Consolidating Workloads: What Works Best

Consolidating Workloads: What Works Best

Many users are still under the impression that mainframe management is an esoteric business requiring a team of lab-coated technicians. Yet, it's possible to reduce staff and save money through workload and server consolidation.

What Is Consolidation?

A server is a computer program which provides services to other programs (and their users), in the same or other computers.

Server consolidation seeks to reduce the total number of servers or server locations that an organization requires. Workloads from separate machines or applications are combined into a smaller number of systems or applications.

Diverse workloads from multiple servers can be moved to a single larger server. Multiple workloads can be combined under a single OS, reducing the number of OS images. And multiple applications such as email systems or database clients can be combined in a single system.

Consolidation evolved in response to the situation where multiple under-used servers took up more space and resources than could be justified by their workload.

And Virtualization?

Consolidation can greatly increase the efficient use of server resources. But complex configurations of data, applications, and servers may result, which can be confusing for the average user.

Server virtualization alleviates this problem by masking the details of server resources from users while optimizing resource sharing.

Virtualization detaches a workload and its data from the functional details of the physical platform on which it's hosted. Workloads can then be matched more flexibly with physical resources, allowing IT administrators to manage infrastructures more efficiently.

Why Consolidate?

Over 70% of the total cost of ownership (TCO) for a typical data center can go towards labor or outsourced services. By reducing IT labor requirements – primarily administration and support – server consolidation can significantly lower the TCO of a company's data center.

Consolidation usually adopts one of four strategies.

1. Physical Consolidation

Here, servers distributed across multiple remote/branch offices and business units are brought into a central data center.

By restricting server access, and exploiting the superior data center infrastructure and security, greater control can be exercised over system configurations. The cost of moves, add-ons, and changes can be eliminated, as may the travel time and expenses associated with break-fix maintenance and support. With physical consolidation, purchases, configurations and management, best practices can be easily standardized.

Implementation costs are typically low, consisting of network enhancements to support the centralization data center modifications to support the consolidation, and physically moving the servers. IT labor savings can reach 10%.

On the downside, poor network planning can lead to reduced performance. The enterprise also runs the risk of catastrophe by having all its server assets in one place – especially if there are no recovery plans available.

2. Re-hosting

This involves porting from older legacy platforms and operating systems to newer, high-performance systems (which can support greater workloads). 

Migrating the operating system to a newer version enhances performance, availability, security, and management features. It also provides better upgrade options. Expensive support and maintenance contracts for legacy systems can be eliminated. And having fewer servers generates a proportional saving in administration and support labor.

Re-hosting costs are often underestimated. If the proposed system is not compatible with the older ones, applications may require porting to another platform, custom code rewrites, procedures and data migration.

3. Logical Consolidation

Individual servers are often configured into "islands", to allow for changes in workload and growth.

Logical consolidation establishes hard partitions for the operating system, applications, processors, and memory requirements, so that individual server "islands" are pooled onto a single server or cluster. Fewer servers are required, and applications can be hosted on a single cluster. Partitions can be modified to allocate more resources as the workload changes.

On a shared server, logical consolidation can save 40% or more of overhead headroom, with a proportional decrease in server assets. This can produce a similar reduction in administration and support costs.

Manually managing the hard partitions can be difficult in a dynamic environment, where workloads change frequently. So logical consolidation can introduce management burdens, complexity, and a reduction in service levels.

4. Workload Optimization

Here, the server operating system or third-party utilities are configured to intelligently manage server resource allocation, based on workload demands. Computing power can be allocated automatically, to meet changing needs. This means fewer CPUs, and in turn, fewer servers, to support multi-application workloads. Fewer, smaller applications that peak at different times give the highest consolidation.

This approach maximizes asset utilization and consolidation. Software licensing requirements, facilities costs and labor can be greatly reduce by 40% or more, based on the application profiles. Administration and support are minimized, as the system itself manages workloads and partitions.

Professional services assistance may be required, as establishing the workload optimization configuration and rules takes time and can be complex.

The Best Option?

Choosing the right consolidation strategy requires careful analysis of current TCO, proposed consolidation options and architectures, required investments, and potential savings. Internal IT teams should consult with independent analysts, performance benchmarking sites, and vendors.

Management commitment for the initiative is essential to provide resources for workload analysis, and for achieving buy-in from the people affected during implementation.

As the consolidation matures, changes in behavior will be required, e.g. getting developers to return server resources to the pool more promptly for more efficient utilization of resources.