Network Servers and Hybrid IT: Refresh, Replace, Recycle
Nothing lasts forever, including server hardware. This is the first blog in a three-part series that discusses some of the options to consider when your servers are nearing the end of their life cycles.
The typical lifespan of a server is up to five years; periodic upgrades and parts replacement can extend its usability up to 10 years. However, the older a server gets, the more prone it is to failure. The challenge for IT professionals is to determine the most cost-effective approach to dealing with aging servers before they can cause business-disrupting downtime.
Hybrid IT: Workloads Matter
To Refresh or Not Refresh
Many companies have hardware refresh policies in place; most seem to be set for five years. However, budget constraints can sometimes force IT departments to keep servers working beyond that five-year mark.
If you have to keep a server working longer than your company’s refresh cycle, you may be able to get by just paying maintenance costs or for replacement parts. The money saved can then theoretically be used to purchase new servers later, delaying that capital expenditure. The downside is that the older the servers get, the more at risk they are.
Technology advances can also affect when servers need to be replaced. Power conservation features, new processors and advanced processor extensions can make purchasing new servers a necessity.
Hybrid IT: The Server Hardware Quandary
The Case for Parts Replacement
If purchasing new servers isn’t an option, parts replacement may be. Regularly replacing or upgrading certain parts can extend the life of the server and is far more cost effective than replacing a whole unit. However, as servers get older, finding replacement parts becomes more difficult and often expensive. Warranties help. Servers typically come with three- to five-year warranties, and most server manufacturers reserve and cover replacement parts for customers whose hardware is under warranty.
However, a warranty can’t help you mitigate downtime or recover lost data — both of which can happen if a server fails. Then there is the scenario when a manufacturer discontinues support for older servers, including carrying the necessary replacement parts. Purchasing them from an aftermarket source can be expensive and result in even more extensive downtime.
Parts replacement isn’t always about preventing failures. Sometimes, aging servers exhibit performance problems. Adding CPUs to unfilled sockets or memory to available slots can significantly increase a server’s performance and at a fraction of the price of a new server. The problem is that not all systems are upgradable, and upgrades can’t always deliver the necessary capacity and performance improvements.
The Case for New Server Hardware
Going with new server hardware can mean new capabilities, features and opportunities to make operations more efficient or boost performance. The upfront investment will cost more than parts replacement or an upgrade. However, some of the costs can be recovered in the savings generated by increased efficiency and performance.
Installing new hardware can be disruptive and that comes with its own costs. However, virtualization makes the process less painful by allowing the new servers to be installed and configured in the production environment next to the working systems. Once the new hardware is in place, workloads can be moved over and tested. The old hardware can then be decommissioned or reconfigured for other use.
Hybrid IT: Make Colocation Your Conduit to the Cloud
Recycling the Old Hardware
Purchasing new servers doesn’t mean the old servers have no useful purpose. Some servers may be able to be used in development and lab environments or configured to handle router or firewall tasks. Some parts, such as hard disk drives and power sources, may need to be replaced, but getting more out of your overall investment may make those minor costs worthwhile.
Where to Start
If your servers are nearing the end of their life span, ask these questions to help determine next steps:
- Can I make a case for not adhering to the hardware refresh cycle (or for not trying to push it out farther)?
- If there are performance issues, will replacing the hardware solve them?
- Will new hardware offer different and/or better features and functionality?
- Will new hardware enable us to leverage new technologies better or to meet changing business requirements?
- Can I get by just replacing or upgrading specific components?
- Is it time to get out of the server hardware management business?