Public cloud services from AWS, GCP and MS Azure have allowed businesses to rapidly deploy applications at low costs while enjoying high degrees of flexibility and scalability. However, they’ve also given IT leaders (and their CFOs) massive headaches due to over paying for services they never use. A common complaint is that public cloud invoices for services IT shops never use and make it too easy to add ancillary services that are wasteful. Or their invoices don’t supply the detailed project-specific resource usage that busy IT Leaders need to manage their infrastructure. So not only are IT leaders being over charged, they feel out of CONTROL. If your monthly public bill leaves you wondering why your cloud infrastructure costs are so high or feel your infrastructure is getting away from you, read on to learn about some common “cloud gotchas” to avoid. Finding the right cloud provider is not as difficult as it seems.
- Not Managing Your Containers or Instances
With an a la carte menu, it’s easy to order too much infrastructure and difficult to manage all the service features. Here are some common mistakes that will result in a big cloud bill:
- Oversized Instances – Most common is IT staff coming from capital acquisition or on premises model are used to buying extra capacity up front and waiting to use it. So when they get to the cloud, they are allocating instances for future or “just in case” just like before. This is just like burning money. Cloud is scalable; you should be able to easily spin up additional resources via a self-service console.
- Unused Containers or Instances – It sounds obvious, but losing track of your instances or old containers, particularly those used for short-term initiatives, such as test/dev, and letting them run when your done, just racks up unnecessary costs.
- Overtime Instances – Similarly, running an instance 24/7 when it has no use outside of business hours will needlessly pad your monthly bill.
- Mismatched Instances – Using the wrong instance for your need also can cost you. Over or under provisioning is common. Instance types optimized for storage, computing or memory are more expensive than general purpose versions.
- Selecting the Wrong Instance Billing Model
Choosing the wrong billing model for your instance also can adversely impact your costs. AWS, for example, offers a few different instance billing models:
- On-demand Instances – Just like paying for expedited shipping when the standard arrives in the same window. This model allows you to buy instances at a fixed rate per hour with no conditions. Using this model over a sustained period can lead to higher monthly costs.
- Reserved Instances – This model offers hourly discounts by committing to long-term purchases. This gives you lower pricing, but it’s only good if you’re planning to use your instance for at least a year. This is similar to managed services model and provides for a good relationship between both the provider and the customer. It's a sustainable win win for both parties.
- Spot Instances – This model lets you bid the price for the instance you want, so you can get bargain basement pricing in non-peak time. But like the stock market, the going rate may not always be in your favor, rocketing past rack rates. Plus, you can lose instances if the price shoots higher than your maximum bid, so it’s risky to use for customer-facing apps.
- Thinking Cloud Services are Managed Services
Assuming that migrating to public cloud will reduce the burden on your IT teams of day-to-maintenance and systems engineering tasks is a huge mistake. Public cloud is an IaaS platform that requires ongoing maintenance and troubleshooting. Make sure you know what your CSP will do and what’s in your court. Take AWS, for example. Its Shared Responsibility Model states that “the customer assumes responsibility and management of the guest operating system (including update and security patches), other associated application software as well as the configuration of the AWS provided security group firewall. AWS operates, manages and controls the components from the host operating system and virtualization layer down to the physical security of the facilities in which the service operates.” That means AWS will not maintain your cloud nor replace an instance in the event of a failure. While AWS provides services such as Auto Scaling Groups to help your team react to incidents more quickly, it will not reimburse you for service disruptions. CIOs need to budget for their IT teams to handle maintenance requests and instance failures.
4. Thinking Hyper Scalers Protects Your Data
With the massive ransomware attacks coming from Russia and China affecting thousands of USA-based companies, Amazon has updated their shared responsibility model (image below). The data security, which is all that really matters since it's the business IP, is 100% in the customers hands. Who cares about the bottom half which can all be reinstalled from source. Not true with the customer data. All that matters is the top of this table but unfortunately you are on your own dealing with a 3rd party who isn't responsible per their own published policy who are right in the middle of everything. Does that sound risky to you? You bet it is.
AMAZON SHARED RESPONSIBLITY - YOUR DATA IS NOT THEIR RESPONSIBLITY
Switch to Enzu to Avoid Wasting Budget and Gain Peace of Mind
Tired of spiraling cloud infrastructure costs? Enzu’s TruCloud self-service management console puts IT Leaders back in control of their budgets and data. We can help you architect a flexible, secure and affordable cloud strategy, including private, hybrid and multicloud environments. We deliver computing, colocation and connectivity services to enterprises worldwide that need to source affordable, large-scale, enterprise-class computing infrastructure on-demand. And our transparent pricing ensures that you get the best cloud for your unique business requirements with no hidden costs and no surprises.
Want Transparent Cloud Pricing?