When I first started as an industry analyst in the 1980s IBM software was in dire straits. It was the era where IBM was making the transition from the mainframe to a new generation of distributed computing. It didn’t go really well. Even with thousands of smart developers working their hearts out the first three foresees into a new generation of software were an abysmal failure. IBM’s new architectural framework called SAA(Systems Application Architecture) didn’t work; neither did the first application built on top of that called OfficeVision. It’s first development framework called Application Development Cycle (AD/Cycle) also ended up on the cutting room floor. Now fast forward 20 years and a lot has changed for IBM and its software strategy. While it is easy to sit back and laugh at these failures, it was also a signal to the market that things were changing faster than anyone could have expected. In the 1980s, the world looked very different — programming was procedural, architectures were rigid, and there were no standards except in basic networking.
My perspective on business is that embracing failure and learning from them is the only way to really have success for the future. Plenty of companies that I have worked with over my decades in the industry have made incredible mistakes in trying to lead the world. Most of them make those mistakes and keep making them until they crawl into a hole and die quietly. The companies I admire of the ones that make the mistakes, learn from them and keep pushing. I’d put both IBM, Microsoft, and Oracle in that space.
But I promised that this piece would be about IBM. I won’t bore you with more IBM history. Let’s just say that over the next 20 years IBM did not give up on distributed computing. So, where is IBM Software today? Since it isn’t time to write the book yet, I will tease you with the five most important observations that I have on where IBM is in its software journey:
1. Common components. If you look under the covers of the technology that is embedded in everything from Tivoli to Information Management and software development you will see common software components. There is one database engine; there is a single development framework, and a single analytics backbone. There are common interfaces between elements across a very big software portfolio. So, any management capabilities needed to manage an analytics engine will use Tivoli components, etc.
2. Analytics rules. No matter what you are doing, being able to analyze the information inside a management environment or a packaged application can make the difference between success and failure. IBM has pushed information management to the top of stack across its software portfolio. Since we are seeing increasing levels of automation in everything from cars to factory floors to healthcare equipment, collecting and analyzing this data is becoming the norm. This is where Information Management and Service Management come together.
3. Solutions don’t have to be packaged software. More than 10 years ago IBM made the decision that it would not be in the packaged software business. Even as SAP and Oracle continued to build their empires, IBM took a different path. IBM (like HP) is building solution frameworks that over time incorporate more and more best practices and software patterns. These frameworks are intended to work in partnership with packaged software. What’s the difference? Treat the packages like ERP as the underlying commodity engine and focus on the business value add.
4. Going cloud. Over the past few years, IBM has been making a major investment in cloud computing and has begun to release some public cloud offerings for software testing and development as a starting point. IBM is investing a lot in security and overall cloud management. It’s Cloud Burst appliance and packaged offerings are intended to be the opening salvo. In addition, and probably even more important are the private clouds that IBM is building for its largest customers. Ironically, the growing importance of the cloud may actually be the salvation of the Lotus brand.
5. The appliance lives. Even as we look towards the cloud to wean us off of hardware, IBM is putting big bets on hardware appliances. It is actually a good strategy. Packaging all the piece parts onto an appliance that can be remotely upgraded and managed is a good sales strategy for companies cutting back on staff but still requiring capabilities.
There is a lot more that is important about this stage in IBM’s evolution as a company. If I had to sum up what I took away from this annual analyst software event is that IBM is focused at winning the hearts, minds, and dollars of the business leader looking for ways to innovate. That’s what Smarter Planet is about. Will IBM be able to juggle its place as a software leader with its push into business leadership? It is a complicated task that will take years to accomplish and even longer to assess its success.
