Feb 25, 2011 (07:02 PM EST)
IT Is Too Darn Slow
Read the Original Article at InformationWeek
Really listen to CIOs these days, and you'll hear something in their voices: impatience.
It's not always overt. And it's more than the anxious ambition of most successful executives, who got where they are in part by pushing and pushing to get things done. This impatience runs deeper. They look at the speed with which IT delivers business value, and they have a nagging sense that it's not close to good enough.
This isn't about the speeds and feeds of technical performance, of gigahertz this and gigabit that. That's IT's comfort zone, and in that arena, CIOs, their organizations, and their vendors have exceeded every rational expectation. In fact, that exceptional technical performance has raised expectations that IT organizations themselves can and must execute faster.
This need for speed is about shortening the distance between great idea and end result. It's the time and effort and cost between 'Hey, what if we …" and employees using a new productivity tool or customers crowing about the latest tech-based innovation.
This need for speed comes to a head when the sales and marketing and product development and HR groups say they can be ready to go in two months--and IT says six. Too often, CIOs are reaching this painful conclusion: IT is just too darn slow.
Listen to Rob Carter at FedEx. A CIO couldn't work for a more enlightened company when it comes to understanding IT's central role in business success. Carter earned his "seat at the table" long ago. When huge strategic decisions get made at FedEx, Carter and his IT organization are there from the start.
Still, the reason Carter's driving a sweeping change to how FedEx runs its data centers and manages applications is in large part because, when the discussion turns to when a business initiative can be completed, all eyes turn to IT. It's nice to be seen as strategic, Carter says, but "it's also difficult to be the long pole in the tent every time we want to go do something."
Listen to Randy Mott, CIO of Hewlett-Packard. Mott has cut the time it takes to complete the typical HP IT project down to six months, but now he thinks he needs to slash it again, to three months. Instead of big-bang IT, he wants IT's impact to be constant. "We ought to be changing the productivity of everybody in our organization every 30 days--in HP, not IT," Mott says. "If we're going to keep up with the growth in front of us, we better figure out how to do that."
Listen to Denis Edwards, CIO of global staffing company Manpower. Edwards sees Amazon Web Services spinning up a server in minutes for a developer and sees Apple and its legions putting out new iPhone apps every single day, and he knows that's the standard by which IT will be judged. So he's making Manpower's enterprise software development and distribution model more like an app store, where country managers can have local IT teams grab software tools, such as a new video interviewing app that's in development, and implement them without it becoming a long-term project.
Listen to Michael Heim, CIO of Eli Lilly. The drugmaker is using Amazon cloud infrastructure to let scientists do early-stage, compute-intensive research. Nothing is more important to this business than discovering new drugs, yet in the past it could take six to eight weeks just to get IT services set up for those researchers. Now it takes days. "It's hard to underestimate the value of letting scientists work at their own pace," Heim says.
Listen to Chris Perretta, executive VP and CIO of financial services company State Street, which is building a private cloud environment in order to, yes, cut maintenance costs, but more important, to accelerate application development, the core of State Street's IT-led competitive advantage. "Our goal is to optimize the time from idea to solution," Perretta says. "The idea is when the stopwatch starts. Really, it's about the customer--when does their stopwatch start, not mine."
Listen to FBI CIO Chad Fulgham, who took control of the intelligence agency's biggest, most important IT project--a twice-delayed case management system called Sentinel--from contractor Lockheed Martin in order to pull it across the finish line. Fulgham, who learned about speed working on Wall Street, has FBI IT staffers employing agile development to break the $451 million project into smaller pieces and make adjustments on the fly. "The smaller team takes out a lot of the, for lack of a better word, bureaucracy," he says. "It is a fundamental change in how the federal government does IT. I want to be on the leading edge."
Why is this need for IT speed accelerating? First, technology and process advances, from cloud computing and virtualization to Web services and agile development, are making it possible for IT pros to deliver applications and infrastructure much faster.
Amazon, for instance, showed it's possible to deliver compute and storage infrastructure in minutes, not months, and Salesforce.com slashed the time it can take to launch CRM applications. Yes, there are caveats galore. Can Amazon meet your security requirements and enterprise-class service-level agreements? Can Salesforce integrate the way you want with other enterprise applications? But these cloud services reset the benchmark when it comes to getting critical IT resources up and running in next to no time. Inside companies' own data centers, virtualization has made it feasible for companies to mimic Amazon-like clouds. They're building private clouds that allow for faster provisioning of servers and more flexibility in moving workloads across servers and storage to meet changing demand.
