What are the differences between supporting applications in an inhouse, server-based computing environment and supporting applications in an application-server environment? To answer that question, I chatted with Eric David Greenspan, founder and CEO of the former Platinum Citrix integrator Make It Work, which in September 1999 evolved into an application service provider (ASP) called Push. Greenspan is a Citrix Certified Administrator (CCA) and is well acquainted with the requirements for supporting a server-based network, whether inhouse or at a centrally located data center.
Before I talked to Greenspan, I thought that the ASP model might present technical problems that don't exist in the inhouse model. As it turns out, I was wrong about the technical differences, but he found strategic differences—all of which translated to "it's easier to be an IT person for an ASP than a thin-client integrator or inhouse staff."
First, in the ASP model, sheer volume is on your side. Make It Work became an ASP for a simple reason: IT departments often don't have the necessary staff to run complex server farms. Network administrators who are trained and experienced in maintaining server-based networks are rare. Sometimes, after setting up a company's server farm, Make It Work had to help the client find a new IT team to help manage its farm. In a nutshell, Greenspan found that the benefits of server-based computing always outweighed the liabilities of managing it, but sometimes that management was too heavy a load for a single company's IT department. When you're problem-solving, it helps a LOT to have 15 other engineers in the same room. Most IT departments are in companies whose main focus is something other than providing IT. As a result, IT departments usually get enough money to function but not enough for all the staff and equipment they could use.
Another strategic difference Greenspan found between server-based ASPs and inhouse models is the degree of control technical staff has over the network. Not only is the ASP completely in control of the network, it can maintain control over the applications to a degree that others might envy. Push users can make normal modifications to applications, but only Push makes changes that could impact application performance or availability. Significant changes don't go live on the production network until they're tested on a staging server.
Finally, Greenspan believes the ASP is more likely than the inhouse IT staff to punctually perform routine administrative tasks. Inhouse staff members are overworked, and the impact of not quickly getting to certain tasks is different. ASPs get paid for new customers, so they have financial incentive to set up new accounts promptly. An inhouse engineer lacks that incentive and has other tasks to perform, many of which are more urgent than getting a new account up and running. Consider system backups, for example. Particularly in smaller or understaffed companies, backups often aren't done properly or at all. On the other hand, ASPs live and die by whether customer data is safe and available, so as Greenspan put it, "We have people who sit there and watch the lights blink \[during the backups\]."
In short, a Push engineer isn't doing anything a person who supports inhouse terminal services would do—IF that person had the resources to do it. I asked Greenspan if one could transfer the ASP model's benefits to people who support server-based computing inhouse. Sadly, he didn't think so, because the constraints on IT budgets prevent the same concentration of experience. Truthfully, the conversation was a bit depressing for those of us who don't work in the same room as 15 other Citrix administrators, but it illustrates why the ASP is the logical conclusion of the server-based computing model. The point of server-based computing is to concentrate resources and provide control. Because some of those resources include equipment budgets and technical expertise, the ASP model seems to have the advantage.