Blog
Cloud News & Insights

Holiday Season, Online Stores And Their Servers – Part 2

e-commerce icons

Why did Hank’s ‘HanKing’ e-commerce store go down like a pack of cards? There are many reasons and numerous variables which decide the fate of a website. Unfortunately no single factor is solely responsible for deciding web server load. Therefore the decision on the number of servers required and their capacity depends on the risk you are ready to take.

How does one decide server load?

There are many ways to estimate load on a particular web server with a certain configuration. Mostly the formula for server capacity is highly dependent on CPU utilization. Other factors such as internal memory, cache and gateway capacity do not come into picture. Simulating load on a web server has its limitations. Nevertheless, transactions per second, average response time and queue length are other factors which come into play. It is clear that a novice like Hank could never have understood the calculations which determine server capacity.

 Load balancing  

When you have multiple web servers catering to a single website, the issues of load sharing become more complex. DNS Round robin is fairly common while there are many other advanced load balancing techniques. The inherent problem with some of these is the fact that  a single point of failure may lead to a cascading effect and quickly lead to a collapse of the entire web server chain. This is probably what happened to ‘HanKing’ when the website simply went down in a short period of time.

If you are running a website with high traffic and horde of users, load balancers are just what makes cloud technology so apt. GMO Cloud America, Inc. is offering load balancers which has a feature of auto-scale and round robin as well.

Server Capacity and efficiency 

It’s rather important to ensure that website visitors experience unhampered and smooth transactions and least frustration in completing their objective. Any deterioration in server performance can lead to visitors abandoning a site which would in turn lead to lost opportunity.  This would become a serious issue, more because estimation of these occurrences due to performance issues is quite difficult. One of the ways to overcome such a problem is to utilize only 80% of server capacity. Here too the estimates vary – some may put a limit of 50% as ideal server load.  Efficiency of transactions also depends on the type of application being run on a server.

Hank’s solution

Obviously server load capacity estimation etc. was beyond the understanding of Hank. But he was a shrewd businessman. Once his technical staff came up with a certain number of servers which would cater to the estimated traffic to his website, he simply doubled up the numbers. However, as a result of doubling up his servers, he had to pay double the cost for the sake of safety. What he didn’t cater for was the sudden spike on server load. Ultimately, the inevitable did happen.

Conclusion

Deciding server capacity is still not a perfect science. There are many factors which ultimately decide your fate or that of your website. Scaling up your servers to meet surge in traffic during Holiday season can lead to unnecessary expenses, which one can least afford during the prevailing highly competitive market.

In my next post, we will look at how Hank could have taken recourse to cloud technology to optimize resources and also save money. ‘Till then, happy holidays.

Be Part of Our Cloud Conversation

Our articles are written to provide you with tools and information to meet your IT and cloud solution needs. Join us on FacebookTwitter and Pinterest.

 

About the Guest Author:

Sankarambadi SrinivasanSankarambadi Srinivasan, ‘Srini’, is a maverick writer, technopreneur, geek and online marketing enthusiast rolled into one. He began his career as a Naval weapon specialist. Later, he sold his maiden venture and became head of an offshore Database administration company in Mumbai. He moved on as Chief Technology Officer of one of the largest online entities, where he led consolidation of 300 online servers and introduced several Web 2.0 initiatives. He holds a Master’s degree in Electronics and Telecommunication.

Share on LinkedIn