business events without programming. This really shows the power of business event processing in a robust and agile way.

Perhaps there is an opportunity to start to standardize this scenario into a benchmark - SpecEventProcessing2008 anyone ?

Beth Hutchison and our new AptSoft colleagues (Steve Lyons and Dave Martin) are pulling together a very compelling product story around BEP, integrating AptSoft technology into the WebSphere portfolio. This will be announced at IMPACT 2008. So, stay tuned.

UPDATE:  At IMPACT 2008, we officially announced the WebSphere Business Events product!  Read more here.  What are you waiting for?  Find that business needle!

Jerry

Zero Sugar is Splenda

FYI: I consume this many packs of Splenda a week!

I'm a huge coffee fan.  I drink about 3 cups a day.    I take my coffee with a little bit of half and half and 2 packets of Splenda.   That's 6 packets of Splenda a day.  So,  I wonder if this stuff is really safe?    I'm now hearing that  evidence is accumulating which suggests the chemical sweetener sucralose (Splenda) may be harmful (and may cause organ damage).  Perhaps I should just stick to sugar.

Speaking about "zero sugar",  Rob Nicholson just posted a very impressive entry on the Project Zero blog, describing how to run SugarCRM Community Edition on Project Zero.    Rob and team have continued to evolve Project Zero's PHP support and running SugarCRM on Zero is a milestone worth noting. (In fact, Rob will be demo'ing it at IMPACT 2008 next week).

I have blogged about our PHP strategy in the past.   I've talked about how running classic PHP application atop Project Zero was one of our goals (like PHPBB).    SugarCRM is arguably the most important,  PHP-based, business application available today.    Sugar is a very popular, full function CRM application that has a vibrant community of users behind it.  With Sugar now running on the Zero platform, things can really get interesting.   Sugar extensions can be written on Zero, using our PHP support.  Extensions should also be possible in Groovy and Java (given Zero/PHP runs on a JVM) and developers can even use the new Project Zero Web IDE to do so.

As Project Zero continues to evolve, applications like Sugar will be able to utilize aspects of the Zero run-time environment, like features I described as "the New Reality Run-time".   Similarly,  the Zero management environment will be able to support scaling and clustering of classic PHP applications like Sugar.

I think this is just the tip of the Sugar cube, so to speak.    So, maybe I can get my sugar fix after all, without using any of those potentially harmful substitutes.

Jerry

The Classics Keep Getting Better

I found a real gem on YouTube. If you’re a classic rock and blues fan, this one “takes the cake”. It’s a clip from a show called the Rock and Roll Circus – Featuring… a band called “The Dirty Mac”. The band includes John Lennon on rhythm and vocals, Eric Clapton on lead, Keith Richards on bass (didn’t know Keith played bass or was ever young) and Mitch Mitchell (from Jimmy Hendrix’s band) on drums. It just doesn’t get better than this…

Or does it? I was talking to Ian Mitchell, one of our lead CICS architects about what the CICS team has been up to. (No, Ian is not related to Mitch Mitchell from the Hendrix band - but he is equally talented). Let me tell you a little about why I'm so excited.

I’ve shared my passion regarding the subject of RESTful SOA. The CICS team is taking this pretty seriously. They are doing a number of things related to REST and ATOM enabling CICS. For starters, the CICS team is providing RESTful remote access to CICS system management capabilities using ATOM. It gets more interesting when they add REST-enablement to CICS managed data. With the REST support in CICS, a developer can use both ATOM Syndication and ATOM Publishing Protocol as means to view (primarily) and update CICS application data. We are hoping to release these capabilities soon via a support pack. There is a great opportunity once all CICS resources have URLs and an XML representation. For example, the synergy with a tagging service is going to be very powerful.

CICS is also participating in complex event processing (CEP). We are developing a capability to "feed" our CEP engine with events from CICS applications. The objective here is to enable customers to get more information out of existing CICS apps with minimum disruption.

The classics keep getting better, for sure.

Jerry

Go Wolfpack!

The NC State Wolfpack playing at Reynolds Coliseum

I vividly remember my first collage basketball game. It was an NC State home game, back when they played at Reynolds Coliseum. To this day I have yet to experienced anything like it again. The auditorium was erupting with pure energy. The volume, of the predominantly sstudent crowd, could have easily out powered the loudest rock concert. I remember thinking how much more dynamic "student power" was compared to watching professionals play.

Well, this week I re-experienced some of that magic, in the form of the positive buzz swirling around Project Zero. The news wires picked up on NC State using Project Zero as part of a computer science course.

One of the real perks of working in RTP is the interaction with our universities. Timo Salo, from the Project Zero team, worked with various NCSU students and professors including Dr. Munindar P. Singh to make this course possible. It's always refreshing to test new software with universities - the students and profs are never short of enthusiastic insights that ultimately make our software better.

This is addictive behavior - and we have similar projects on various topics going on at Duke U and UNC. Our university collaboration goes beyond RTP. In fact, we have many university hotspots all over the world, which we call. Centers for Advanced Studies (CAS). For example, On of the most vibrant university programs is run from our Toronto CAS. I'd like to blog more about these activities as the year goes on.

I guess James Governor, of Redmonk, got it right when he said in a recent blog entry - “IBM doesn’t want to be your father’s IBM, it wants to be your son’s”.

Go Wolf pack!

Fishing, REST and my recent video interview with INFOQ

Jerry at REST, Fishing - 2007 (saturated photo by GCC 2007)

I was talking to a group of people the other day about REST, when someone joked - "Jerry really knows little about the subject - he is always on the go." This is simply not true - I know how to relax.

In fact, I love to surf cast. It's when I do some of my best thinking. I really don't want to catch fish. In fact, when I do, I find it to be a real bother. You know- My son, daugther and relatives getting all upset- call me a "fish murderer", etc. But, let's be real here... It's not like the fish are jumping out of the ocean, into my hands. In fact, I think I've caught just one fish in the past five years. And for the record, I threw it back... Actually, now that I think of it- fishing is kind of stressful, not RESTful.

Speaking about REST... Over the Holidays, I had Floyd Marinescu from InfoQ drop in with video equipment in tow. Floyd does some really nice work, and asked me some decent questions regarding Project Zero and Restful SOA. It was kind of weird sitting in front of the camera with a bright light shining in my face… but after a while I forgot the camera was there and started doing my thing. The real trick was for Floyd to get me to shut up (The interview is 32 mins long :-)

The article is titled Interview: IBM CTO Jerry Cuomo on REST & Project Zero.

Enjoy... Jerry

WebSphere Technology Trends for 2008

 

Is that Agile Jerry doing a jumping back kick? (by SKC 2005)

Each year I outline a handful of technology treads that the WebSphere team is aggressively pursuing.  The following article provides a brief outline of seven technology trends that are energizing the WebSphere team in 2008. 

The 2008 trends are these: Restful SOA, Event Processing, SOA Management, Extreme Transaction Processing, Application Virtualization, Business Machines, and Business Rules

There are a few overarching thoughts to these trends – not the least to mention on continued maniacal focus on SOA.    However, as I outline these trends a single thought is in the forefront of my mind – that is Agility. 

• Agility in unleashing the content in the enterprise to the web. 
  
• Agility in creating new composite and reactive applications. 
  
• Agility in allowing a business analyst to dynamically customize a SOA 
  
• Agility in managing a complex data center in a cost efficient and green manner.

So, here goes...

[+] Click to read the more...