Second is the juggernaut of consumer IT, particularly mobile devices. The pace of change in consumer tech--led by Apple's iPhone, app store model, and now its iPad--is raising the expectations of what people can do at work, and IT must keep up.
The third and most important driver is the changing nature of demand for enterprise IT. It's what Carter alluded to--business just can't move without IT anymore. IT is embedded in new products and new product development. Consider FedEx's sensor product that can relay an alert if a package is opened en route. Consider the video interview clips Manpower will send to clients along with the resumés of job candidates. Consider the digital mock-ups Procter & Gamble uses to show a new shampoo bottle to a focus group. If IT's too slow, business is too slow to innovate. It's that simple.
HP's Randy Mott wants 90-day IT projects. "If we're going to keep up with the growth, we better figure out how to do that."
A related factor is that IT is doing a lot fewer long-term process projects, like ERP, and a lot more iterative projects, like new tools for collaboration or customer service or customer-facing Web sites. The Corporate Executive Board has studied this phenomenon and found that these shorter-term projects, which it calls "information projects," now account for 45% of IT's new project budgets.
With these kinds of projects, business units don't know exactly what they want at the start. IT needs to move with the business units to get something in front of customers or employees, listen, and then iterate, either by fixing what's wrong or adding more features. Last holiday season, Guess CIO Michael Relich gave the clothing retailer's salespeople iPod Touches equipped with bar-code readers so they could tally customers' purchases on the store floor. As soon as the salespeople got those devices, they flooded Relich with requests to do new things with them: Let me sign customers up for the loyalty program. Let me check inventory at the mall on the other side of town.
In such a world, "velocity is more important than perfection," says Peter Hinssen, a consultant and author of The New Normal. More and more products are experienced through a digital filter, Hinssen argues. He recounts buying a new car, trying to plug an iPhone into it to play music, and being told his "accessory" wasn't compatible with the phone. This is the digital world IT must keep up with--one where Apple can view a $40,000 car as an accessory to the iPhone.
Speed over perfection can be a tough notion for IT to accept. But if IT is going to drive business results and maintain its strategic presence, it can't slow the business down. Warns Hinssen: "If you don't like change, you'll probably like irrelevance even less."
The Innovation Feedback Loop
Technology doesn't just meet demand; it creates demand. That's another reason IT can't live in long, orderly project cycles.
Guess' experience is a great example of what HP's Mott calls the "innovation feedback loop," and it's a big reason he's trying to slash the average time it takes to complete an IT project to three months. IT can't set priorities for 10 projects spread over the next two years because, once projects one, two, and three are done, that will change what would have been four, five, and six. Guess could have tried to anticipate every feature it needed on its handhelds--or it could get devices out with core functionality and add features as salespeople figured out what's most important. "There's an innovation feedback loop that right now, in these long development cycles, IT constrains," Mott says.
OK, so 90-day projects are a nice theory. How will Mott get there?
First, it's worth noting how HP got down to six-month IT project cycles, since many companies aren't even there. One step was to staff each project for faster results. That means having 10 people on a project for six months rather than five people for a year. Mott also measured the number of IT locations involved in any project and tried to keep that number low, under the theory that proximity helps people work faster.
In measuring on-time delivery, Mott weighted projects based on their value--so four little projects on time and one huge one late doesn't equal 80% on-time delivery. Also, he used what he calls "time-phase boxing," comparing projects of similar length to see if they're hitting milestones on time, in order to spot delays sooner. So if the average three-month project moves into testing phase at six weeks, and project XYZ still isn't testing by week seven, the time-phase boxing metric would be flashing red. HP will keep leaning on these practices.
To squeeze IT projects down to 90 days, Mott is putting a lot of faith in automating more of his organization's testing scripts, "because too much time is spent just making sure everything still works right." (HP sells testing automation, via its Mercury acquisition.) Mott also wants developers using fewer tools and languages, rather than letting them pick whichever ones they like best. Having more people trained on a smaller set of skills will let HP move people across projects more freely. "We'll have multiple skills," he says, "but we can't have 500 skills."
IT also needs to understand which elements of speed matter most to their companies. Consider Eli Lilly using Amazon's infrastructure service for drug research. The discovery of a single new drug--and getting that drug through the entire research, development, and regulatory pipeline sooner--can change the future of the company. "For us, it's pipeline, pipeline, pipeline," said Heim, explaining the company's cloud strategy at last September's InformationWeek 500 Conference. "Anything we can do to further our knowledge, get products into the pipeline, and develop those more quickly is crucial to us."
That's why Lilly saw big value in trying Amazon's infrastructure as a service to speed the earliest stages of drug discovery, when a researcher might want to quickly model and test an idea. (The fact that early-stage research often involves relatively nonsensitive data that's not much of a security risk also played a part in Lilly moving those resources to the cloud.)
