Mary Kay is a cosmetics business which faced the problem that every company wished they had. They were growing so quickly that the system they have in place was no longer a viable option to process the orders they had coming in, which ranged from 25-30,000 per day. At the outset of the company, they had a decentralized system. This meant that the orders were scattered across 4 different mediums: web, phone, mail, and desktop computers, and many mistakes were made as a result.
With over 850,000 IBCs each having the capability to create their websites (150,000 created), the burdens on the decentralized servers was becoming too large. So, Mary Kay switched to a centralized system which could handle the increased need for storage and could manage the network. With the change from Compaq Alpha servers to the Unisys Enterprise Server ES7000 mainframe with 32 Intel processors, Mary Kay found that IBCs could enter orders in half the time and was set up to simplify the management of online operations as sales grew.
But, there is always the underlying question of whether implementing a centralized system was the best option for Mary Kay. There is no doubt that the decentralized system needed to be changed, but Galen Shreck states that “a lot of data centers can’t be broken up to run multiple tasks at the same time. ” Also, the system was chosen not necessarily because it was best, but because it was convenient to stay with the Windows 2000 operating system with the transition from servers to mainframe.
Instead of choosing the centralized mainframe system, which is already outdated in today’s technology world, I would have chosen to implement a distributive network infrastructure. For many reasons that I will illustrate later, the centralized system will prove to be inefficient for the tasks that Mary Kay wishes to perform in the future. Mary Kay has a simple problem, but there is not a simple answer. They have expanded at an increasingly fast rate with 4 different entry systems, which has become a problem for their network. So, they make the switch from a decentralized, server based system to a centralized, database system.
But, although there is the issue of which infrastructure system to choose, the main issue at the heart of the company is the complexity of having 4 different order entry methods. By having telephone, mail, web, and desktop computers bringing in orders, this vastly complicates the entire process. So, before Mary Kay begins to consider the problem of the infrastructure system, they should first consider streamlining the way they take orders to just one or two methods. If they had done this in the first place, there may never have been a need to change to the centralized system in the first place.
Second, the case states that although orders are taken throughout the course of the day, the vast majority are taken at days end. This puts a great strain on the servers as they have to process a large amount of information in a short time period when those orders could have been spread out over the day and put less strain on the servers. So, a change in the timeframe that IBCs input orders could do a great deal to simplify the problems Mary Kay had been facing. The final and biggest question is whether they chose the right system for their new infrastructure.
In order to best identify the correct system, you must know a great deal about the way the company operates. They have 850,000 beauty consultants which operate in all parts of the United States. They process up to 30,000 orders a day and receive these orders through 4 different entry systems. So, they seemingly have 3 options. First, they could have stayed with the decentralized system. Under this infrastructure, there is little to no IT sharing, and each server is connected to a system of hardware which would provide it with the order information.
Under this system, it proved difficult to process the orders, as each of the 40 parts of an order was processed inefficiently by the one server to which it was connected. Change was needed. Mary Kay chose to go with option 2, the centralized system, where although there is information sharing, it all occurs on one mainframe. Although this is slightly better than the previous system, it is still not entirely efficient because tasks cannot be broken down to run at the same time. In my final paragraph, I will state my recommendation as to which system the company should go with.
I believe that under the current structure of Mary Kay cosmetics, the best network infrastructure to use would be a distributed network infrastructure. By creating a network of servers across the country, each individual task could be allocated to the most efficient place, meaning that the problem that Shreck highlighted of tasks not being able to be run simultaneously would be solved. Because there are still many input methods, this remains the most feasible option, as cloud, tiered, and client/server networks would not be compatible with the telephone, pc, and mail entries.
With a distributed network infrastructure, Mary Kay can use the servers they already had in place under their decentralized system and spread them across the country to the main hubs of operation. By connecting these servers together, you create a network which can process the data at a more efficient rate and allows for the expansion that the company has continued to see. Although this is not the only option, because of the ability to allocate tasks to the most efficient places and still connect with the 4 input methods, I feel that a distributed network would have been the best choice for Mary Kay cosmetics.