IT strategists and visionaries peering into 2004 and beyond see a potentially catastrophic flaw in existing enterprise solutions: complexity.
How serious is it? Every business that hosts or outsources large-scale solutions will take a hit.
Even those least affected will be slammed with rising subscription costs for high-volume, technology-based services, such as mobile phone communications and Web hosting, as providers rebuild and simplify their infrastructures. In the worst cases, IT operations will suffer service degradations and acute failures that will be difficult to diagnose and expensive to fix. Complexity isn’t the only factor limiting the lifespan of existing IT solutions, but it is the easiest to describe, identify, and fix.
Change is the stressor that transforms complexity into failure. With all the gear-grinding that chief technologists are doing to adapt to this freakish economy, change is already bearing down hard on IT staffs and their cobbled-together solutions. These solutions scale and they’re secure, but they are nailed down, fixed to their roles and dependent on specific combinations of components. That’s what it means to say that a server is “in production” -- it is presumably so fragile that nobody is allowed to touch it. But chief technologists will require that their IT assets move where they’re needed, when they’re needed, without regard for boundaries between departments, roles, or even platforms.
A new paradigm for IT architecture is taking shape that makes such flexibility possible.
It values simplicity and manageability over traditional design goals. Paradigms come and go, and for the most part IT leaders are free to embrace them or stay their course — but not this time. The dynamic-enterprise paradigm, which dictates that all computing assets be virtualised and effortlessly reconfigurable, will stick because it’s what companies need.
Where to find it
Vendors are hurriedly creating dynamic (also called “agile” or “adaptive” by various vendors) solutions in the form of products and initiatives such as N1, the heart of Sun Microsystems’ new dynamic-enterprise strategy. Hewlett-Packard and IBM are recasting existing initiatives for grid and utility computing, intelligent storage, and automated management software as building blocks for dynamic infrastructures. IT will see a flurry of vendor activity around this subject in 2003, but you might have to squint to recognise it.
Complexity is an enemy to business, but it’s a powerful ally of IT vendors and consultants who are paid to unravel it. But nothing is so appealing as the idea that, based on business requirements, a server on one side of the building can be yanked from its current role and virtually teleported to where it’s needed, switching platforms from Windows to Linux along the way. The administrator just clicks and drags. That’s what Sun’s N1 does now, and it is a model approach.
Sun isn’t the only vendor that knows it needs to help customers get ahead of this problem. HP, IBM, SAP, Tibco Software, Sonic Software, and other forward-thinking enterprise players have strong technology portfolios they can quickly bring to bear on this challenge. Although it’s a radical change for most vendors, it’s well worth it.
But at the same time, vendors are careful not to upset their customers — after all, the technology IT uses today is precisely what vendors, analysts, and consultants recommended; everyone involved should focus on getting ahead of the problem instead of assigning or avoiding blame. The rules are changing in ways nobody could predict.
Wildly fluctuating business conditions are forcing companies to make huge, short-notice adjustments to their products and services. Wave after wave of radical requirements changes are rolling in as executives scramble to stay afloat, exploit a narrow opportunity, prevent a shareholder revolt, consolidate staff, or bail out of an unprofitable product line. It’s vitally important that IT staffs get each request right on the first try, but sometimes there isn’t enough time, money, or manpower to do it right. Still, the company’s band of high-tech heroes pulls a series of all-nighters and somehow, miraculously, gets the sorely needed services online in the nick of time.
If the requirements you hand down to your IT implementers are always met, it probably looks as though your operation is dynamic enough. If you ask the technical lead to stick his or her head into a conference room to explain how it was accomplished, you’ll likely hear reassuring phrases such as “Web services” and “component-based architecture”. Maybe you’ve got a wizard on staff who will toss in “business-process management”. They’re all real, workable, and important concepts, and without them half of the world’s IT solutions would have quit working already. Those heroes who keep you in business make their miracles happen by actually using the technologies these trendy buzzwords describe. Amazingly, solutions to most business problems can continue to be built by wiring together the hardware and software components available right now. What’s wrong with that?
Through the 90s, companies treated technology upgrades as business imperatives. To rush new capacity and new services onto the network, IT staffs stacked new systems on top of old systems, plugged integrated appliances into discrete black boxes, added vendor B’s best-of-breed point solution to vendor C’s packaged application. The integration craze demanded that every box be wired to every other box, and then all applications wired to all other applications. If you could just get information from one end of your server room to the other and through all the silos and murky points in between, your enterprise would be integrated and your work done.
