Optimierung der Cloud-Kosten: 6 "Negative" von Cloud Computing erklärt
3. Oktober 2019
- Geschrieben von: Simon Croyden
Every business decision comes with an element of risk. A consideration of the pros and cons and a plan for how to effectively monitor and adapt to change. Cloud-Migration is commonplace in the modern business environment. Not only for core operations but across a diverse portfolio of services.
We have already discussed in this series how the key challenges of enterprise cloud services have been addressed and resolved over time and with new developments in the technology. Now, the more pressing challenge is upholding an efficient and cost-effective service for the long term, once cloud migration has been successfully achieved.
Here are 6 reasons why cloud suddenly seems so hard and expensive. We consider these so-called “Negative” of cloud computing and offer solutions that will keep your cloud costs optimized.
AWS, Microsoft and Google all have different price models and, what’s more, they frequently change as the providers try to retain commercial advantages.
While cloud pricing can seem simple on the surface – a by minute or hourly cost for a cloud instance or a cost per GB-month for storage – the reality is that there is a dizzying array of options from which to choose.
Add to this the challenges of choosing IaaS or PaaS, and there are tens of thousands of pricing options across the three leading cloud providers. Instances can also have significant price differences based on the region in which they run. With IaaS, older versions of instance families can be more expensive than their replacements. Even instances with similar amounts of CPU power and memory can vary markedly based on other add-on characteristics.
Then there’s the workload issue to consider – PaaS can be cheaper until workloads become more and more persistent to a point where IaaS becomes a better option.
Organizations who are using cloud have to stay on their toes to make sure that they’re conforming to the current best deals. It’s not easy to achieve. Some tools only show billing for one provider. A full understanding of what your organization actually needs from cloud computing helps when regularly comparing and monitoring costs.
If you have chosen to use an IaaS model, you will need to decide which instances to provision as your engineers and IT staff are building and deploying applications. In many cases, your staff may be unfamiliar with the performance characteristics of the cloud instances or of the applications they’re deploying.
When migrating instances from on-premises infrastructure, IT staff and engineers may not know what the equivalent instance sizes would be. They will often take a “better safe than sorry” approach and select a larger size.
Once the infrastructure is overprovisioned, it rarely gets downsized. In addition, resource owners don’t have full visibility into cost implications.
The only way that you can know this is to have a full understanding of your performance and availability requirements.
Reviewing when cloud instances are and are not being used and making changes based on this performance data is one of the biggest savings that companies can make. Systems such as development or test can be spun down and up when needed, making for large savings. Organizations must monitor consistently to gather the information needed to make these decisions.
There is no useful middle ground in vendor billings. Most vendors provide a high level summary, while some go into granular detail. In both cases, organizations will find it difficult to make any actionable insights from billing data.
Cloud providers’ billing tools provide information about the assets rented from them. What they won’t tell you is exactly how they’re being used and won’t provide any insight, for example, on how a company could balance workloads.
A good comparison is with mobile phone bills. Billing data is much the same as what you would receive from your network provider – either a very high-level summary or the deep dive minute by minute breakdown of everything that’s been used and hence paid for.
The detail is all there, but you need to analyze it to get what you need from it. It’s difficult, time consuming, and often impossible to get insight on how your company is using the tools that run on cloud, including when and how intensively they are used.
Also, vendor bills don’t have any split between cost centers or other company cost charging, which may make cross-charging extremely complex.
Use the data from cloud computing vendors to get down to the details of what the company is paying for, then compare it with insights and data from how your organization utilizes cloud computing resources.
At the time of provisioning, resource owners often have little to no visibility into what their applications will cost in the cloud. The hourly cost of cloud instances can seem low, so they may not understand the full impact when they run instances for weeks, months, or years.
This limited visibility into costs can also be exacerbated during agile development processes when teams are automatically provisioning and tearing down deployments for development and QA.
If templates or automated scripts are used, instances can be repeatedly overprovisioned, resulting in a continual recurrence of wasted costs.
To reduce losses from overprovisioned instances, avoid using templates and instead opt for custom options. Once applications are running, resource owners should be able to access reporting that enables them to see the cost implications of potential overprovisioning.
Cloud computing brings the benefits of flexible and agile scaling but also the need to continually rightsize. Dynamic monitoring of cloud usage and provisioning of instances is necessary for organizations to address issues with wastage.
Cloud is a new way of doing IT and just using it as a replacement for an internal data center could be the wrong decision. Pre-planning before cloud migration is essential, but so is continuous monitoring and being open to change.
Once migration has taken place, the cost shock that companies can receive is often a useful driver for cost base analysis against re-architecting their current services to be designed with cloud in mind.
Without resource owners being able to have easy to access and understand billing they can’t understand their general cost benchmark and understand the drivers for an increase or decrease in cost.
Resource owners normally have a tolerance of a certain percent of cost increase/decrease but outside of that need to be alerted straight away to review and take the best course of action.
Optimization is an ongoing challenge in the enterprise. Even after waste is identified and resolved, the dynamic nature of cloud use means that waste reoccurs.
Deep dive usage and utilization data is critical to dynamically monitor and respond to waste.
For cloud governance teams to ensure cost-efficient cloud use, they need analytics tools that work across all of their cloud resource pools. They need to intelligently identify specific areas of waste and use these insights to collaborate with resource owners to take action.
We hope you agree that cloud computing is the way forward for your business. Reduced costs are a hopeful promise before migration, but the reality is that cloud services are often not used effectively due to a lack of proper monitoring. With the right knowledge about what aspects of your services to monitor and how your organization utilizes cloud deployments, you can mitigate the negatives of cloud computing and achieve cost efficiency.
In the next post, we’ll share with you the best ways to optimize cloud computing costs and give you practical pointers about how to reduce cloud spend over time. In the meantime, you can download our Smart Guide, Grundlegende Strategien zur Optimierung der Cloud-Computing-Kosten – Durchschneiden der Cloud-Komplexität to learn more about managing cloud services beyond migration.