Perhaps no industry's IT organizations are feeling--or should be feeling--this speed pressure more acutely than those in the U.S. healthcare sector. If healthcare providers are going to cash in on federal subsidies for deploying electronic health records, the clock is ticking--not to just get them implemented, but to make sure doctors, nurses, and other clinicians apply them in specific ways that meet federal "meaningful use" guidelines. Oh, and they'll also be looking for ways to make patients healthier and lower the costs of treating them in the process.
IT needs to judge itself on whether it's fast enough at the things that matter most to the company--whether that's integrating new suppliers needed to launch a product or providing the IT resources to design and spec a product. "Those are real-world provisioning. … It's how fast can you go from zero to a billion in a product," Mott says. "Those companies that make the transition and cut down that cycle time will become the leading players in the field."
Learning From Consumer Tech
To see the effect of consumer technology on corporate IT, there's no better place than Manpower's headquarters in Milwaukee on a recent January day. The lobby is full of suitcases and foreign accents as the company wraps up its global leadership meeting, where executives from around the world get the Manpower religion. Throughout the event, the IT organization sets up shop in a large conference room to show its latest ideas. From nearly every test project or prototype, this much is clear: This IT team is racing to keep up with consumer technology.
The expo is part of CIO Edwards' attempt to get more development happening faster, by tapping the company's smaller IT teams worldwide. In conjunction with the leadership meeting, IT teams from Bulgaria, France, Canada, Argentina, and elsewhere showcase new applications and prototypes they think all of Manpower could use. From these prototypes, execs will decide which to develop and test. Then those software tools will be put into Manpower's version of an app store, available to country managers to implement when there's a business need in their markets.
Inside the expo, there's a kiosk that looks like a very large ATM. It has a touch screen that a person can use to look for job openings, create a Manpower profile, or even access a LinkedIn account, and use that to apply for jobs. It's something of an interactive billboard for Manpower that could be placed in a mall or an airport, or possibly even replace some of the company's low-traffic branch offices. What it really is, though, is a giant iPhone app. Manpower had already developed an iPhone app to make it easy for people to apply for jobs. For that, the big effort was getting the company's Web application to respond to touch-screen commands. Using the smartphone app as a base, it built the ATM-sized prototype in about three months.
Next is a wall-mounted computer display showing Google Earth, integrated with Direct Talent, Manpower's back-end software that manages job candidates' applications. So while looking at a Google Earth map of Europe, a person can zoom in on each country and see data on how many new people have recently applied to Manpower in Italy. What's the point? That's what IT VP Mark Bistersky asked when IT director Ken Rheingans suggested they build it. "I said that I don't see the business value, but Ken said, 'Yeah, but it looks cool, and it won't cost any money,'" Bistersky recalls. They built it, and several salespeople loved it, saying they wanted it to bring some sizzle to sales presentations.
Manpower demos a prototype Facebook app that lets people friend a virtual recruiter. It's not a real person and doesn't pretend to be; it's an app that lets you ask questions using chat, and it responds by providing Manpower Web resources. The results it provides are really no different from the ones you get with Web search, but some people are just more comfortable using Facebook and the chat interface. "Fish where the fish are," Bistersky explains.
Manpower's Edwards is building an internal app store. "We can get things done faster and get more things done in parallel."
Manpower also shows off an "advertising wizard," essentially an analytics tool that assesses and scores the copy in job ads written by recruiters based on the effectiveness of past ads with similar wording. There's Interview Buddy, a smartphone app that a job candidate can check before heading to an interview. The app provides information about the company and offers simple tips, like asking if your shoes are shined. And there's the app we mentioned earlier, video interviewing, where a recruiter provides a series of questions, a candidate tapes responses, and those answers are shared with relevant hiring managers.
The big change here for Manpower is twofold. One is that these ideas are inspired by consumer technology. Whether it's building or using smartphone apps or Facebook searches or Google Earth links, the company is determined to keep its enterprise IT as relevant and powerful as the personal computing experiences of employees, partners, and customers. The other is that all this work is happening in parallel by distributed IT teams. Rather than wait for a central IT team to get around to each of these ideas, Manpower's letting the Canada IT team develop the video interview, for example. Once the app is tested in other markets and deemed good enough to implement worldwide and integrate with all of the company's systems, it will be offered to all country units via the app store.
IT needs speed, but it needs the right kind of speed. With Manpower's video interviewing, the demand is overwhelming--all country managers want it as soon as they see it. However, if each country unit went out and found its own video interviewing software and ran with it before it was tested and stabilized, Manpower would end up with a fragmented candidate pool, with some video in one format inaccessible to people in another.
