Skip to main content

Hosting applications in the cloud rather than locally can save money… if you pay close attention. What are the best tactics for controlling cloud costs?

In the cloud, there are many sources of unproductive expenses. Some illustrations:

Unused resources

Active but unused virtual servers are a form of waste in the digital industry. This is the case, for example, of a server activated for a test phase, which is never put back to sleep, or of databases used by an application at a given time, and which are no longer linked to an instance afterwards. These “abandoned” databases on the servers can generate significant cumulative costs. They should not be overlooked. In the cloud world, the potential savings are considerable.

Oversized servers

A server with more allocated CPU resources than it actually needs, with an over-sized budget allocation, represents an illustration of misused spending in the cloud.

Poorly calibrated storage capacities

Most cloud providers offer expanded ranges of options with service offerings that grow in tiers. Typically, the speed of data export from a cloud service can vary considerably from one offering to another. In general, the most economical cloud offerings are for simple data storage. Indeed, archives or backups are not intended to be consulted frequently. Retaining cloud options that are not directly required for the business contributes to unnecessarily increasing the bill.

Unjustified data transfers

Most server leases include a charge for adding or removing data (egress) from the cloud, except in the case of data transfer within the same cloud. In some ways, transfer costs are inevitable, as is access to data hosted in the cloud. On the other hand, the additional costs incurred by unfounded or unnecessary access to data are superfluous.

Controlling cloud expenses

The right tools can combat these sometimes inconspicuous costs that plague cloud providers’ invoices.

This is already possible with the reporting and monitoring tools offered by default by all cloud providers. They generally offer fairly minimal control features, but let’s face it, a cloud provider isn’t looking to lower bills as a priority. For example, these tools do not usually offer to identify the origin of expense variations in cloud billings.

On the other hand, cloud management solutions designed by independent vendors can help identify unnecessary expenses within a cloud infrastructure. These solutions are aimed at the continuous optimization of cloud spending.

What actions to take to regain control of cloud spending?

Right-size the servers

Cost optimization starts with the careful tuning of server instances. Choosing the type of instance that best fits the actual workload is the first step. The functioning of a body with too few resources will be degraded in periods of high demand. However, the addition of unnecessary resources to the activity generates unnecessary costs. Thus, right-sizing – the work of adjusting instances to actual needs – requires a careful assessment of the proven workload required, and an ability to forecast future changes in needs.

Right-sizing requires the ability to anticipate changing needs.

Secondly, optimization requires identifying the instance best suited to the business requirements. Most cloud providers offer several dozen types of instances. In a way, right-sizing should remain a fairly small step to methodically align server resource allocation with actual business needs.

Cutting unused resources

Intimate knowledge of what cloud resources are actually being used is critical to managing cloud costs. However, this mapping of all the cloud services subscribed to is often neglected. Only this precise knowledge guarantees that all cloud resources can be identified and cut off when they are not needed. From this point of view, cloud monitoring solutions are valuable. They allow you to alert on anomalies: a database that is not connected to any application, a virtual server that is not requested during a certain period, etc.

Intimate knowledge of the cloud resources actually used is crucial

Of course, to avoid discovering unused resources, it would be best to never have created them. This is why it is appropriate to establish principles that describe the conditions under which managers should open – or not – a resource, or close those that are no longer in use.

Automatic size change

But beyond the right-sizing methodologies, there are solutions that can automatically increase or decrease the storage capacities. These automations make it possible to vary the cloud resources according to the real needs of the business. With these tools offered by most vendors, a default low usage setting can be implemented, with adaptations for occasional peaks in activity.

Data transfers require special attention.

Usually, data transfers trigger charges from the cloud service provider. However, the nature of the movements that trigger these billings depends on the offers. Knowing how to read the offers and contracts proposed is essential, especially the price lists for data transfer. For example, if a vendor charges foregress, you need to make sure that this does not affect the applications that need to access this data on a daily basis. Theegress costs that are imposed on data transfers to a locally hosted application can be extraordinarily high.

The benefits of cloud storage.

Cloud providers’ terms for storage offerings vary widely. It is recommended to identify the appropriate offer for each specific use, taking into account the frequency of access. Regardless of the offer chosen, knowing the time required to export the data in order to access it is fundamental. This allows you to be prepared for a disaster recovery scenario with full knowledge of the facts.

Discount offers adapted for certain uses.

Most providers offer virtual servers at discounted prices under certain conditions. For example, AWS offers “spot instances” that provide access to additional infrastructure at discounts of up to 90% compared to comparable infrastructure with on-demand access. In the case of a “low cost” infrastructure, permanent access to data may not be guaranteed. These instances are not relevant for applications with continuous workloads. On the other hand, an organization can realize significant savings by taking advantage of discounted instances for regular functions, such as data analysis, that do not require instant availability or high performance speed. Similarly, vendors offer more aggressive commercial terms for storage availability booked in advance, compared to the same servers available immediately.

Cloud cost management is a complex subject. No standardized scheme, or miracle tool, will solve all the subtle constraints encountered in integrating a cloud solution. On the contrary, controlling cloud spending requires customization and constant attention to the many sources of potential waste. The unproductive expenses of the cloud can be well taken care of.

Article published on Search Cloud Computing.