We use cookies and other similar technologies (Cookies) to enhance your experience and to provide you with relevant content and ads. By using our website, you are agreeing to the use of Cookies. You can change your settings at any time. Cookie Policy.


Capacity Planning: Whatever Happened to Predictability?

Originally published June 16, 2005

Once upon a time there was predictability in the world of computing. We had OLTP systems and peak periods of operation. Those peak periods typically were from Monday to Friday from 10:30 am to 2:30 pm, day-in and day-out.

With this kind of predictability, we had people with the title of capacity planner. The capacity planners looked at trends and determined when the corporation was going to need the next hardware upgrade based on usage. The capacity planners looked at past trends and factored in new applications and new business, mixed everything together and came out with a proclamation—“our current computer is going to last until May 14, 2005. At precisely that moment we will need more capacity for our business.”

Everyone felt so comfortable with this predictability. But today, there is no predictability. What happened to it?

Today all around us is unpredictability. We still have OLTP systems and their predictability. But today, with OLTP systems, we have other systems. Take the web environment, for example. If we have a website with any amount of traffic at all, we have noticed that one minute there are many users and the next minute no users at all. There simply is no pattern to the traffic that is driven to the website. For one thing, web traffic can come from around the world, where the working day is constantly changing. Stated differently, it is always prime time somewhere in the world. So there is no convenient 9:00 to 5:00 for the website.

Another factor is dictated by products which come and go. What is in vogue one day is out the next day and we just can’t tell with certainty when that will happen. Trying to predict the number of users at any moment in time is a game of chance.

But there is another major way that unpredictability creeps into our lives. And that way is through DSS systems. DSS systems and processing are notorious for being unpredictable. Even the most basic statement is fraught with difficulty –

  • Select * From Table ABC Where name = smith.

On one set of data 10 rows of data are returned. On another table 10,000 rows of data are returned. This is with the same code. When you are dealing with access to sets of data, you never know how many rows of data will be returned. You have to actually execute the query before you know what the results will be. This is like playing Russian roulette, where you breathe a big sigh of relief after the hammer has hit on an empty chamber.

You just never know how many resources are going to be used when you submit a set-based query until that query comes back with its results.

There is also the unpredictability when statistical analysis has been mixed in with standard DSS queries. One query may take one second. The next query may take 10 seconds. The third query takes two hours. The workload is so mixed and unpredictable that trying to do performance and capacity predictions on the workload becomes an exercise in futility.

But processing capacity is not the only thing that is unpredictable and out of control. Today, we have volumes of data in the data warehouse that never before have been encountered. The rate of growth for the data warehouse is unpredictable. There are so many factors that go into the growth of the data warehouse.

So, what’s a capacity planner to do? In many ways a capacity planner today is like a sailor adrift on a life raft in the middle of the Pacific. The capacity planner is tossed about at the whimsy of larger forces. About the best that the capacity planner can do is operate in a reactive mode. How do capacity planners operate in a reactive mode? The best bet that capacity planners have is to over-buy. If a capacity planner thinks that 10 units are needed, you should buy 20 units. That way you won’t be unprepared. True, you may spend more money than what you wanted, but at least you aren’t caught short.

The capacity planner should indeed keep track of resource utilization. A crude form of extrapolation can be attempted. This is the least that can be done in a day and age of unpredictability.

  • Bill InmonBill Inmon

    Bill is universally recognized as the father of the data warehouse. He has more than 36 years of database technology management experience and data warehouse design expertise. He has published more than 40 books and 1,000 articles on data warehousing and data management, and his books have been translated into nine languages. He is known globally for his data warehouse development seminars and has been a keynote speaker for many major computing associations.

    Editor's Note: More articles, resources and events are available in Bill's BeyeNETWORK Expert Channel. Be sure to visit today!

Recent articles by Bill Inmon

 

Comments

Want to post a comment? Login or become a member today!

Be the first to comment!