The hybrid utility architecture
May 31, 2006
A year ago, when I wrote my article The End of Corporate Computing, I made only a passing reference to the software-as-a-service movement as an example of one of the forms utility computing was taking. Despite a lot of startups in the space, and some growing success stories like Salesforce.com and RightNow, it seemed like the expansion of the SaaS model would proceed very slowly. Nervousness about entrusting sensitive data to outsiders appeared to a major sticking point for a lot of companies, one that would hinder their adoption, or even consideration, of SaaS alternatives.
But that was then, and this is now. Executives' nervousness is dissipating rapidly in the face of clearer evidence of the benefits that the SaaS model can provide - not only in freeing up capital and achieving cost savings but in enhancing flexibility, radically streamlining installation and upgrades, and even enhancing data integration through Web 2.0-style "mashups." The SaaS tipping point is fast approaching, if it hasn't already arrived. The latest evidence comes from the conservative consulting giant McKinsey & Company, which has just placed its imprimatur on software-as-a-service. In a new article for the McKinsey Quarterly, McKinsey IT consultants Kishore Kanakamedala, Vasantha Krishnakanthan, and Roger Roberts write about SaaS in glowing terms:
The software-as-a-service model can cut the total cost of deploying some classes of enterprise applications by 30 to 40 percent as compared with the total cost of purchasing and maintaining them in house. Of the senior IT executives we talked with, 38 percent said that they plan to use the software-as-a-service approach during the next 12 months ... Software as a service differs from the fad of the late 1990s for application service providers (ASPs) because the most successful companies offering this latest generation of hosted software have redesigned their applications for scalable delivery over the Web. In this way, these companies innovate more quickly and thus have lower total costs - and pass the benefits on to their customers. Contrary to some expectations, the acceptance of this model isn't limited to midsize companies with understaffed IT departments; some very large enterprises are among the earliest adopters.
The consultants also report that "IT executives are shifting to the software-as-a-service model for some applications not only for lower licensing and maintenance fees but also because implementation is usually quicker and companies don't have to maintain special skills in software-specific areas." They go on to conclude that the adoption of SaaS, combined with the rapidly growing use of hardware virtualization, suggests that "a technology architecture transformation" is under way in mid-sized and larger companies:
In the past, CIOs deployed their own self-contained application architectures on their own servers and storage systems. This old model is giving way to a hybrid application architecture that combines hosted functionality with in-house applications running on consolidated and virtualized commodity servers. We believe that this transformation will drive efficiencies across the full stack, from business processes to physical infrastructure, while increasing IT's ability to meet new demands in a rapidly changing business environment.
To put it another way, larger companies are reconfiguring their traditional IT assets as centralized internal utilities and then drawing in new or improved IT capabilities from outside utility suppliers such as SaaS firms. This hybrid utility architecture, as I would term it, enables much greater efficiency in running mature enterprise applications while also allowing companies to tap into the new generation of true Internet-based software. The hybrid model also provides a way for CIOs to defuse the tension between IT Departments focused on legacy applications and employees looking for new Web 2.0 capabilities - a tension that Peter Rip describes very well.
SaaS is looking more and more like the pivot between the IT of the past and the IT of the future.
Having just listening to the 5/31 Gillmor Gang episode, "Gartner Gang-Up Part I," I'd think that maybe you're the surprise guest that Steve references. Your post sounds just like the discussion they had about the relationship between SaaS for the more commoditized IT activities and the new mashup/scripting/AJAX work for in-house IT activities critical to competitive advantage.
Rather than referencing the McKinsey article, though, the guys gang up on Gartner as now crying fire in the theater after having milked/sold SaaS ambiguity over the past few years. Now Gartner feels we've reached the tipping point and they're trumpeting SaaS as the next panacea for software development.
As an aside, I always enjoy Gillmor Gang when you're on the call. Thanks.
Posted by: chrish at June 1, 2006 01:41 AM
I've read a fair bit about ASP, on demand, SaaS, (and now heaven forbid SaaS 2.0). I'm stumped.
I'm trying hard to understand what the differences are between what Salesforce.com does and what ADP has been doing since the late 1940's (well, 1957 when they moved to punchcard computing).