As I was pointing out yesterday, there are many unintended consequences from any emerging technology platform — the cloud will be no exception. So, here are my next three picks for unintended consequences from the evolution of cloud computing:
4. The cloud will disrupt traditional computing sales models. I think that Larry Ellison is right to rant about Cloud Computing. He is clearly aware that if cloud computing becomes the preferred way for customers to purchase software the traditional model of paying maintenance on applications will change dramatically. Clearly, vendors can simply roll in the maintenance stream into the per user per month pricing. However, as I pointed out in Part I, prices will inevitably go down as competition for customers expands. There there will come a time when the vast sums of money collected to maintain software versions will seem a bit old fashioned. In fact, that will be one of the most important unintended consequences and will have a very disruptive effect on the economic models of computing. It has the potential to change the power dynamics of the entire hardware and software industries.The winners will be the customers and smart vendors who figure out how to make money without direct maintenance revenue. Like every other unintended consequence there will be new models emerging that will emerge that will make some really cleaver vendors very successful. But don’t ask me what they are. It is just too early to know.
5. The market for managing cloud services will boom. While service management vendors do pretty well today managing data center based systems, the cloud environment will make these vendors king of the hill. Think about it like this. You are a company that is moving to the cloud. You have seven different software as a service offerings from seven different vendors. You also have a small private cloud that you use to provision critical customer data. You also use a public cloud for some large scale testing. In addition, any new software development is done with a public cloud and then moved into the private cloud when it is completed. Existing workloads like ERP systems and legacy systems of record remain in the data center. All of these components put together are the enterprise computing environment. So, what is the service level of this composite environment? How do you ensure that you are compliant across these environment? Can you ensure security and performance standards? A new generation of products and maybe a new generation of vendors will rake in a lot of cash solving this one.
6. What will processes look like in the cloud. Like data, processes will have to be decoupled from the applications that they are an integral part of the applications of record. Now I don’t expect that we will rip processes out of every system of record. In fact, static systems such as ERP, HR, etc. will have tightly integrated processes. However, the dynamic processes that need to change as the business changes will have to be designed without these constraints. They will become trusted processes — sort of like business services that are codified but can be reconfigured when the business model changes. This will probably happen anyway with the emergence of Service Oriented Architectures. However, with the flexibility of cloud environment, this trend will accelerate. The need to have independent process and process models may have the potential of creating a brand new market.
I am happy to add more unintended consequences to my top six. Send me your comments and we can start a part III reflecting your ideas.
What a difference a year makes. The past year was filled with a lot of interesting innovations and market shifts. For example, Software as a Service went from being something for small companies or departments within large ones to a mainstream option. Real customers are beginning to solve real business problems with service oriented architecture. The latest hype is around Cloud Computing – afterall, the software industry seems to need hype to survive. As we look forward into 2009, it is going to be a very different and difficult year but one that will be full of some surprising twists and turns. Here are my top predictions for the coming year.
One. Software as a Service (SaaS) goes mainstream. It isn’t just for small companies anymore. While this has been happening slowly and steadily, it is rapidly becoming mainstream because with the dramatic cuts in capital budgets companies are going to fulfill their needs with SaaS. While companies like SalesForce.com have been the successful pioneers, the big guys (like IBM, Oracle, Microsoft, and HP) are going to make a major push for dominance and strong partner ecosystems.
Two. Tough economic times favor the big and stable technology companies. Yes, these companies will trim expenses and cut back like everyone else. However, customers will be less willing to bet the farm on emerging startups with cool technology. The only way emerging companies will survive is to do what I call “follow the pain”. In other words, come up with compelling technology that solves really tough problems that others can’t do. They need to fill the white space that the big vendors have not filled yet. The best option for emerging companies is to use this time when people will be hiding under their beds to get aggressive and show value to customers and prospects. It is best to shout when everyone else is quiet. You will be heard!
Three. The Service Oriented Architecture market enters the post hype phase. This is actually good news. We have had in-depth discussions with almost 30 companies for the second edition of SOA for Dummies (coming out December 19th). They are all finding business benefit from the transition. They are all view SOA as a journey – not a project. So, there will be less noise in the market but more good work getting done.
Four. Service Management gets hot. This has long been an important area whether companies were looking at automating data centers or managing process tied to business metrics. So, what is different? Companies are starting to seriously plan a service management strategy tied both to customer experience and satisfaction. They are tying this objective to their physical assets, their IT environment, and their business process across the company. There will be vendor consolidation and a lot of innovation in this area.
