All large multi-location enterprises have a tendency toward server sprawl. Multiple, under-utilized servers take up more space and consume more resources than can be justified by their workload.

Industry analysts D.H. Brown Associates estimate that servers in many companies spend much of their time running at 15-20% of capacity. Server consolidation has become one of the shortest routes to cutting unnecessary costs and lowering the TCO of a company's data center.

There are several forms of consolidation. Heterogeneous workloads from multiple servers can be moved to a single larger server. Multiple workloads may be combined under a single OS, reducing OS images. Multiple applications (email systems or database instances) can be combined into a single system.

Re-Hosting
One form of consolidation called “re-hosting” involves porting from older legacy platforms and operating systems to newer systems. Fewer new high-performance systems are typically needed to support the workload. Savings include the elimination of older, expensive support and maintenance contracts. Since the cost of administrators and support labor is correlated to the number of servers, fewer servers means less overhead.

The downside is that applications may require porting to another platform due to incompatibility with prior systems. Resulting costs can come in the forms of custom code rewrites, procedures and data migration. Porting costs are often underestimated and can be expensive.

Logical Consolidation
Logical consolidation establishes hard partitions for the operating system, application, processors and memory requirements on a single server or pool of clustered servers, so individual server "islands" are pooled onto a single server or cluster. This also reduces overhead and administration costs. Plus, as needs change, the team can change the partitions to allocate more resources as needed for workload demand changes.

Unfortunately, manually managing the hard partitions is difficult in a dynamic environment where workloads change frequently. There are political costs, too. Many business units will not support logical consolidation where they must share servers. They’ll have to be "sold" on the business merits. Having more eggs in one basket increases the importance on availability and business resilience.

Workload Optimization
Another consolidation tactic incorporates configuring the server operating system or third-party utilities to intelligently manage server resource allocation based on workload demands. Partitions can be established based on demand and schedule rules so that the system automatically meets needs.

Benefits are fewer CPUs and fewer servers to support multi-application workloads. The approach maximizes asset use and reduces software licensing requirements, facilities costs and labor. Establishing workload optimization configuration and rules will take some time and can be complex, requiring the help of professional services. As with logical consolidation, business unit apprehensions and business resilience best practices apply.

Physical Consolidation
Probably the least expensive form of consolidation involves relocating servers distributed across multiple remote/branch offices and business units into a central data center. By reducing complexity and standardizing purchases, configurations and management best practices, organizations can save 10% or more on labor costs. Other key benefits include improved configuration control by restricting server access and strengthened business resilience.

Performance degradations due to inadequate network planning create problems under this technique. Companies also risk business resilience by having all of the server assets in one basket, particularly if the data center does not have adequate recovery plans.

Virtualization as a Component of Consolidation
Server virtualization separates a workload and its data from the functional details of the physical host platform. Virtual servers act like independent hardware units, allowing companies to consolidate the number of physical servers they use. This increases the flexibility of matching workloads with physical resources. Administrators can develop business-driven policies for delivering resources based on time, cost, and service level requirements.

Traditionally, x86 servers have been used in one-box to one-application scenarios. In that model, servers typically use about 10% of their computing capacity. Virtualization turns one server into many, allowing it to run several operating systems and applications. Leading virtualization provider VMware promises they can:

• Increase utilization of existing hardware by as much as 80%
• Reduce hardware requirements by a 10:1 ratio or better
• Reduce energy costs by as much as 80%

It’s probably easier to virtualize workloads than it is to consolidate them but there are cases where it won’t fit. Complicated applications that require servers to be inside and outside the firewall may not work well. Heavily utilized systems may not deliver acceptable performance when virtualization overhead is added. For most organizations, some combination of consolidation and virtualization offer the largest savings in the short term.

VistaOne can Arm You against the Downside
VistaOne has supported a number of server consolidation efforts, employing tools that eliminate latency, identify platform incompatibilities, establish usage policies, secure access and reduce hardware requirements in branch offices.

Both Blue Coat and Riverbed products enable consolidation of IT from branch offices to central sites without affecting the end user experience. Both manufacturers have solid working relationships with VMware.

A joint Blue Coat and VMware solution allows IT organizations to consolidate servers and perform disaster recovery operations in real-time without performance degradation due to latency or limited bandwidth. The result is improved uptime of systems, accelerated recovery from failure, quicker provisioning and the elimination of redundant file and application servers through virtualization.

The enhanced Riverbed Services Platform (RSP) provides customers the capability to run as many as five additional services and applications virtually on VMware in a protected partition on the Steelhead appliance. Customers can deploy local services in all their branch offices without the need for actual servers to run the applications.

Riverbed Cascade combines network flow data already existing in an IT infrastructure with Network Behavior Analysis to provide the information IT teams need to minimize the risk of datacenter change by:

• Passively discovering which servers run which applications
• Providing server-server dependency maps assuring that planners reassemble servers correctly the first time
• Providing baselines of user activity by application and server to identify the optimal change window and minimize service disruption
• Providing application performance reporting and analysis to streamline troubleshooting of unforeseen problems on return to service

Helpful Resources:

WAN Optimization Is Growing Up and Spreading Out

Accelerating Virtualized Environments with Wide-area Data Services (WDS)

Server Consolidation with Blue Coat

Blue Coat and VMware Facilitate Real-Time Disaster Recovery and IT Consolidation

Riverbed Services Platform

Riverbed: Notes on Server Consolidation

Riverbed and VMware VI3