WavebreakmediaMicro - Fotolia
Planning your data center budget seems to be getting harder. Gone are the days of just assuming some form of growth -- by resource, volume or power usage -- and allocating financial resources toward that area. Now, a whole raft of variables must be taken into account.
For example, will the next change to the data center take place through an expansion of the facility to accommodate more equipment? Or will it decrease in size due to greater use of intelligent virtualization or a move of at least some workloads to a public cloud platform?
A data center manager who plumps the budget for a larger facility only to discover that 50% of new workloads will now go to the cloud will see operating costs grow wildly. The decision to opt for a smaller facility -- only to find that performance, security or some other aspect of the cloud doesn't meet user needs -- could result in major constraints on the business.
It used to be that approximately 25% of a data center budget would be used for capital expenditures. But a cloud-first IT organization may have no capital expenditure because it is paying subscription costs only. This makes many of the old rules-of-thumb calculations go out the window. The best advice is to not get hung up on such metrics; spend your budget in a manner that makes the most business sense.
Evaluate design and upkeep concerns
Here are the main areas to evaluate when it comes to your data center budget:
Modularity. By focusing on modular systems -- server, storage, networking or support systems -- the resulting platform will be built with variability in mind. Resources that are not being used can be removed, reducing the need for power, cooling, licensing and admin costs. Likewise, if power and cooling systems are modular, they can be elastically managed to support the equipment in use.
Monitoring and automated systems. Simple desktop reporting portals are no longer a fit for today's complex platforms that need constant monitoring. Advanced systems that use machine learning and AI to predict where issues will arise and can fix them before they become problems should be considered a vital investment.
Updates, patches and changes. Planned downtime is no longer acceptable and unplanned downtime must be avoided at all costs. Therefore, processes and software that enable continuous delivery are a necessity. Orchestration software from the likes of Electric Cloud, HashiCorp or Stonebranch ensures that platforms maintain high levels of availability even when changes are undertaken.
Licensing and ongoing operating costs. Old-style license plus maintenance software agreements are running out of steam. More vendors are responding to customer needs by providing subscription-based pricing. Some of this will be via software as a service, which removes underlying stack costs from the data center budget, which may well be worth considering. Other vendors offer a subscription-type model for on-premises software, but there can be drawbacks. Per-core or virtual core models are useless when mass cloud adoption lies beneath, while per-transaction models encourage limited use of systems and therefore limited benefit as well. Look for unlimited use per user or role and make sure there are no hidden costs.
It is also important to track service contract costs. These contracts are seen by vendors as essentially ongoing recurring revenue as few companies review them to make sure they are getting value for their money. Review your current contract and ask yourself the following questions:
- Does the service contract still make sense? For example, are you looking at retiring the hardware within the next few months? If so, the contract may be superfluous.
- Is the service contract being used? Over the time that you have paid for it, have you had to use it? If not, it might be more cost-effective to pay when or if a problem occurs, rather than taking out what can be very expensive insurance.
- Is a third-party contract a better option? Although there are cowboys out there, there are plenty of well-regarded companies with considerable (independent) skills in maintaining your kit at a fraction of a vendor's service contract rates.
Funding requests should focus on three areas
When it comes to making the business case, remember that the business does not care about the technology itself. It only cares about the business outcomes. If a wet string attached to tin cans will work better than a hybrid cloud system, then string and cans it is.
Put your funding requests in terms the business can understand. Focus on these three areas:
- Risk. What impact does the requested change have on the business' risk profile?
- Cost. What will be the overall cost of carrying out the business process?
- Value. How will the budgeted expense help the business gain time-to-market benefits in improved or new products or services?
If you have a good enough understanding of the business, then you can balance these three concerns effectively. Some businesses may happily carry a little more risk if the cost benefits are strong enough. Others may pay more for risk to be driven down.
C-level executives will require pretty solid data center budget numbers from you. This may not be completely possible, as the pace of change in IT continues unabated. However, moving toward a subscription-based Opex approach means figures will be easier to come by and to predict over the period of the budget. Rather than estimating the hardware required in addition to maintenance, licenses and power, it becomes more a matter of the number of seats multiplied by the type of usage license. The number of users may change over time, but at least the incremental costs will be better known.
With data center technology changing, the way budgets are constructed and requested must also change. The more you can shift your budget to a private or public subscription approach, the better. This minimizes the hard-to-calculate costs on variable physical workloads and makes those estimates a much smaller proportion of the budget.