Editor’s note: This is the last in a series of four articles on server hardware equipment refreshes. Part one detailed strategies for getting the timing right; part two showed you how to garner support for the project; and part three coached you about hardware selection. In part four, Daniel Eason explains some tactics to use that will lay the groundwork for a refresh proposal down the road.
In most circumstances, a server refresh proposal is solid enough to get approval. But what if a data centre manager doesn’t get budget signoff for a hardware refresh?
Some viable lower-cost alternatives can suffice and involve a lower expenditure, but these options do not eliminate the risk of performance problems. Include this caveat in any alternative, lower-cost proposals to avoid stakeholders assuming that your new revised proposal eradicates risk.
Server component upgrades. A first option is a server component upgrade. Upgrade activity involves buying RAM and/or CPU at a lower price point than is available with a completely new server. Upgrades can also give a much-needed performance boost to support new business-line demands.
Hardware upgrade options are possible only if you have sufficient additional RAM/CPU slot capacity available to perform the upgrade. If hardware lacks slot capacity, review whether the price tag of replacing smaller RAM slots with larger ones is more cost-effective than a complete server refresh.
Upgrades of server components can be viewed as wasted investment. But an upgrade may be your only option until the next financial budget year. With server upgrades, consider the following:
- an upgrade is cheaper, although beware paying a premium for rare legacy components;
- ensure that server hardware can support larger memory volumes; it may be unable to support the desired amount of RAM that your business demands; and
- if your hardware supports larger RAM volume, ensure that your OS is 64-bit for greater than 4 GB of memory.
Reconditioned servers. Consider procuring pre-owned or reconditioned servers. These servers may be one or two years old and have another two years of shelf life. Additionally, some leading server manufacturers have their own online marketplaces. Buying refurbished servers directly from a vendor offers the peace of mind that the server was rehabilitated correctly.
Decommissioned/failed projects. Evaluating your existing environment may highlight that certain decommissioned servers are sitting idle and no longer used for their original purpose. Consider reusing such unused server’s platforms to fulfil a new refresh requirement.
Virtualisation extends server life. Virtual machine workloads do not always demand up-to-date underlying hardware. Most VMs are heavily underused and don’t need the benefits of newer platforms and can run just fine on older hardware. In these situations, you can stretch out the underlying server assets until you have sufficient budget to purchase new hardware.
Functionality within most leading virtualisation technologies means you can be agile and exclude older underlying hardware from clusters of hosts that may have newer chipsets and tier-one virtualised workloads. Finally, high availability and disaster recovery failover capability of virtualised technologies minimise the risks of failure in running a VM on older hardware -- indeed, the risks may be fewer than if the platform were completely physical.
Roadmap. Even if you don’t get budget approval for a server refresh in the near term, continue planning for an eventual refresh. To better prepare, document your proposed technology roadmap with facts on what needs to be refreshed, along with the justification through collated gap analysis.
During the roadmap phase, prioritise by creating a master document to audit server inventory. Gathering audit inventory provides a reference point to use throughout the gap analysis stage. Additionally the audit inventory gives IT substantial documented proof that server refresh requests are valid. Consider performing inventory with the following high-level metrics:
- a complete server list of all servers that reside within an estate, along with associated name and business owner;
- information on key dates, such as when servers were originally purchased or from your procurement teams’ inventory;
- obtain the EOL dates from your server manufacturer’s website; most manufacturers have this information available. If not, ask the manufacturer directly; and
- finally, establish which servers warrant a higher-priority refresh.
Once the IT audit has been compiled, consult with business-side stakeholders on future budget considerations and on priorities that may not be visible to IT from a technology perspective. Examples include a planned business transformation activity or even sell off of certain parts of the organisation that would involve a reduction in overall hardware.
Further, if IT faces reluctant buy-in for server upgrades, the process itself becomes valuable. By building relationships with key stakeholders and application owners, IT gains trust, which enables data centre managers to articulate the benefits of a server refresh in terms beyond a budget figure.
Once you perform due diligence on your environment’s considerations, create a timeline for which servers will be refreshed and when. Plan and project manage the roadmap scope and structure the transition to be easy for key stakeholders to understand.
Look before you leap
Deciding when to upgrade your hardware or meet additional functional performance requirements with other refresh tactics is a critical but potentially costly decision to gain approval for. As you weigh server performance, business needs and other technology dictates against the cost of new hardware, document your process so you can make your case to key stakeholders. Even if you decide to delay your refresh or your business case is not approved, you have laid the groundwork for a hardware refresh down the line.
As you make these choices, keep business needs at the centre of the equation. Consider the cost of potential downtime against the cost of new servers. Consider the benefits of improved performance with the costs of installing new hardware, and consider the compatibility benefits of using a single vendor as well as the tradeoffs of lock-in or operating multiple platforms. Making these cost calculations at an earlier phase will pay dividends later on.
Daniel Eason is an infrastructure architect at a multinational company and is based in the UK.