Other than alot of acronyms, I can't find any differentiating factors, except that ADP do payroll, car dealer services and brokerage services and have had 167 or whatever quarters of continuous growth, are big, and are very very profitable. Also ADP seems to have a more quietly spoken, mild mannered CEO.
Please excuse me banging my SAP drum but you may have missed the deal announced yesterday at SAP Sapphire with IKEA 35 countries, 85 000 employees on the ADP-SAP globalview offering.
(I think there are now more users on the ADP-SAP global view HR-payroll platform than there are salesforce.com users.)
Imagine ADP start doing CRM.
Posted by: Thomas Otter at June 2, 2006 05:47 AM
That's the first time I've heard "SaaS 2.0." I pray it's the last. I agree with you that ADP is the original SaaS company. The difference between ADP and today's SaaS companies is evolutionary. ADP's service was essentially a batch-processing service (at least originally) that didn't have to handle a lot of constant queries and entries and messages. The internet just opens up a lot more applications to the SaaS model. But I think ADP continues to be a good "proof of concept."
I think it's interesting that today you hear a lot of nervousness about having an outside company handle sensitive corporate data, yet for decades companies have been happily letting ADP and other payroll processors outsiders have free access to some of the most sensitive business data of all: salaries and bonuses.
Posted by: Nick Carr at June 2, 2006 08:37 AM
Alas, Nick too late.
when is roughtype 2.0 coming out !-)
ADP's model today has really moved on. It is tempting to think of them as a batch business, but that was decades ago. If I was a "new" SaaS player, ADP would be my nightmare competitor. Until recently I had no idea how big and smart they are. They just quietly get on with it.
Posted by: Thomas Otter at June 2, 2006 11:21 AM
I think, the trouble with a lot of discussion about utility computing, it is too large a scale for use to comprehend. The systems are vast, and too complex to grasp with our little brains. Remember the story about Jack Welsh first being showed how to use the Internet by his wife while on holiday. Up until that point, Jack has listened to the hype, but just didn't get it!
To understand the situation for CEOs with regard to grid computing, it might be useful to imagine a much smaller situation, like that of a boyfriend and girlfriend having coffee on their sunday morning together. Yesterday morning, I was drinking a cappuchino outside a cafe in the center of Dublin city. A couple were sitting at a table beside me. While the man had brought something to read, I noticed the women wanted to talk a lot.
The man asked his girlfriend, if she ever bought newspapers to read. She answered no, because she always read her news online. She asked her boyfriend to accompany her to a web cafe. The boyfriend refused. He said he would prefer to sit in the sunshine and read his book. She could go to the web cafe if she liked. After a few minutes more, the boyfriend got a brainwave. He rose from his seat and walked across the street. He returned a minute later carrying a newspaper he had bought from a shop. He offered the newspaper to his girlfriend.
She spent a second or two leafing through the pages, only to begin chattering again. Still unsatisfied. She persisted in begging her boyfriend to accompany her to the web cafe. No, she would not accept a compromise, whereby the boyfriend could sit outside the cafe reading his book, while she went off and browsed the internet. For this young woman, reading news on sunday morning needs two things. To obtain the news in her preferred bit format, while still having the company of her boyfriend.
For her boyfriend, it meant sitting in sunshine, outside a cafe. The current situation does not afford couples the opportunity to do both. The story of the couple and their sunday morning coffee, is the story about society in the twenty first century. Namely, the struggle for supremacy between bit-based and atom-based media. With all the abundance of technology, we still do not have the capability of consuming our bits and and our atoms in the one place, at the one time. Why is that, that news cannot be received in bit format and atom format at the same time?
If you want another historical reference for bits and atoms coming together, think about the following. What are most males attracted to? They are attracted to beer, food and football. Fast cars and loud music too. But somewhere at the top of a man's list, you will find beautiful women. What are beautiful women but a very fine collection of atoms?
I ask you to notice carefully the effect of a beautiful female on a man. The female can extract information from a man with insulting ease. A very strong relationship exists there between bits and atoms. Some females are so successful, their ability to extract information overtakes their ability to process the information. Hence, we get the dumb blonde jokes.