Let’s say that describes the backend portion of your CRM system. Now wade into that rat’s nest with your smartest, most honest technical person. Point to a fibre switch, a scattered handful of servers, and a network storage device — no, half of the storage device. Accounting is closing the books for the quarter and it needs to borrow those resources for a while. Ask that technical person, “How long will it take to move all this, and can it be done without disrupting CRM or accounting?”
That’s the kind of request that executives have learned not to make of their IT operations. The business person asking that question would be derided for his failure to appreciate the complexity involved. Reflect on that for a moment. It’s time to remind your technology why it’s there in the first place. Your IT assets don’t exist to shape your company’s business needs and capabilities, yet complex, fixed technology makes modern business objectives difficult to meet. It’s time for some changes.
A dynamic-enterprise appliance, service, operating system, or middleware layer will never transform an overly complex, inflexible network into a simple, adaptable one. Don’t start down the road toward a dynamic infrastructure by shopping for new products. Instead, start by looking for your points of greatest risk — in particular, that which is fixed to its present function, dependent on technology only available from one source and inadequately understood by your IT staff — and whittling down your unique assets to the smallest possible set.
Every IT shop must assess its particular vulnerabilities. Some will find, as they fear, that the existing architecture cannot be finessed toward a simpler, more flexible approach. Hunks of technology will sometimes have to go. Yes, re-architect, rip, and replace. Or just rip if that’s all you have time to do; often the consequences of the old tear-and-toss are minimal because most networks have a lot of unnecessary pieces.
Start with a meeting of the sharpest minds you can gather across your IT and business operations. Recognise that complexity is a potentially crippling problem. Go around the table and ask for examples of vulnerable solutions: contrived or over-engineered solutions, proprietary or undocumented links between systems and applications, and anything that no one can describe concisely. Then decide what you’re willing to sacrifice to solve the problem.
Most IT leaders are already attacking the issue of complexity at a basic level by reducing the number of discrete boxes on the network. Even if you don’t yet subscribe to the broader philosophy behind eliminating complexity, basic asset consolidation is worth doing. As you go along, the dynamic-enterprise idea will naturally take shape. Servers and storage are ripe for consolidation. Anything you unplug from your network is a step in the right direction. But don’t equate consolidation with simplification — the total number of distinct assets on your network is not a good measure of complexity. If you have 30 routers presenting seven distinct management/monitoring interfaces, seven is the number you should reduce first.
Take the first step toward reducing that number by enlisting your vendors’ help to push out firmware updates. If a device can’t be updated because it’s too old or if the vendor isn’t keeping up with relevant cross-vendor standards, rip it out even if you can’t afford to replace it right away. That sounds extreme, but it’s a prime example of a brittle point in your infrastructure. In a year, if you need to move it some place else, your dynamic management solution may provide only limited interaction with it. As much as possible, make assets in a given category look alike from a management perspective, which can be done without selling out to a single vendor.
The aspect of Sun’s Network Computing strategy that resonates most is the idea that manageable assets — starting with systems, storage, and networking but eventually working down to the granularity of processors, memory, and services — should not have roles or personalities. Let’s say your accounting database server runs Windows 2000 and Oracle 7. You can’t hope to virtualise or relocate that asset because it’s too bound to its complex identity. Take away any piece of it and the rest falls apart — that’s too fragile.
The software each server is running is irrelevant in a dynamic setting because fixing those details makes the server nearly impossible to pool and reallocate. In time, you want your dynamic management system to be able to shift that box to customer service to run desktop thin clients during weekday banker’s hours, accounting at night for database crunching, and the e-commerce site on holidays. The server can’t belong to any department, role, or vendor; it can’t even matter where it sits in the building, or whether it’s in the building at all.
You have a lot of work to do before any part of your infrastructure can be considered dynamic. Vendors are obviously motivated to help out, but be cautious. Virtualisation that operates only within a vendor-specified platform or brand boundary doesn’t create a dynamic infrastructure. Investing in proprietary or restricted technology creates the likelihood that you’ll have something new and expensive to rip out of your network later. Sun still has to prove itself through execution, but it certainly sets the gold standard for commitment to the concept of the dynamic enterprise: If you choose to, you can use N1 to make Sun’s brand invisible on your network or to kick Sun completely out of your shop. If you understand why that has to be so, you’ll probably make good choices as you simplify your infrastructure to start building your company’s dynamic enterprise.