Five. The desktop takes a beating in a tough economy. When times get tough companies look for ways to cut back and I expect that the desktop will be an area where companies will delay replacement of existing PCs. They will make do with what they have or they will expand their virtualization implementation.
Six. The Cloud grows more serious. Cloud computing has actually been around since early time sharing days if we are to be honest with each other. However, there is a difference is the emerging technologies like multi-tenancy that make this approach to shared resources different. Just as companies are moving to SaaS because of economic reasons, companies will move to Clouds with the same goal – decreasing capital expenditures. Companies will start to have to gain an understanding of the impact of trusting a third party provider. Performance, scalability, predictability, and security are not guaranteed just because some company offers a cloud. Service management of the cloud will become a key success factors. And there will be plenty of problems to go around next year.
Seven. There will be tech companies that fail in 2009. Not all companies will make it through this financial crisis. Even large companies with cash will be potentially on the failure list. I predict that Sun Microsystems, for example, will fail to remain intact. I expect that company will be broken apart. It could be that the hardware assets could be sold to its partner Fujitsu while pieces of software could be sold off as well. It is hard to see how a company without a well-crafted software strategy and execution model can remain financially viable. Similarly, companies without a focus on the consumer market will have a tough time in the coming year.
Eight. Open Source will soar in this tight market. Open Source companies are in a good position in this type of market—with a caveat. There is a danger for customers to simply adopt an open source solution unless there is a strong commercial support structure behind it. Companies that offer commercial open source will emerge as strong players.
Nine. Software goes vertical. I am not talking about packaged software. I anticipate that more and more companies will begin to package everything based on a solutions focus. Even middleware, data management, security, and process management will be packaged so that customers will spend less time building and more time configuring. This will have an impact in the next decade on the way systems integrators will make (or not make) money.
Ten. Appliances become a software platform of choice for customers. Hardware appliances have been around for a number of years and are growing in acceptance and capability. This trend will accelerate in the coming year. The most common solutions used with appliances include security, storage, and data warehousing. The appliance platform will expand dramatically this coming year. More software solutions will be sold with prepackaged solutions to make the acceptance rate for complex enterprise software easier.
Eleven. Companies will spend money on anticipation management. Companies must be able to use their information resources to understand where things are going. Being able to anticipate trends and customer needs is critical. Therefore, one of the bright spots this coming year will be the need to spend money getting a handle on data. Companies will need to understand not just what happened last year but where they should invest for the future. They cannot do this without understanding their data.
The bottom line is that 2009 will be a complicated year for software. There will be many companies without a compelling solution to customer pain will and should fail. The market favors safe companies. As in any down market, some companies will focus on avoiding any risk and waiting. The smart companies – both providers and users of software will take advantage of the rough market to plan for innovation and success when things improve – and they always do.
I have been following Salesforce.com since its founding in the mid-1990s. Initially the company started by creating a contact management system which evolved into the sales force platform it offers today. Last month I attended a small dinner meeting in Boston hosted by Marc Benioff, Chairman and CEO of SalesForce.com, for some partners and customers. I met the Steve Pugh, CEO of CODA Financials, a subsidiary of Coda, a UK based developer of accounting software. I was intrigued that the company had built its new generation financial application on top of Salesforce.com’s infrastructure. In my next post, I’ll talk about Coda and why they made this decision. But before that I wanted to take a look at the Salesforce platform.
What is most interesting about Salesforce is that it intended to build a platform from day one. In my discussions with Marc in the early days he focused not specifically on the benefits of CRM but rather on “No Software”. If you think about it that was a radical concept ten years ago.
Therefore, It goes without saying that Salesforce has been a Software as a Service pioneer. For example, in June 2003 launched sforce, one of the first web services based SaaS platforms. It offered partners a published SOAP-based API. Rather than viewing Salesforce as an application, it views it as a “database in the sky.” It interprets this database as an integration platform. Likewise, from a customer perspective, Salesforce has designed its environment to “look like a block”. What does that mean? I would probably use a different term maybe a infrastructure blackbox.