Intel have a strong demand for their atoms, just as women do. Intel makes one of the sexiest collections of atoms - the fast silicon microchip. The silicon chip is dressed by the fashion experts at Dell and Microsoft. It is delivered to your home for a modest fee. Games and applications that run on these PCs have proven so addictive, they cause men to ignore women!
Intel has run into the dumb blonde's problem too. Their success in extracting peoples' information, has far out-stretched their ability to process the information in some useful way. Microsoft's operating system has become a dumb blonde's brain.
Consider lastly, the story of Apple CEO Steve Jobs. His story is about mixing bits and atoms. Steve Jobs initially founded a company called Apple Computers. One which grew to become a 'big mover' of atoms and bits in the computer industry. Much later, Jobs found himself in charge of the largest mover of CG entertainment bits ever, Pixar Animation Studios. Some say, Jobs invented the personal computer industry all by himself. It is like saying Elvis invented Rock N' Roll. Who knows, it is open to much debate. Was it the sound of the music, or the shaking of those hips? Who one knows if Apple is a software or a hardware company.
Brian O' Hanlon.
I think you are right that enterprise IT is now ready for SaaS for certain classes of application, but not yet all. Some of the most interesting work in the application of 'read-write Web' ideas to the enterprise will require the exposure of existing croporate data as objects and feeds that can be re-combined by users. The problem is that SaaS /ASP applications currently find it hard to access this content and data ina secure way across the firewall. That, I think, will mean some of the most intersting enterprise social software will need to be internal.
I think a bigger question is how we can invent a new delivery mechanism that allows companies to benefit from the rapid pace of innovation whilst also having stable applications that don't get thrown out every few months.
Our view is that this will need to combine application integration, architecture work, SaaS and external online services for data, with a stable UI and a stable architecture providing the anchor for a fluid middle tier of apps and services.
We have asked the questions here: http://www.headshift.com/archives/002916.cfm (but we don't have all the answers yet!) This is based on our experience over the last few years integrating social software for large organisations,and it woudl be interesting to hear form other practitioners whether that also reflects their own experience.
One thing's for sure, and I agree with you, that IT is evolving towards an enabling utility service and away from its previous role as the technology Central Committee who must approve everything more sophisticated than a digital watch within the firm.
Posted by: Lee Bryant at June 7, 2006 05:30 AM
I believe we've already reached the tipping point. System integrators, hosting providers and SaaS providers tell us they're being asked for SaaS and hosted applications by customers of all sizes, and from all industries. This is no longer an SMB or high-tech market.
However, there are still hurdles to offering the services in the way customers want to buy them. Computerworld recently published a list of 10 SaaS "traps" and several items on that list are related to infrastructure. SLAs, access to storage, and even the ability to customize aren't supported easily by current architectures. Computerworld left off my favorite, and one of the hardest for a SaaS provider to support, the ability of the customer to choose when to upgrade.
As a result, when companies begin to gain traction, they immediately face a crises in scaling service reliably. Even bloggers don't seem to understand the situation, though, typically berating the down service for lack of vision or proper coding. The few companies that have grown to significant scale find themselves spending as much on infrastructure as the old telcos used to. For SaaS to mature, new infrastructure architectures need to be developed that are supportive of the business model.
I've written a couple times that SaaS and Web 2.0 providers will determine the future of the datacenter. I realize this is an uncomfortable statement for both suppliers and CIO's, but the needs of these businesses already exceed that of most traditional enterprises for uptime and scaling. Just consider that a 100 person SaaS supplier can have the IT footprint of a medium sized bank.
For the moment, few established vendors seem to recognize this shift or are willing to step up to the challenge, leaving the market for us startups.
Posted by: barmijo at June 12, 2006 04:33 PM
Post a comment
Thanks for signing in, . Now you can comment. (sign out)(If you haven't left a comment here before, you may need to be approved by the site owner before your comment will appear. Until then, it won't appear on the entry. Thanks for waiting.)
"Riveting" -San Francisco Chronicle
"Rewarding" -Financial Times
"Ominously prescient" -Kirkus Reviews
"Riveting stuff" -New York Post