Until fairly recently, data center operators typically installed just one application (or program) on each physical server. When taking the additional servers dedicated to development/testing environments and disaster recovery into account, a ratio of three to five physical servers per application was commonplace in data centers.1 Because of the “one workload, one box” approach, most servers run at a low "utilization rate" – meaning that just a fraction of their total computing resources are engaged in useful work. A 2014 report by NRDC estimated average server utilization at 12 to 18 percent and noted that it had largely remained static from 2006 through 2012.2
For most datacenters, an inventory of all servers and an assessment of their utilization rates will uncover servers that are performing single, infrequent, or limited tasks. Consolidating these servers will allow data center operators to remove unnecessary server systems and reduce energy, hardware, and support costs in the process. There are several ways to consolidate servers:
A "virtual server" is a software simulation of a server and an operating system that executes programs just like a real server. Server virtualization offers a way to consolidate servers: it allows you to run multiple different workloads on one physical “host” server. Virtualization saves energy by reducing the required number of physical servers in a data center. For more information on virtualization, see the Server Virtualization section of this website.
Combining applications onto a single server and operating system instance
This approach involves hosting multiple workloads within the scope of a single operating system.3 Consolidation is generally most effective with homogeneous workloads. Two or three lightly utilized file servers, for example, can be consolidated onto one machine.
Server clustering reduces the number redundant or standby servers needed by a primary server system. Traditionally, data centers deployed one server to “stand by” as an automatic server failover solution for each primary server. That is no longer is necessary. “N+1” server “clustering” allows for only one standby server (“+1”) per cluster of servers (“N”). For example, instead of deploying six additional standby servers as “failover” servers for six primary servers, with N+1 clustering you need only one standby server. In this example, you need just 7 total servers instead of 12. Applications in the cluster are continuously monitored, and if an application fails on one server, it's activated instantly on another server within the cluster. Advanced clustering software can even include the capability to "direct traffic," or move applications among servers and storage devices to fine-tune performance and maximize workload.
Downsizing the application portfolio
Large organizations often run a number of very similar applications, some of which are underutilized or unused. This can happen as organizations add applications to their networks over a number of years, without a formal process for removing any. For example, when a government CIO surveyed the applications in use at a state agency in Massachusetts, he found no fewer than 14 different inventory asset management or fleet management applications.4 Fewer servers are needed when redundant applications are eliminated.
Savings and Costs
Consolidation enables you to use fewer servers, thus directly decreasing electricity consumption. Removing servers from a data center also means that less energy is consumed by air conditioners, power distribution units, UPS systems, and building transformers. As a result of these indirect energy benefits, saving one watt-hour of electricity at the server level typically results in an additional 1.9 watt-hours of electricity savings at the facility-level!5
According to the Uptime Institute, decommissioning a single 1U server (a standard-sized rack server) can save $500 annually in energy, not to mention $500 in operating system licenses, and $1,500 in hardware maintenance costs.6
Standardization through consolidation also means fewer spare parts are needed to keep the data center running. If you are only using two server vendors instead of ten, it is easier to get the cables, power supplies, fans, and other parts you will occasionally need.
Costs associated with server consolidation may include staff time for:
- Conducting a thorough inventory of the data center
- Migrating applications from one server to another
- Removing old servers and deploying new servers (if necessary).
There may also be end-of-life disposal costs for server systems that are not repurposed, sold, or donated. Many hardware manufacturers have recycling or trade-in programs, some of which may involve disposal fees.
Tips and Considerations7
There is always some risk of downtime when migrating applications and files from an existing server to a new system. Proper preparation can substantially mitigate this risk.
- Take inventory. It is easy to lose track of servers as time goes by, as mergers and acquisitions take place, as people procure servers without going through central IT, or as servers are retired or repurposed. The first step toward consolidation is identifying all servers in the organization. Take an inventory of all systems and servers, their associated application workloads, and note the equipment type, location, and Service Level Agreements (SLAs). Server usage can change often, so establish a policy for regular SLA reviews and inventory updates.
- Identify servers that should not be considered for consolidation. Servers that have privacy, security, or regulatory restrictions, or require ultra-high service levels, may not to be the best candidates for consolidation.
- Group remaining servers by OS. Servers running the same OS at low utilization are generally good candidates for consolidation.
- Combine workloads that have complementary characteristics. For example: Workload A consumes 10% processor capacity during business hours and very little at night. Workload B consumes 10% processor capacity during the night and very little during business hours. Combining these two workloads on a single server will have almost no impact on the performance of either workload.
- Keep an eye on cooling. Consolidating workloads onto a smaller footprint can create "hot spots" in your data center. You should monitor for possible hot spots and you may need to work with HVAC experts to ensure an adequate and efficient cooling configuration.
1 The Economics of Virtualization: Moving Toward an Application-Based Cost Model, IDC, 2009.
2 Scaling Up Energy Efficiency Across the Data Center Industry: Evaluating Key Drivers and Barriers, An Issue Paper by NRDC, August 2014. https://www.nrdc.org/sites/default/files/data-center-efficiency-assessment-IP.pdf (PDF, 485 KB)
3 Consolidation and virtualization: The same, but different, TechTarget.com, by Tony Iams, April 2005. http://searchdatacenter.techtarget.com/tip/Consolidation-and-virtualization-The-same-but-different
4 Application Consolidation — Where the Real Savings Is, by David Raths, Government Technology, June 27, 2011. http://www.govtech.com/pcio/articles/Application-Consolidation-Where-the-Real-Savings-Is.html
5 New Strategies for Cutting Data Center Energy Cost and Boosting Capacity, Emerson Network Power presentation, 2012, p.8. https://www.vertivco.com/globalassets/documents/blog/is03947_2012_energylogic_fin_62359_0.pdf (PDF, 1.4 MB)
6 Decommissioning as a Discipline: Server Roundup Winners Share Success, by Matt Stansberry, Uptime Institute, undated. https://journal.uptimeinstitute.com/decommissioning-discipline-server-roundup-winners-share-success/
7 Section was developed using the following document as a guide: Using Virtualization to Improve Data Center Efficiency, Green Grid, 2009.