Impact of server virtualization on TradRetail Inc.

Business profile

The case study will now focus on the impact on the number of servers that early adoption of server virtualization would have had. 
 
Below is a recap of the business profile of the company in this case study as well as the core IT services that apply.

TradRetail Inc. is a consumer goods retailer with a head office that includes associated production and warehouse facilities.  Goods are assembled, stored and dispatched to a number of branch offices that are consumer-focused retail stores in a range of satellite locations. 

The following summarizes the company’s core IT services:

•The IT department has 15 staff. 
•65 Intel x86 servers are held in the head office data center and 11 servers in small server rooms in each of the other 2 sites.
•The growth rate is around 10-20% per annum.  The expectation is to initially add at least one new retail location a year.
•Server analysis shows the average utilization of non-critical systems is only around 20% and 10% in terms of CPU and memory (RAM) utilization respectively.  The critical clustered application servers typically make use of less than 25% CPU and RAM, allowing for peak utilization loads to be up to 2 times higher before running into resource constraints.  In addition, the spare cluster capacity mentioned above reduces this by between 30-40% depending on the cluster design since there are between one and two spare systems in each cluster.  This means that the clustered systems operate at only 15-17% of capacity.

The initial move to adopt virtualization has come from pressure from within the IT department and from software developers. They are finding it increasingly difficult to work on testing production systems and see that virtualization might remove some of their problems. 
 
As part of an overall hardware refresh cycle that is taking place, IT management has agreed to consider an initial project to migrate systems that do not have a direct impact on retail operations, finance, HR or line-of-business functions for the manufacturing, sales and warehousing systems. 

Review the list of servers in the environment and come up with a proposal to move as many systems as possible to a virtualized solution using standalone hypervisors. 
 
(Note that 30-40% spare capacity should be retained when consolidating separate physical servers as multiple virtual machines on single servers.  For example, if 100 physical servers run at 10% load, on average 100 servers could be consolidated into 10 physical servers.  In practice, consolidating them into 13-14 servers would provide spare capacity.)

For the purpose of this activity, the following assumptions can be made:

•All physical servers have an identical configuration with two dual-core CPUs and 8 GB of RAM.  (While improbable, the conclusions will remain valid.)
•The hardware refresh project will replace all servers in the environment.

•13 x Infrastructure (Identity Management, user accounts, security, 8 are for test and development)

•Standalone servers (22)
a.3 x HR 
b.4 x Education and Training
c.6 x File Shares (3 are for non-critical archives)
d.6 x Internal Web servers (2 are for non-business critical services)
e.3 x CRM server

•Clustered Applications and Services (30) (Vendor won’t support virtualization (17))
a.4 x Email Servers
b.3 x Enterprise Database systems
c.5 x Core ERP servers cluster. 
d.5 x Sales / POS management
(Vendor supports virtualization (13))
a.5 x Inventory 
b.4 x Accounts / Finance
c.4 x Document Management

• Standalone / Infrastructure Servers

a.2 x Infrastructure
b.1 x Inventory
c.1 x File Share
d.1 x Document Management

•Clustered Servers

a.2 x Database
b.2 x ERP Systems
2 x Sales / POS Management (on site)

• Identify low risk server candidates for virtualization.

• Calculate the capacity requirements for the new hypervisors. 

• Calculate the amount of compute (CPU) consolidation that can be achieved safely.

• Explore other potential benefits, such as energy consumption savings, that would result.