Salesforce’s approach to creating its ecosystem has been incremental. It began, for example, by allowing customers to change tabs and create their own database objects. Next, the company added what it called the AppExchange which added published APIs so that third party software providers could integrate their applications into the Salesforce platform. Most of the applications on AppExchange are more like utilities than full fledged packaged applications. Many of the packages sold through the AppExchange are “tracking applications” for example, there is an application that tracks information about commercial and residential properties; another application is designed to optimize the sales process for media/advertising companies; still another package is intended to help analyze sales data.
But this is just the beginning of what Salesforce has planned. The company is bringing in expertise from traditional infrastructure companies like Oracle and BEA — among others. It’s head of engineering came from eBay. Bringing in experienced management that understands enterprise scalability will be important — especially because of Salesforce’s vast ambitions. I have been reading blogs by various Salesforce.com followers and critics. Josh Greenbaum, whom I have known for more than 20 years has been quite critical of Salesforce and has predicted its demise (within 18 months). He makes the comparison between Salesforce.com and Siebel. While any company that has risen as fast as Salesforce.com has will be a target, I do not believe that Salesforce.com is in trouble. There are two reasons I believe that they have a good chance for sustainability: their underlying SOA architecture and the indications that ISVs are beginning to see the company as a viable infrastructure.
So, what is the path that Salesforce is following on its quest for infrastructuredom (is that a real word — probably not). One of the primary reasons for my optimism is that Salesforce.com has a combination of traditional development through a procedural language it calls Apex that is intended to help developers write stored procedures or SQL statements. While this may disappoint some, it is a pragmatic move. But more important than Apex is the development of a standard XML based stylesheet interfaces to a service designed for use with Salesforce applications. This allows a developer to change the way the application looks. It is, in essence, the interface as a service. A third capability that I like is the technique that Salesforce has designed for creating common objects. In essence, this is a basic packaging that allows a third party to create their own version of Salesforce for its customers. For example, this has enabled Accenture to create a version of Salesforce for its customers in the health care.
But what is behind the curtain of Salesforce? First, Salesforce uses the Oracle database as a technique for serving up file pages (not as a relational database). But the core Intellectual Property that sits on top of Oracle is a metadata architecture. It is designed as a multi-tenancy service. Salesforce considers this metadata stack as the core of its differentiation in the market. The metadata layer is complex and includes an application server called Resin. The Resin Application Server is a high-performance XML application server for use with JSPs, servlets, JavaBeans, XML, and a host of other technologies. On top of this metadata layers is an authorization server. The metadata layer is structured so that each organization has a unique access to the stack. Therefore, two companies could be physically connected to the same server but there would be no way for them to access each other’s data. The metadata layer will only point to the data that is specific to a user. The environment is designed so that each organization (i.e., customer) has a specific WSDL-based API. In fact, the architecture includes the approach of access APIs through the WSDL interface. There are two versions of WSDL — one general and one for a specific customer implementation. If a customer wants to share data, for example, they have to go through the general WSDL interface.
Salesforce’s approach is to use XML based interfaces as an integration approach. It has used this to integrate with Google Apps. Salesforce has already begun partnering with Google around Adwords. This move simply deepened the relationship since both companies are faced with competitive threats.
The bottom line is that I think that Salesforce.com is well positioned in the market. It has an underlying architecture that is well conceived based on a SOA approach. It has created an ecosystem of partners that leverage its APIs and rely on its network to build their businesses. Most importantly, SalesForce.com has created an application that is approachable to mortals (as opposed to software gods). Companies like Siebel, in contract, created a platform that was complicated for customers to use — and therefore many purchased the software and never used it.
Salesforce.com is not without challenges. It needs to continue to innovate on its platform so that it does not get caught off guard by large (Microsoft, SAP, and Oracle) players who aren’t happy with an upstart in a market they feel entitled to own. They are also at risk from upstarts like Zoho and open source CRM players like SugarCRM. If Salesforce.com can collect more packaged software vendors like Coda to build their next generation applications on top of Salesforce’s environment, they may be able to weather the inevitable threats.
