Why is it that the problems mainframe shops solved years ago are still major issues in open systems shops? I hear terms like "virtualization," "improved utilization," "capacity planning" and "information life-cycle management" as if they are brand new, but we've had these technologies for years. B.F. -- Springfield, Mass.
Why indeed? ILM in the mainframe world is hierarchical storage management -- been around since just after the advent of electricity. The only difference is that ILM is for the networked world, and HSM/mainframe stuff is direct-attached. Granted that poses a slew of new problems, but the point is, why did this take so long?
Virtualization? Please. Mainframers have run virtual arrays before Al Gore invented the Internet. LPAR anyone?
Utilization? Capacity planning? These are concepts that have been nailed in the mainframe world since the Franco-Indian war. Mainframers can tell you within 10 percent how much storage they will need next year, based on 57 years of historical data. They run their shops at 80 percent utilization. In the open-systems world, we don't have a clue how much we'll need and tend to run either at 4 percent utilization or 103 percent -- rarely in the middle.
The problem can't be people; it must be tools. Why aren't open-systems vendors modeling their "solutions" on the known good practices of the past? Why reinvent the wheel time and time again? I like what MonoSphere is talking about -- mainframe-caliber capacity planning tools for open-system environments. If vendors can provide tools to perform that kind of analysis and enable mainframelike processes and best practices, wouldn't the world just be a better place all around? I think so. The AppIQ stuff that Hewlett-Packard now owns has the same mainframe attributes for managing a storage environment. Mainframers use tools to enable better processes and best practices -- not as the end-all.
Don't get me wrong, I was one of the guys saying the mainframe was dead in 1989. It was too expensive, too limited, too restrictive, blah, blah, blah. Instead, it turned out that while all of that is true, it also works. People don't run into the computer room every 11 minutes plugging new things into the mainframe, then wondering why the entire Eastern seaboard crashed. Boring maybe, but there is something to be said for that.
So, it seems we can learn some things from our geriatric Cobol friends. You could ask them, but just make sure it's before 5 p.m. -- they get to go home when the workday ends.
-- Prior to founding Enterprise Strategy Group Inc., where he is senior analyst, Steve Duplessie founded and was CEO of Invincible Technologies Corp., a manufacturer of fault-tolerant network-attached storage systems. He has also held positions at Clearpoint Research and EMC Corp. Send your questions to email@example.com.