How can you determine the performance impact on SLAs after an N+1 redundant hosting configuration fails over? This question came up during Dr. Neil Gunther’s Guerrilla Capacity Planning class. It can be addressed by referring to a multi-server queueing model.
N+1 = 4 Redundancy
We start by considering a small-N configuration of 4 hosts where the load is distributed equally to each of the hosts. The load distribution is assumed to be performed by some kind of load balancer. The idea of N+1 redundancy is that the load balancer ensures all 4 hosts are utilized. The common misconception is that none of the hosts should consume more than 75% of their available capacity. In other words, the total capacity is assumed to be 4×3/4=3 or 300% of the configuration, rather than 400%. When a host fails, its lost capacity is compensated by redistributing the load across the remaining 3 hosts.
Please click here to read rest of the article.
- Why The Linearizer is called a Linearizer The Mean Value Analysis algorithm for Queueing Network Theory, “Linearizer: A Heuristic Algorithm for Queueing Network Models of Computing Systems” by Chandy and Neuse is one of the most popular and extensively used algorithm if not the most popular. In Section 4 of the paper, the […]
- Don’t Panic About The Math When you read about performance you are often presented with complex laws and elaborate equations. All that can be intimidating, but fear not. I paid off my mortgage and retired early by doing useful and productive performance work with the ideas presented here. The most complex formula […]
- Do Applications Have Performance DNA ? – Part I About This Series - This series of articles is a how-to manual on applying performance modelling techniques without any needless discussion of internals or mathematics (with the exception of the necessary but simple Little’s formula which will be included in the 2nd installment of this […]
- Applications Have A Usage Volume – Part II About This Series - This series of articles is a how-to manual on applying performance modelling techniques without any needless discussion of internals or mathematics (with the exception of the necessary but simple Little’s formula which will be included in this article which happens to […]