Partners are getting more and more important to the major software players. IBM announced a very interesting relationship with Kana, a $60 million solution provider of multi-channel customer service software. This is indeed a growing area in the market. Kana sells its software to about 60% of the Fortune 100. The company started in 1996 and has managed to survive some rough times and come out strong.
While IBM, like other major industry players rely on their partner ecosystem as an important go to market strategy. Some partnerships work better than others. What I thought was particularly interesting about the Kana partnership is its depth. Kana has decided to embed IBM’s DB2, WebSphere (including the WebSphere Process Server) into its solutions. SOA is an important new direction for Kana and the two companies plan to do some joint development in this area. Relationships like this don’t just happen. More than half of Kana’s customers are also IBM customers. This is important because increasingly the customers that I am talking to are looking to buy solutions from one trusted provider rather than trying to get a bunch of individual vendors to work together.
IBM has had a strategy for more than a decade of partnering with packaged software providers rather than being in that business. On one level, this can be viewed as a risky strategy. One only has to look at the roles of Oracle and SAP in the market to wonder if these packaged offerings will swallow up the entire ISV partner ecosystem like a black hole. I guess that my conclusion is that it just isn’t that simple. Customers that I have spent time with look at software packages from a different vantage point than infrastructure software. Because Oracle or SAP provides an excellent package for supply chain management or accounting management does not necessarily mean that they are the right choice for middleware or SOA infrastructure.
IBM’s partner strategy with ISVs has evolved over the past several years. I see a change from the desire to have lots of partners who will enable their software to run one or more IBM software offerings to deeper more strategic relationships. The Kana relationship is an OEM relationship — not a simple membership in a partner program. In fact, IBM has more than 30 of these OEM partnerships with vendors including Fair Issacs, Cisco, Nortel, and PTC — to name a couple. I expect that OEM partners are going to became an important center focus of IBM’s partnering strategy in the coming year.
I was talking to a CIO the other day about the whole area of Service Oriented Architectures. It was one of those interesting probing discussions around key players, emerging technologies and the like. One of the interesting topics that came up was around packaged software. This CIO was confused about a major issue. What is the benefit and danger of implementing a package software offering that has all the industry best practices, business process, and middleware integrated together. What are the opportunities and risks of this approach? Likewise, what are the risks of buying piece parts and integrating them together?
This is an important question and one that I have obvious opinions about. I think that it can be dangerous for companies to buy a too well integrated SOA environment from a single vendor. While it might seem like the path of least resistance might be to just buy an entire software suite from a company like SAP or Oracle and be done with it. While this may seem like a pretty straight forward question it actually is much more complicated. On the plus side, a customer could get a head start by using a SOA model where everything is designed to work together. On the other hand, I would submit that this approach is antithetical to the reason companies are approaching SOA in the first place. Companies are moving to SOA in order to create a flexible, modular environment where it is easier to add or subtract components based on either a new business initiative or a new innovative technology. If the SOA platform is too well integrated, change becomes hard.
So, what did I suggest to my CIO friend? I told him that it is better to look at packaged software as components in an overall SOA strategy rather than the lynch pin of that strategy. It is better to begin with the overall business strategy and an Enterprise Architecture and select technologies that are designed with standardized interfaces. The foundation should be based on loose coupling of services.
A packaged offering can work if customers finds a package that is standards based and extensible does not lock them into one perspective on the world. I think we’d all like to have a world where you just buy what you need off the shelf and life is good. But unless you are buying a commodity, I think the world is still too complicated for packaged SOA. Are there SOA commodities? Of course, for example, a set of best practices that are well understood across a broad spectrum of customers can be packaged as a business service and used broadly. Even a large service such as those offered by ADP is an example of a service offering that is well understood and not differentiated. Who would want to write their own service for managing payroll.
I do think that there will be a time when the SOA software market has matured to the point where building blocks are mature and well structured enough to be able to link together services smoothly and easily. But I don’t think we are there yet…do you? Let me know what you think.