Edwards manages to look both energized and exhausted, hugging a Diet Coke as the IT expo wraps up on Friday afternoon and delegates stream out of the lobby to head back to their home countries. He knows what he has done. He has revved up demand. And now his team needs to pick the right apps; deliver technology that's secure, works globally, and integrates with other company systems; and deliver it before the excitement withers away. If video interviewing is still in the prototype room next year, the feedback to Manpower's IT team will be, "Didn't you promise us this last year? When will we finally have it?"
100 MPH--The Wrong Way?
Once IT decides to focus on speed, two obstacles get in the way: security and governance.
Security's often the trump card IT plays. It's certainly true that security and regulatory compliance make business IT more complicated than consumer IT, but security can't be the overriding excuse for not moving faster.
CISOs must bring more of a business point of view to their security judgments, says James Dipasupil, a consultant and former CISO of financial services company Ameriprise and health insurer WellPoint. Having worked in financial services and healthcare, Dipasupil knows some risks are nonnegotiable. You'll never risk breaking a regulation or knowingly put a Web app out with a cross-site scripting error. But for some security problems, CISOs must weigh a delay against the risk and decide if the app can be rolled out and any problems resolved along the way. Again, it's velocity over perfection--and it's heresy to some security pros. "CISOs need to get comfortable with that," Dipasupil says.
CISOs also need to get comfortable automating more security testing, he says. (Dipasupil does some consulting work with a security automation firm, Veracode.) Skilled security staff will always be in too short supply to hand test every element. "Reserve these really good security people for the really difficult security problems," Dipasupil says. Another time saver: Have legal, compliance, and IT agree on key elements of a law or regulation. Too often, the interpretation of a law or regulation gets debated anew with every security problem.
For FBI's Fulgham, agile development is a "fundamental change in how the federal government does IT."
Governance covers a lot of ground: budgeting, priority setting, aligning IT and business sponsors. Done wrong, it's a lightning rod for criticizing IT for being too slow, and a major reason business units start their own rogue Salesforce and other projects without IT input. Project management offices generally don't do a good job of being flexible, says Shvetank Shah, who leads the Corporate Executive Board's IT practice. Too often, the PMO's methodology is built to handle complicated, long-term projects, and that puts too much overhead on small, quick projects. Having flexibility gets more important as companies do more iterative projects, those that have to factor in customer and employee feedback. "You want IT to be in constant test-and-learn mode," Shah says, "and IT isn't built for that."
But IT leaders who try to move faster without sound governance practices can end up going 100 miles an hour in the wrong direction, says George McKinnon, CIO of China-based IT outsourcer Bleum and former CIO of Nationwide Insurance and VP of IT at Expedia. To have credibility with business units, CIOs need good data on IT operations, he says--which projects are on time, or how speeding up one project will cause others to be slowed down. And IT must be transparent about quality trade-offs--does everyone agree that this app can go out as "good enough" and revised on the fly? Getting speed "without sufficient quality, you lose trust," McKinnon warns.
HP's Mott doesn't plan to give an inch on governance, even as he tries to slash IT project times. To force business leaders to prioritize IT projects, Mott demands that every one of them gets a cost-benefit analysis--no IT effort goes forward without a CBA. "If it's a one-week project, the CBA isn't that big," Mott says. "These things scale together." Yes, the first CBA employees do takes time, as they struggle to define the business outcomes they expect from a project. But with experienced people, a CBA on a small project can take 15 minutes. "It's not where most of your time gets spent," he says.
Architected For Speed
One major area that holds back IT speed is legacy IT infrastructure. FedEx's Carter, speaking at a CIO event last fall, told the audience not to spend time looking back at how they got where they are--everyone has some mix of legacy software, hardware, and architecture. In fact, any CIO who doesn't have this dilemma is either way beyond good or has been keeping too tight a rein on IT to let the business grow, he says. The bottom line: CIOs must look ahead and decide if their companies' IT architecture can keep up with the pace of business.
Carter didn't feel he had that. But before he tried to change it, he wanted his fellow business leaders to agree. So his team did some research, identifying the thousands of applications, servers, and custom interfaces that stitched together the many businesses FedEx had acquired and built over the years. Carter made the business case that FedEx couldn't do all the things it wanted to do--provide a unified customer experience, launch new products rapidly, make and quickly integrate acquisitions--without a more unified IT infrastructure.
