How to avoid unexpected costs when migrating to the cloud | NTT DATA

Tue, 23 February 2021

How to avoid unexpected costs when migrating to the cloud

CTOs and IT teams love the public cloud (IaaS and PaaS) – because of it’s elasticity, feature-set, ease of use and flexibility that make their lives easier. Plus they get to the tick the “cloud native” box which looks cool.

What doesn’t make their CFO’s happy is unplanned or unexpected costs and budget overruns.

Many IT leaders are chasing promised savings as they migrate application and data workloads to the public cloud - but are they getting the savings they expect?

At the push of a button, people want to be able to move from legacy on-premises software to the cloud – but that is still just a dream and not what happens in reality. Many IT leaders and buyers are disappointed at the actual reality that turns out to be a painfully slow migration process, often with an unexpectedly large invoice at the end.

Finance teams need training to support IT leaders, to help them plan and forecast in a cloud-native world. So what are the three key focus areas?

  1. Cost models
    Every cloud workload migration should come with a cost model based on “full-lifecycle” total cost of ownership (TCO) metrics, as with any other IT workload. As part of this, there needs to be an “as-is/on-premises” versus “in the cloud” comparison – as far as possible using the same cost-drivers and real data.

  2. Consumption based financials
    The days of raising and approving large purchase orders (POs) for procuring IT assets are distant history. In the new world of cloud computing – only having strong IT-Finance collaboration ensures robust budgeting, chargeback, and resource accounting is adopted.

    As part of this, the IT budgeting process has to be re-aligned to mirror the shift from large cyclical capital expenditure (CapEx) spend to ongoing operating expense (OpEx) spend.

  3. Existing fixed costs
    On-premises costs won’t magically vanish after cloud migration – no it won’t. It’s necessary to budget for some retained workloads on-premises and to plan how to manage the disposal, resale, and reuse of any new kit that still has usable life.

What to expect

It’s important to be wary of a fixed fee after a set migration period. It’s better to tie the migration period to actual workload units, not timescales in months.

Payments should be based on actual and fully operational migrated workloads, not the pre-deal estimates on which your minimum spend commitment is based. Look at a mix of ‘compute-units tiering’ and a ‘volume X stock keeping unit (SKU) based model’ with a charging formula and rate-card – not a fixed price.

Equally, be sure to hold your hyper-scaler to account on any post-migration value and cost benefit claims made in the sales cycle.

It’s also worth taking time to get to grips with the platform’s pricing model and how your workloads drive the costs you’ll be charged for – do note that the pricing models are in constant flux.

Tagging

There are several ways you can further ease the process of migrating to the cloud. “Tagging” helps users organise and make sense of cloud data. It is surprisingly easy to tag cloud resources by assigning metadata values to simple things like name of application, project or service, user group, cost centre, instance type and workload group.

Tagging is hugely powerful, but it requires an enterprise-wide naming-convention that is adopted at scale. It is the cornerstone to enable budgeting, billing alerts, usage monitoring and the ability to spot and suspend orphan services/workloads before you get an unexpected bill.

Choosing services

As your cloud strategy matures, you’ll need to make decisions about which mix of services work best for you.

For pricing, there are three options: on demand, spot or reserved. On demand requires no commitment. It costs the most but offers the best service. Spot provides the highest level of savings, but it is unpredictable and unsuitable for long-term projects. Reserved requires a one-year or three-year commitment in exchange for discounted service.

You’ll also need to decide on storage type. File storage is the simplest form of storage, and it works best for smaller data sets. Block storage stores volumes filled with files that have been split into equal-sized data chunks. And object-based storage stores data in isolated containers known as objects.

There are other considerations are connectivity, availability, etc. where the hyper-scalers all offer services unique to each one of them. I am not going to even try and cover this higher value services here related to databases, AI/ML, big data, edge computing, content delivery and DevOps.

“Infrastructure as code” and auto-scaling

“Infrastructure as code” enables you to manage your IT infrastructure using configuration files. This increases speed, consistency and accountability. “Auto-scaling” is a way to automatically allocate or remove computing resources.

These are both useful features and the ability to automate them takes a lot of drudgery out of IT Operations. However, without adequate control and monitoring, these nifty features can lead to unforeseen and sudden large increases in your monthly bill.

Expertise

Experts for hybrid and multi cloud are hard to come by. Finding a good partner with real expertise in hybrid and multi cloud is key. Equally important is that this should be someone who also understand legacy IT and its modernisation.

With the right technical expertise, cloud-native tooling and automation can maximise the pace and scale of adoption for your enterprise. A credible systems-integrator (SI) will also be comfortable signing up to a predictable cost model based on payment for each application or workload migrated successfully.

Right now, there are lots of cash-strapped firms. Many public cloud solutions providers are offering ‘migrate and run’ managed services with all upfront migration costs being amortised into a minimum term OpEx fee based on the number, size and complexity of migrated workloads.

Keeping the CFO happy

Finally, there are a few key metrics that can are critical to a public cloud commercial and financial construct – these are: 

  • Transparency is crucial; there should be a clear correlation between costs and types of services used.
  • Predictability of cost increases or decreases; with well-understood and quantified cost drivers. 
  • Independence; the buyer seamlessly by able to easily flex capacity, add/drop features, and dial up/down charges in line with business needs, changes and cycles.
  • Pricing and charges should also trend with current market pricing to ensure market-competitiveness – this is always downwards by the way. 
  • Verifiable cost savings; going cloud native has to be cheaper than staying on-premises on a ‘like-for-like’ basis.

If these criteria are fulfilled, you are very likely to reap the benefits of the cloud without the burden of an unexpectedly large bill.


我々の強み

How can we help you

Get in touch