Considerations for modelling physical and virtual servers in a data centre
The flexibility, efficiency and reduced cost of ownership that virtualisation provides, makes it extremely compelling to large and small organisations alike. Increasingly organisations are contemplating virtualisation across all platforms. As this trend makes its way deeper into the data centre, organisations are starting to leverage the fact virtualisation also lifts many of the constraints that govern which platform an application needs to run on. Different types of applications possess different workload ‘personalities’ and these heavily influence how well an application will perform on a given virtualisation model.

By modelling transformations in terms of the constraints that govern them, it’s possible to chart a course that reaches your end goal. These constraints can be classified into three categories:

1. Technical – what can go together?
2. Business and Process – what should go together?
3. Utilisation Analysis – what fits together?

Technical Constraints
Technical constraint analysis deals with compatibilities and affinities between hardware and software components:

• Hardware and Software Compatibility
  Certain types of guest operating systems simply are not supported on certain virtualisation platforms, and for those that are, there are typically a number of constraints that must be considered. For example, in many virtual environments there are limitations on the use of virtual symmetric multiprocessing (vSMP) that vary based on the type of guest OS, and ignoring these can lead to unpleasant surprises. Likewise, any transition must take into account whether the source systems employ specialty hardware such as token rings, faxes, and USB devices that may be difficult to move and/or support in the target environment.
• Server and Application Affinity
  In many cases it is important to either keep workloads together on the same physical servers or to spread them apart, conditions referred to as “affinity” and “anti-affinity” respectively. For example, when optimising memory sharing in VMware it is important to analyse the software and services running on each VM in order to group similar OS images on the same server. Similarly, by grouping applications that talk to each other, it is possible to realise tremendous benefits by promoting “crosstalk” that doesn’t hit the network. The reverse is also true, and for VMs that are performing application-level clustering it is typically desirable to ensure that they are never placed on the same servers.

Business and Process Constraints
These constraints help organisations determine what can and cannot be done from a non-technical perspective due to regulatory requirements, internal politics, or other real-world considerations.

• Process-Oriented Constraints
  The proper functioning of production IT environments often relies on tight process controls. Change freezes, maintenance windows and other controls must be respected, and transformation of IT environments must take into account the rules that have been put in place.
•  Security & Regulatory Compliance
  To protect intellectual property and other sensitive information, many environments segregate data types and apply security and access control policies accordingly. Virtualisation changes many things, but rarely does it change the basic security requirements of an environment, and businesses should therefore avoid virtualisation scenarios that violate security zones, intermix sensitive data or cause other security vulnerabilities.
• General Business Constraints
  Organisations are also often resistant to sharing infrastructure between departments, partly due to the lack of chargeback models that enable cross-department hosting. Even with these models in place, however, there are sometimes other practical considerations that affect sharing, such as the complexity it creates when isolating faults (namely finger-pointing) and even the desire to phase implementations to allow users to become comfortable with the new operational model.

Workload Analysis
Because of the diverse workload personalities present in the data centre, and the varying characteristics of the platforms, getting the right workload analysis is no easy feat. CPU utilisation must be properly normalised, and analysis must take into account I/O rates, memory utilisation and context switching to ensure accurate results.

• CPU Normalisation and Benchmarking
  By employing multiple benchmarks that represent the relative performance of source and target systems for a given workload personality, it is possible to construct a single analysis that optimises utilisation across multiple source/target combinations in a single step. Each source workload is then able to be normalised independently based on its distinct personality, and the target endpoints filled to capacity based on a reasonably accurate projected utilisation.
• I/O Analysis
  In any consolidation scenario it is desirable to combine workloads that dovetail in a way that promotes a balanced use of resources. I/O constraints should be analysed in parallel with CPU and other metrics to give a balanced result that makes optimal use of resources. Also, the impact of I/O activity on CPU utilisation, a form of overhead present in some forms of virtualisation, should be factored in to produce truly accurate results.
• Memory Analysis
  Memory is often a deciding factor when determining your strategy, yet this can be tricky to analyse. Using the measured memory utilisation to determine what can fit on a target will provide a safe answer but might require the use of more memory than is actually required. Specifically, technologies such as overcommit and page sharing can produce significant memory savings if VM placements are tuned properly, and this effect should be reflected in the analysis phase, particularly if multiple types of virtualisation are being compared to each other.

Conclusion
Ultimately, all constraints must be analysed together to determine the optimal data centre. When considering this array of constraints against a series of source and target servers, the analysis becomes a three-dimensional optimisation problem. By properly assessing the suitability of applications to consolidate, the results they will give, and the overall TCO of the solution, it is possible to uncover significant opportunities to simplify IT infrastructure, improve reliability, increase resilience, decrease power consumption and ultimately drive down costs.