3 ways to approach legacy hardware maintenance

Defining legacy hardware

With an average lifespan of three to five years, servers make up a large percentage of an organization's legacy data center hardware. Though legacy servers often operate alongside current-generation servers without issue, there are certain logistical problems and costs associated with aging server upkeep.

Legacy servers limit an organization's ability to run current workloads.

Today's business use cases make extensive use of machine learning and big data, which is processor-intensive. Legacy servers may lack the processing power to effectively run such workloads, which limits an organization's ability to dynamically respond to changing business needs.

Data center sprawl is another challenge. It takes several legacy servers to do the same amount of work as a newer server. Older servers also consume more rack space and switch ports. Plus, they require software licenses for OSes, applications and management tools, which are costly for organizations.

Legacy storage arrays are another relic. Capabilities differ considerably among storage array models, but the more common problems associated with supporting legacy arrays include siloed storage and capacity limitations.

Aging storage arrays are notorious for consuming large amounts of power. Legacy storage arrays produce a considerable amount of heat, which means that the data center's cooling systems must work harder to stabilize less energy-efficient hardware.

Data centers still support tape drives. Even if an organization no longer backs up to tape, it likely maintains its tape backup system in the event that admins must get information from older, tape-based archives.

Tips for managing aging hardware

There is no silver bullet to make legacy hardware management easier, but there are a few approaches admins can use to make it intuitive.

A big problem with hosting mission-critical workloads on legacy hardware is that, if the hardware fails, nobody sells an exact replacement. That might not be a problem for virtualized or containerized workloads, but it is something that admins must consider for processes that use physical hardware.

Many organizations still use old virtual address extension (VAX) systems, even though those systems are at least 30 years old. It is nearly impossible to port a workload off of an old VAX system and onto a VMware host in the event of a critical failure.

One option is to require any service contract providers to keep an inventory of replacement hardware on hand. This hardware can help the organization to bring workloads back online after a failure.

Organizations can also transition workloads off legacy hardware and onto emulated legacy hardware in the cloud. Stromasys has a cloud service that runs VAX, Alpha, HP 3000 and Sparc systems on emulation hardware within the cloud.

Repurpose legacy hardware

It is possible to repurpose old or aging hardware. Rather than retiring aging storage arrays and servers, for example, the two can be combined into a software-defined storage system. The system probably won't run high-performance workloads, but this is an attractive option for organizations that need general-purpose storage.

Legacy storage systems can have greater power and cooling requirements than modern systems. Some organizations have found that they can channel cool air through a raised floor setup to deliver more to the most needed areas; this can help reduce cooling costs.

Recognize legacy hardware is still useful

Organizations are demanding new levels of innovation and performance that are impossible to achieve with legacy hardware. Plus, legacy hardware management has a reputation for being expensive and time-consuming to keep functional.

As true as these ideas may be, there is a difference between a piece of seven-year-old hardware and a system that the company has used for 30 years. In the case of a server, storage array or tape drive that is five to 10 years old, it's a good idea to consider how to replace or repurpose the hardware.

On the other hand, if a system has been running for decades, then it may be best to leave it alone unless the organization wants to run the data on emulated hardware. The workloads that are running on decades-old hardware tend to more stable and secure, simply because they are far less complex and are more self-contained.