Problems With Enterprise Applications Today
IT organizations face several problems that outsourcing addresses. Expense is the first: Enterprise IT is costly, not just in hardware purchases and software licensing costs but in maintenance, support, and staffing—all the factors that are rolled up into the so-called TCO (total cost of ownership).
The expense is exacerbated by the necessity to overinvest in the data center to accommodate peak CPU, storage, and bandwidth needs that may be reached only infrequently. (Having a fixed infrastructure also gets in the way when your operation exceeds capacity. How do you add further capacity in a pinch?) Additionally, those investments often require significant up-front capital expenditures that must be depreciated over an extended period, not expensed as they occur.
Beyond the direct costs, theres also the challenge of finding and retaining highly skilled IT staff for niche needs. Maybe youre a Windows shop that needs occasional Unix expertise, for example. Or you need to support legacy applications but no longer have the domain knowledge in-house. Perhaps you want to tap the best-of-breed experts in a particular field.
Implementing enterprise applications takes time, too, and the risk of failure is significant. Additionally, rapid changes in technology can make maintenance an ongoing hassle and slow a companys ability to adapt to new conditions. For nontechnology companies in particular, the reality may also be that IT, though important, is not a crucial competitive differentiator.
Smaller operations have even more basic problems. With an IT staff of one or two and a budget already stretched to the limit for maintaining network connectivity and databases, where do you turn for additional resources?
With these problems in mind, major consulting operations like IBM Global Services, EDS, and Accenture regularly ink multibillion-dollar deals in which they outsource entire divisions of large enterprises. But not all outsourcing efforts have to be so grandiose. On a smaller scale, a variety of providers let companies rent, in effect, items such as computing capacity, applications, or services on a pay-as-you-go basis.
Unraveling the available options can be confusing. Part of the problem is that the basic terminology is far from standardized. Here are some of the terms that occur frequently, and the contexts in which they appear.
Utility computing. A utility is a commodity thats available as needed via a standardized interface. You can plug any appliance into any electrical outlet, more or less, and get access to the power supply. Utility computing aims to make computational, network, and storage resources available on demand. Sun Microsystems N1 software, for example, "virtualizes" several physically distinct machines so that they behave as a single resource.
Other companies, such as EMC, HP, IBM, Verari, and Veritas offer similar capabilities. Utility computing may be appropriate for custom or customized applications of the sort that you might otherwise run in-house.
The sources of a true utility product would be fungible—you dont know or care which power plant generated the electric-ity you consume—though no IT utility today yet approaches that ideal. Its not particularly in a vendors interest to commoditize its own products to such an extreme.
Grid computing. The term grid applied originally to the large arrays of comparatively cheap commodity processor subsystems used in some supercomputer designs, as well as to the loose confederation of machines whose spare CPU cycles are harnessed by distributed-processing projects, like the SETI@Home initiative. Now, though, grid computing is sometimes used as a rough synonym for utility computing.
On-demand. Often used to describe hosted software, the term on-demand is particularly prevalent in the CRM (customer relationship management) market, where companies like NetSuite, Salesforce.com, and SalesNet offer Web-native approaches, unlike more traditional client-server implementations. (For a comparison of leading on-demand CRM products, see "CRM on Demand: Have It Your Way," First Looks, July, page 62.)
However, the term on-demand has a second meaning, closer to that implied by utility computing. As Dean Douglas, vice president of IBM Global Services puts it, "On-demand is about having the capability—tech infrastructure, software, et cetera—available as you require it."
Application service provider (ASP). ASPs are, loosely speaking, companies that host software and make it available remotely. They can be further divided into two subcategories: those that host packaged applications that could be installed in-house (like mail servers or accounting applications), and those with Web-native applications like on-demand CRM.
Managed service provider (MSP). As a subset of the application service provider category, MSPs typically offer network management and monitoring services. Managed security service providers offer services such as intrusion detection, perimeter antivirus filtering, and managed firewall support.
Software as a service and pay-as-you-go IT. These are variations of the on-demand and ASP theme of subscribing to software. Software as a service sometimes carries the further connotation that functionality is exposed as a collection of Web services components. In this way, companies like Amazon.com and Google, both of which publish Web services APIs (application programming interfaces), could even be said to provide a limited form of free software-as-a-service infrastructure.
Amazons API, for example, lets developers access catalog data and manipulate shopping carts using REST (Representational State Transfer) or SOAP (Simple Object Access Protocol) queries—essentially exposing Amazon functionality via Web services for anyones use. Googles API, which is currently limited to noncommercial use, provides an interface to Googles search results, page cache, and spell-checking engine.
Although Web companies like these are best known for being destinations in their own right, some analysts think that someday such companies may be well positioned to provide broad software-as-a- service capabilities. Could Googles Gmail, for example, be an opening salvo, just the first of many on-demand applications that the company might ultimately launch?
Regardless of the terminology used, its clear that a wide variety of capabilities are amenable to outsourcing. Bill McNee, founder and CEO of Saugatuck Technology (a Westport, Connecticut–based research firm) believes that the applications most likely to be outsourced in the near future are those "where configuration can meet your needs, rather than tremendous amounts of customization."
The most popular applications under consideration today are those that manage personnel and finance-related issues: benefits and personnel administration, travel services, payment processing, payroll, and the like. CRM, SFA (sales-force automation), ERP (enterprise resource planning), and customer service applications are much lower on the scale, even though CRM has probably the highest profile.