FedEx is overhauling its infrastructure using a services-based approach. The apps will remain but will call on common data sources for 24 core transportation-related services, tasks like providing an address. The apps are also being rewritten to use common infrastructure such as databases and messaging, so they can move across servers. FedEx built two data centers based on highly virtualized, x86 environments--a private cloud for what Carter calls "general purpose" computing, letting application workloads be shifted more efficiently to any part of the data center. No apps get into those new data centers until they run on this modern services architecture.
FedEx long ago made IT a strategic asset. Now CIO Carter is transforming its IT infrastructure for greater speed and flexibility with an architectgure for "general purpose" computing.
Manpower's Edwards also revamped the company's architecture for speed. Edwards decided not to consolidate the company onto a single ERP platform near term. That remains a long-term goal; today, high-impact apps such as video interviewing and tablet-based apps for recruiters are the more urgent priority. However, Edwards did need to establish a company-wide integration architecture for the app store concept to work, so a new application could be plugged into any number of legacy Manpower systems used worldwide. In the past, Manpower would build a unique, one-to-one interface to connect each country's legacy system to any new app. Now, Manpower's using a services architecture, where apps publish data in common XML formats, so local IT teams can more easily integrate them to their legacy systems.
The speed comes from teams around the world working on small projects in parallel. "We're doing a ton of proofs of concepts and pilots, and we can do that because we understand what we want our architecure to look like," Edwards says. A year ago, he wouldn't have been comfortable allowing all this distributed development, because the architecture wasn't ready. "We have to make sure our impatience doesn't lead us to make mistakes that cause problems down the road," he says.
From Agile To Open Source
While cloud computing's the trendiest tool for IT speed, many IT pros will point to some long-standing practices as the best way to pick up the pace.
Agile development tops the list. Agile's had passionate advocates for decades, stipulating quick iterations of development followed by close collaboration with end users to confirm that developers are delivering what's needed. Companies such as UPS and General Motors are adding visualization software to the agile process, letting IT mock up what interfaces and transactions will look like to get business feedback before any code's written. Yet that also points to a limit to agile's ability to speed up IT: Business unit partners must be equally committed to it. IT can't do iterative development--what Shah calls constant "test and learn" mode--if business colleagues aren't putting in the time.
New converts to agile also tend to bring a certain religious zeal to its use, and Bleum's McKinnon warns that some projects are still better in traditional waterfall development, like those with very clear specs such as porting an old Cobol system over to .Net. A shop's goal should be development agility, being able to switch among styles. "A one-size-fits-all methodology will inhibit speed," he says.
Using open source software is another long-standing practice with potential to increase speed, since the code can be used in prototypes without a lot of vendor negotiations. Manpower is using open software for one proof-of-concept project, knowing it might go to something else later. But open source lets it get started at low cost and maximum flexibility.
Outsourcing, as always, is a flashpoint in the debate over speed. Some will turn to outsourcing to get the variable capacity or skills they need to move faster. Others will avoid outsourcing, believing speed comes from in-house IT pros who know the business and can iterate relevant software more quickly. None of these methods, from private clouds to agile development to outsourcing, guarantees speed. Companies need more than tools: They need an IT strategy that emphasizes speed.
State Street's Perretta measures time to when customers see results. "When does their stopwatch start, not mine."
Meet The Need
IT hasn't been slow for a lack of effort. Speed--in terms of delivering end results to customers--simply hasn't been central to the IT mission. Over the last decade, cost cutting has been more likely to dominate the agenda, both in terms of cutting how much IT spends and in using IT to squeeze costs out of business operations.
That tendency has left a huge strategic mismatch in companies. In a survey last year of chief operating officers at electronics and high-tech companies worldwide, Accenture found that 60% of COOs agreed that "building flexible and efficient IT systems" to expand ties inside and outside the company is more important after the economic downturn; yet only 7% said that building those flexible systems is "driving operating model decisions." And money isn't the problem: 95% said they have enough capital to execute their global strategy.
Perretta, State Street's CIO, thinks IT leaders are getting what they ask for. He notes that at most companies, the IT architecture group reports to the CTO, whose main charge is to optimize infrastructure costs. At State Street, chief architect Kevin Sullivan's team reports to Perretta, and its mandate is to collapse the time it takes for the company to deliver IT-based business results.
The demand for IT is unquestionably there. The reason IT's under the gun is because business leaders know they can't get anywhere without IT. Listen again to those COOs in the Accenture survey. Seventy percent said IT is critical to global expansion, though only 21% think their corporate IT systems are capable of supporting it.
This gap explains why CIOs are so impatient. IT is seen as too slow precisely because the business demand is so intense. IT needs to step up to that challenge. --With John Foley, Doug Henschen, and Rob Preston