The IBM IT Economics team is a worldwide group of technical and financial consultants who work with clients to optimise their IT operations. The team focuses on identifying areas for efficiencies, cost reductions and increased business value for client business objectives.Clients ask the team to find infrastructure and solution improvements to minimise overhead and maximise qualities of service. Areas of analysis include hardware and software purchase and maintenance costs, disaster recovery, security, datacentre costs such as networking, floorspace, energy, and labour.
The study recently conducted summaries labour efforts observed by the team in client environments using IBM LinuxONE and distributed servers. These observations are based on IBM IT Economics studies for clients either running or considering LinuxONE as an alternative for their business critical workloads on distributed servers.
While each client’s IT environment was different due to service level agreements, server types and workloads, one or more of the following practices were found to drive two behaviours 1) use of more distributed servers than actually required and 2) use of more labour for distributed servers than for LinuxONE servers. Common practices in distributed server environments
- Some applications or lines of business do not allow sharing with other departments. Clients adhering to this practice indicated they require a higher aggregate number of servers than for a shared resource topology in which all users consume common resources as needed.
- Most clients confirmed development and test are rarely run on the same distributed server with production, and deploy additional physical servers dedicated to development and test use only.
- Multiple clients indicated that they have up to five duplicated environments with some combination of production, development, test, high availability, quality assurance, and disaster recovery.
- Most clients stated that they host their application and database workloads on separateservers.
- In some accounts the lines of business create separate copies of data to run analytics, each hosted on its own servers.
- For tier one applications, many clients indicated they adhere to a three-year server upgrade cycle to leverage the latest technology as well as tomitigate potential hardware failures.
- In some cases the client’s procurement department negotiates advantageous terms with a vendor for a specific server type (same number of cores and fixed amount of memory). Yet the same client’s IT department indicated that a ‘one size fits all’ server profile does not satisfy all core to memory ratio requirements. Depending on the workload, a server can run out of memory before exhausting CPU capacity. To mitigate resource constraints, their staff deploys one or more additional servers to support the workload.
- Some clients indicated that new server purchases can be challenging. If more than a year has elapsed since their last purchase, they may not find the same type of server with the same supported version of operating system. As these differences proliferate, more planning and deployment efforts are required.
- For some clients with larger networking environments, full-time dedicated specialists are required for capacity planning and the physical effort of router and switch management.
- A few clients with exceedingly intricate and large distributed server centers expressed challenges diagnosing and resolving outages
To read the full report check it out here.