Cloud providers benefit from economies of scale that they can pass on to customers. This results in immediate cost savings, replacing the cost of buying and hosting hardware with a lower-cost alternative. However, Cloud hosting offers opportunities for wider commercial benefits.
With so many opportunities for cost savings, organisations should gain significant commercial benefits from moving to the Cloud. Undoubtedly many do, but this is not guaranteed. So, what are some of the risks that could mean cost savings are not realised?
It is striking that the level of due diligence performed by many customers when selecting us as a Cloud provider is far reduced from that previously undertaken for “conventional” hosting. In part this is positive – customers no longer challenge the minutiae because hosting has become a service, thus the focus must be on the service.
However, many organisations don’t perform the level due diligence on Cloud services they would have historically performed on hosting facilities. Given that Cloud is all about service, it’s vital to ensure the Cloud supplier has a track record of delivering services to the required standard. Otherwise, poor service will result in higher costs for the customer through increased management time and reduced system availability or performance.
Organisations may also see Cloud services as infallible. While Clouds are generally very reliable it is a mistake to overlook good I.T. practices on the assumption that failures won’t occur. In March 2021 Europe’s largest Cloud services provider, OVHCloud, suffered a fire that devasted one data centre and damaged a second. It was reported that OVHCloud’s affected data centre had neither a VESDA (very early smoke detection apparatus) system nor a fire suppression system.
Rather, staff relied on smoke detectors and fire extinguishers. If true, this was a poor design. The OVHCloud example demonstrates why customers should still perform some due diligence on the hosting facility even if the focus is on service. If the Cloud supplier can’t support that then are they are suitable host for your corporate data?
Netcraft estimated the OVHCloud fire caused 3.6 million websites across 464,000 distinct domains to be taken offline. Two weeks after the fire, 18% of IP addresses attributed to the affected data centre were still not online. Sports activity tracker Sportstracklive was down for over a week and for some time it was in doubt whether they could restore all historical data. Sportstracklive posted on Facebook:
“I know this is crazy, but we can't do anything about it. I could never imagine a Cloud provider could have such a disaster on a such secured site.”
It is vital that those hosting in the Cloud do imagine such disasters can happen and plan for service failures. Cloud does not obviate the need for backups and DR and ignoring such good practices will lead to significant costs when a failure occurs.
The cost of Cloud service failures can be high, potentially leading to business failure, and, realistically, will never be adequately compensated through service credits. Thus, organisations must choose suppliers where risks are minimised and solutions must be designed to mitigate the remaining risks. But it is still important to demand a strong SLA / KPI regime that is supported with service credits.
Some Cloud providers’ SLAs, and particularly those from global public Clouds, offer little real value to customers. They are “availability objectives”, often opaque and with no real consequences for failures.
That may be acceptable for small or non-critical systems where the customer can accept failures given the long-term low service cost. However, for critical systems or those that cannot easily be migrated to another provider this is an important issue.
A further potential cause of cost savings not being realised is that existing teams or services are not effectively re-deployed. Engaging a Cloud service will only deliver cost savings if higher cost services are replaced. This is easy to deliver and measure for the core hosting service. But where the Cloud includes additional services, e.g. provided via PaaS, it is essential to ensure the service they replace is terminated or the re-deployed. Failure to do so not only means the cost saving isn’t delivered but, worse, costs are duplicated.
Performance must also be considered when choosing a Cloud. The low cost of vCPUs is in part achieved through clever virtualisation technologies. Few systems run continuously at 100% CPU utilisation, thus the physical CPUs underpinning a Cloud can be oversubscribed with virtual CPUs.
When configured correctly, the total demand for compute resources won’t exceed the physical CPUs’ capacity and customers benefit from good performance at a reduced cost. Of course, performance may still be affected at busy times.
Some Cloud providers go further and provide yet more vCPUs by capping vCPU performance, i.e. the GHz of processing power available to each vCPU. This will always limit the performance of any system that isn’t bound by other bottlenecks, e.g. disk performance. Such performance limits may be mitigated by purchasing additional vCPUs but that will require additional Oracle licenses. Given the cost of Oracle licenses, the cost saved by using a lower cost Cloud with reduced performance is likely to be a false economy.
Such Oracle customers would be better served by hosting on a “hard partitioned” Cloud where CPU resources are not shared with other customers. The small increase in Cloud costs will be more than offset by the benefits of increased performance and the need for fewer Oracle licenses.
Finally, there are potential hidden costs when hosting in a Cloud. For Oracle users the largest potential hidden cost is for licensing. Compared to on-premise hosting or “hard partitioned” Clouds (such as Claremont Cloud) additional Oracle licenses are required on Amazon and Azure (the “ACE” Clouds or “soft partitioned” Clouds.
Twice as many Oracle licenses may be required on AWS and Azure compared to other Clouds. Furthermore, fewer vCPUs may be used with Standard Edition Oracle software on AWS and Azure. Not only are more licenses required, but performance may be limited. Given the cost of Oracle licensing, this additional cost and potential performance impact may outweigh the affected public Clouds’ cost savings.
Clearly, there are significant opportunities for saving costs by moving Oracle systems to Cloud hosting. However, careful due diligence and planning is required to ensure these cost savings are fully realised and unexpected additional costs are not introduced.
If you are looking for an Oracle partner who can help you with your Oracle support requirements, goes about it the right way, and can back up the talk, then contact us.
And if you would like to find out more about the Oracle Hosting with Claremont or have a question, you can email us at info@claremont.co.uk or phone us on +44 (0) 1483 549004.