Make a technology plan for 2005
- 31 December, 2004 07:06
The end of the year is a traditional time for re-evaluation of, well, just about everything. But one area in particular that gets re-evaluated in most organizations is technology strategy.
Many of the technology strategy issues you need to address in 2005 fall into five overlapping areas: server platform architecture, e-mail architecture, security architecture, enterprise analytics, and customer interaction technology. Here's a quick planning checklist of the top issues in four of the areas.
Server platform architecture. As usual, the most interesting issues in server platform architecture relate to database management systems. Included among those is this huge question: Is the row-based relational model approaching the end of its useful life? The evidence in favor of nonrelational and/or column-based DBMS architecture keeps mounting -- especially for XML-based applications. And while top-of-the-line DBMSs are still clearly superior for data warehousing, complex queries or nontabular data management, often that's exactly because they're getting increasingly far from simple third-normal-form data structures. Yet conceptually, there's little reason to see any limit to relational systems' continued evolution and, hence, continued market domination.
Meanwhile, at the other end of the DBMS spectrum, what about commodity database management systems? They do a good job for many smaller online transaction processing applications. Indeed, between MySQL and the once-excellent Ingres, even open-source DBMS offerings may warrant consideration.
But there's a yet bigger set of concerns: Just which kinds of platforms are "strategic" in your enterprise? Viable candidates include traditional hardware, storage hardware and software, DBMS, operating systems, application servers, enterprise information integrators and several kinds of management superstructures. And are the choices different for the idealized grid/blade future and the big-central-server present?
One good choice of strategic center is fourfold: DBMS + analytics engine + operating system(s) + network management infrastructure. But there are many viable alternatives.
E-mail architecture. Now here's a depressing subject. At most companies, e-mail represents a huge cost. Employees each spend hours per day on e-mail, grumbling all the way. E-mail administration is a huge nightmare. And vendors aren't helping much - Microsoft promised key fixes over a decade ago that it isn't even close to delivering.
So how about a radical gambit? E-mail is a constant processing stream with serious security vulnerabilities and regrettably little in the way of DBMS underpinnings at this time. That's exactly the kind of processing that does well on dedicated appliances, rather than general-purpose computers. And you shouldn't just consider e-mail appliances; e-mail outsourcing is also worth a serious look.
It's also time for serious planning in the area of e-mail archiving. Regulators want you to keep huge amounts of e-mail; lawyers, for very similar reasons, want you to throw it all away. Meanwhile, a tremendous fraction of an organization's knowledge is buried in all that e-mail, and archiving technology offers better potential access to it all the time.
Security architecture. At most organizations, security architecture is a nightmare. Dedicated security appliances (which work well), centralized security management (which doesn't) and general network operations management systems don't play together nearly as well as they should. Meanwhile, wireless networking and employees' personal Internet use punch ever more holes in the security perimeter.
You can't fix the whole shebang, but you can and should do at least this: Break your network into conceptual pieces, then move quickly to secure the most critical pieces as tightly as you can. Dedicated intrusion-prevention systems, network segmentation firewalls, newfangled application-level security appliances and so on -- you can't protect your whole network with those technologies, but your key databases can be protected very well.
Another area in which your planning should probably rev up (it's OK for implementation to be slower) is authentication and identity management. LDAP integration will eventually be a necessity and, despite the inherent hassle, many enterprises will eventually adopt smart-card/badge-based two-factor authentication as well.
There are other large, troubling challenges that I could mention -- especially issues relating to outsourcing. But since this is the holiday season, I'll end on a happy note instead.
Customer interaction technology. This area should be on your list for re-examination every year, and you should be happy about it. Technology for managing customer interactions keeps getting better -- on the Web, in call centers and even in hoary old direct mail -- and the impact on profits is almost immediate. So run, don't walk, to your marketing executives, and ask them to dream up a technological wish list. Chances are, you'll be able to give them a lot of what they want.
But this can all wait until the new year. Until then, have a great holiday season.