Forgetfulness is the biggest cause of cloud price shocks

Procurement functions that cannot handle pay-as-you-go and daily payments are another hurdle, say web seminar panellists

The difficulties in ensuring that you're getting the best price for cloud services and the most common causes of large bills were among the topics discussed during a Computing web seminar this week.

Procuring public cloud services at the right price is made more difficult by the pace of change.

"The customer needs to evaluate cloud continuously," said John Abel, Oracle head of technology and cloud for UK, Ireland and Israel.

"The price you pay today needs to be constantly re-evaluated or you lose the power of cloud which is agility and choice."

Recent Computing research into the subject found that human error is the most common cause of "billshock", the unwelcome and unexpectedly high monthly invoices from cloud suppliers that many firms receive. A quarter of the 100 IT heads polled said this had happened recently to them.

Forty-six per cent of these put the cause down to people forgetting to turn off VMs and storage. This was exacerbated where firms have a high turnover of staff: someone can commission a load of services in the cloud and then leave the company, with no-one being any the wiser that the services are still running.

Underestimating future demand was another common issue. Some organisations allow for this, and for fluctuations in demand, by overprovisioning services. While this was lower on the list of causes among those polled, it is something that data centre group Equinix Services sees quite frequently, said managing director Michael Winterson.

"We've seen clients overprovisioning by four or six times. It's hard as a supplier to know if they made an intelligent choice around price because it's not a priority for some folks."

While suppliers may be able to advise on the right service levels and provide tools to allow customers to optimise usage ultimately "the client must take responsibility," he said. "I don't know if the vendor can create a perfect TCO model. We will be biased."

Procurement unprepared

It's more than just having someone monitor a console, Winterson went on, the setup of the back-office functions must be able to accommodate the pay-as-you-go and recurring payment models.

"The biggest post-honeymoon period discovery is the client saying that procurement can't handle daily contracting," he said "So make sure the back office can handle the idea of scalable services, otherwise you will generate internal cost structures in your own company,"

Adam Sewell, CIO at the Copyright Licensing Agency, advised people to evaluate their requirements carefully, in light of current and anticipated requirements and those of incoming legislation such as the GDPR.

"Make sure you get into bed with the right vendor, and make sure you understand the Ts&Cs", he said. "Get legal advice if you have to."

Claiming back service credits after the fact is very difficult Sewell went on: "I don't know anyone who's managed to do that."

The panel agreed that the hybrid model, with core IP kept on premises and the rest moving to public cloud will become the norm. However, even with the back office prepared, it can be difficult to monitor and manage in terms of costs, especially when a large number of services are involved. Abel said that pricing issues often occur at the edges of IaaS, PaaS and SaaS models and where the legacy and cloud models come together.

"You need to understand the cost matrix," Abel said. "So you would use an unmetered service for regular workloads and a metered services for bursting. Don't just look at it on terms of peak workloads; look at expected use and deploy an unmetered service for that, and for extreme use you go metered."

Sewell said that vendors could do more to provide real-time visibility into the costs being incurred, along with cost projections - especially for backup.

"It would be really useful if they showed you that," he said.