Your search did not match any results.
We suggest you try the following to help find what you’re looking for:
By Sasha Banks-Louie | July 2020
Most cloud infrastructure vendors offer companies a stable and secure environment on which to build and run their applications without having to manage a data center, invest in hardware, or install and update software. While many cloud vendors provide the same types of services, they differ in how they charge for and deliver those services.
As your company evaluates cloud infrastructure vendors, consider these four hidden costs:
Among the most essential elements of cloud infrastructure, compute services provide companies with the raw CPUs/GPUs they need to process data and workloads. What makes the cost of compute so difficult to estimate is that companies need varying degrees of capacity at different times. Small variations in cloud vendors’ pricing models can mean huge differences in customer costs.
Bay Area adtech firm Widget learned that lesson the hard way. “We were spending enormous sums of money with Amazon Web Services on server capacity that we weren’t using,” says Vik Mehta, cloud evangelist at VastEdge, Widget’s software implementation partner.
To get truly “elastic” capacity, the pricing of cloud-based servers needs to reflect the exact capacity a company actually consumes, says Brent Juelich, director of cloud strategy and business development at Oracle. But many cloud vendors aren’t truly elastic and force you to buy in ‘shapes’—predetermined sets of CPU cores, memory, and storage. These shapes are often significantly larger than the actual consumption of the application, leading to waste. Oracle allows custom shapes on its E3 service, so you can specify exactly how many cores you need, eliminating waste.
With AWS, “you’re forced to decide up front on an elasticity window,” Mehta says, adding that Widget had wanted a lower capacity for day-to-day operations with an option to increase capacity for brief periods during testing. “But AWS didn’t allow that.”
Cloud storage is becoming increasingly attractive as companies collect ever-greater amounts of data. But selecting the right type of cloud storage can be tricky.
That’s because some vendors peddle bulk discounts for block and object storage, often locking companies into a type of storage capacity that doesn’t perform for their applications, resulting in significantly higher costs when they move to higher performance storage.
After hitting a wall in its on-premises data center, data conversion platform Adlib initially migrated to Google Cloud Platform. But communication latency delays between different data center locations could hit 6 milliseconds or more, requiring more local storage while data was being processed. While that might not sound like much, Adlib vice president of product engineering, Mike Grainge says this latency increased the amount of high-cost storage resources needed to meet specific time-sensitive processing service-level agreements for customers.
Last year, Adlib switched to Oracle Cloud Infrastructure and the latency has dropped to about 1.5 milliseconds. “When you’re talking about millions of customer documents, that significantly reduces the amount of temporary storage we need to meet our customers’ processing requirements,” Grainge says.
“We were spending enormous sums of money with Amazon Web Services on server capacity that we weren’t using.”
Cloud networks are the backbones of modern digital businesses. Just be aware that all such networks aren’t created equal.
One area to watch out for, warns Vinay Kumar, vice president of product management for Oracle Cloud Infrastructure, is the cost of moving data out of the cloud and onto the public internet, where it can be made available for analysis, reporting, and/or sharing. While all cloud providers let customers upload their data to their clouds for free, prices for downloading that data vary and can be quite expensive.
Most of the cloud providers built their pricing model around the notion of ‘data gravity,’ where the incentive is to aggregate more data into a specific cloud and export less, securing lock-in for the vendor. Notably, AWS, Azure, and Google have significant data export fees to drive their data gravity strategy. By comparison, Oracle’s model allows for fast, easy, and affordable data processing between cloud and customer.
While all cloud infrastructure vendors let their customers access online documentation and community forums for free, many charge hefty fees for hands-on, expert support—the kind you’ll need to fix a latency problem or network outage. “Companies should look for a vendor that provides one enterprise-level support package, which includes tiered escalations, a dedicated support manager, and a performance guarantee that is more than availability, but guarantees responsiveness, all as part of the service cost, not an add-on charge,” Kumar says.
When Widget began using AWS to run its mobile communications platform, the company was paying a few hundred dollars each month for infrastructure support, Mehta says. In less than a year, however, it was paying more than $20,000 per month as the company struggled with recurring latency problems, server outages, and a lack of dedicated support for its more than 16 open source technologies. “We had no choice but to shut everything down,” Mehta says.
Because Oracle provides premium support for free and offers at least 99% uptime for every workload a customer runs on its cloud infrastructure, hidden fees are no longer a concern for Widget. “We are now able to focus on our core business and not worry about escalating support costs,” says Mehta.