Capacity Planning Primer
Running benchmarks is only part of the application performance picture. “If you want to know at what load the program will bottleneck, how big a machine you need to do 100 requests, or how soon you need to upgrade the server, then you can measure it,” writes David Collier-Brown.
In this article, Collier-Brown provides detailed examples using the textbook response-time curve, or “hockey-stick” curve. “The only people who always use it are capacity planners and performance engineers. They use it because they can take a few measurements, plot some lines, and credibly estimate both curves,” he says.
“You don't need to do a full-scale benchmark any time you have a performance or capacity planning problem,” Collier-Brown explains. “A simple measurement will provide the bottleneck point of your system.”
Read more at ACM Queue.
Related content
-
All you ever wanted to know about hard drives but never dared ask
As a prelude to a deeper dive into storage tuning, this month we explore the basics of hard disk drive performance. -
Beyond monitoring: Mathematical load simulation in Perl
Simulations and the ability to predict the future can show what effect adding new hardware, or changing the load, can have. Armed with the right know-how, anybody can use this tool. -
Energy efficiency in the data center
Storage systems are one of the biggest factors in power consumption, so data storage can make a massive difference in operating costs. We look at how you can achieve savings through technologies such as flash, tiered storage, or even cloud-native container environments. -
End-to-end monitoring – Measuring what the user perceives
Continuously monitoring the performance of applications helps ensure the service quality in multilayered cloud services. - New TOP500 List Pushes the Limits