gCloud Genie Delineates Discounting

Recommended: Please read the Introduction here: Tech blogs with a twist!

It was a beautiful morning! The sun was shining brightly on the crisp November morning, streaming in through the windows of Zach Kennedy’s corner office overlooking the city. The hotshot Enterprise Architect was relaxed as he checked a few mails and completed a few actionables that had come his way.

All in all, life was good. A little over a year since CIB began their journey towards Digital Transformation through adoption of Google Cloud Technologies, everything was running smoothly. The cloud was purring like a kitten, humming along as it handed the ever increasing workloads that were now rapidly being tossed on to it. Even the audit from the National Reserve Bank, headed by Angela “Hell’s Angel” Ramirez, which many were dreading, had gone off smoothly. That had made waves in the industry as well, for many banks who had adopted a ‘wait and watch’ policy, now were jumping onto the cloud bandwagon, having seen CIB clear all the regulatory hurdles smoothly.

Zach looked back at what he had accomplished, and had every reason to feel proud!

There was a beep on his laptop. His digital friend, philosopher, savior, guide — gCloud Genie — appeared.

“Mornin’ genie”, said Zach

“Hey Good Morning Zach! How’re you today?”

“Top of the world!” Zach exulted!

“Great! I just wanted to discuss something rather exciting with you”

Zach loved these conversations. The Genie explained the features and benefits of various aspects of Google Cloud, and helped Zach navigate the rigors of transforming one of the largest banks in the country. They were always super exciting times!

“I’m all ears!” said Zach, a grin lighting up his chiseled face.

“Well, let’s play this out as a quiz”, said the Genie.

“Can you explain what CUDs, or Committed Use Discounts, are?”

“Sure! CUDs are basically a discount provided by Google in return for a commitment to use a particular resource for one or three years. These resources can be CPU, Memory, Local SSD Disks and GPUs. You basically commit a number of vCPUs or say a specific quantity of RAM or disk space. You pay a fixed monthly cost through the end of the commitment term, for the quantity of the selected resource, and in return for the steady revenue stream, Google discounts the said resource for you, through to the end of the term. CUDs vary by the resource type, but go up to 57% for most machine types”

“Correct, ten on ten”, grinned the Genie, a twinkle in his eyes!

“Now tell me one thing: What happens if you purchase a Standard CUD for an N2 Machine Type and a few months later, the App owner wants to switch to N2D machine type?”

Zach thought for a moment. “Well, since the commitments are N2 resource based, the commitments won’t carry forward to N2D. N2D will be billed at on demand rates, and the committed N2 bills will get accrued anyway. Other project owners will then be able to use the N2 commits by enabling discount sharing, but the fact of the matter is you’re gonna need another commit for N2D and takers across the enterprise for the original N2 commit.

“Correct”, said the Genie.

“Now the second question: Today, most project owners commit in only one region, which is in the same city as CIB’s own data centers. This becomes the CIB’s GCP production site. As we get into Disaster Recovery discussions, how would Standard CUDs work there?”

Suddenly the conundrum dawned on Zach. “Damn! I hadn’t thought of that before! Since Standard CUDs are regional in nature, standard CUDs applied in one region cannot carry forward to the other region. That means, the customer will be billed on demand when they invoke a DR! Besides, there’s no value in downsizing the production during DR invocation, since they’ll get billed for it through the commit anyway!”

“Which is fine, if you folks run out of the DR for a few days at a time, just for compliance. But now, I’m sure you know the National Reserve Bank is thinking seriously in the direction of mandating a 50–50 split between Production and DR runs for banks in the country! Which means you’ll have to run out of DR for months at a time, at on-demand billing because the CUDs are all applied in the production region!”

“You’re right man! Oh dear God, the CFO’s gonna kill me”, moaned Zach!

“No he won’t, he’s going to love you even more, and I’ll come to that later. For now, final question: Do your teams switch off UAT systems at night?”

“No they don’t. What’s the point? We moved the UAT estate under committed discounts, and we’ve already paid for the VMs for 3 years — so why switch them off?”

“Zach, you do know that CIB has recently appointed a Chief Sustainability Officer, don’t you? His charter is to drastically reduce your carbon footprint and make CIB carbon neutral over the next five years. He’s going to make a beeline for the IT Team, not to optimize costs, but to optimize the carbon footprint. The fact that you’ve already paid for your VMs and hence you don’t need to shut them down is not gonna cut any ice with him. You know that, don’t you?”

Zach buried his head in his hands! For the umpteenth time in the past year, he thought “What have I gotten myself into!”

But he knew there was a point that the Genie was driving towards — and from past experience, it usually turned out to be a bright shining light at the end of a very dark tunnel!

“All right, I know you well enough to understand that you’ve got something up your sleeve, Genie! Out with it! How’re you gonna get me out of this one?”

“Well Zach, let me introduce ‘Flexible CUDs’ to you! Google Cloud has recently launched ‘Flexible CUDs’ or ‘Spend based commitments’ as opposed to resource based commitments.”

“Okay, sounds exciting, carry on”, said Zach, sitting bolt upright!

“Oh it is! It is very exciting! Basically instead of saying ‘I hereby commit to using say xyz number of N2 VCPUs for the next three years’, you say ‘I hereby commit to using $abc per hour, for the next 3 years’.”

“Aha!” said Zach, the brilliance of the idea finally dawning upon him!

“Let me explain. Flexible CUDs give a fixed discount of 28% for one year, and 46% for three year commitments. Consider a scenario where you have multiple projects with VM instances across different machine families, located in multiple regions.

• Suppose you purchase a Compute Engine flexible commitment for this Cloud Billing account and commit to spend on resources whose on-demand prices are worth US$100, every hour, for a 3-year term.
• The US$100 amount becomes your hourly committed spend amount for the commitment term.
• You receive a 46% committed use discount on the three year US$100 commit and are charged an hourly commitment fee of US$54. Thus you get US$100 worth of eligible SKUs for US$54, every hour, throughout your commitment term.
• If you use less than $100 worth of eligible SKUs in a particular hour, well, you still get charged $54 because that’s your commitment to Google.
• If you use more than $100 worth of eligible SKUs in a particular hour, the additional usage beyond $100 would be billed at on demand rates.
• The other good thing is that Flexible CUDs operate at the billing account level by default, not the project level. So the discounts are applicable Enterprise-wide across CIB. This dramatically simplifies administration and helps you cover the maximum estate under Flexible CUD discounts.
• Finally, and this is the sweetest part — Flexible CUDs are not restricted to a particular region! They are shareable across regions!”

Zach sat back, stunned with the brilliance of this new feature from Google Cloud! He could see it clearly! The App teams would be thrilled at the flexibility and the CFO would be thrilled with the savings! It was truly the best of both worlds!

“All right Genie, things are crystal clear now! I now see why you asked me all those questions. Let’s replay them.

If an App owner wants to switch from N2 to N2D, with this Flexible CUD functionality, he can just switch, since this is not bound to any particular machine type.

We can easily run for six months from the primary region and six months from the DR region, since these Flexible CUDs are not restricted to a region.

Since Flexible CUDs are spend based rather than resource based, I can switch off a particular set of VMs (like UATs which run on N1 VMs) at night, and deliver additional compute power to the data warehousing team for their nightly batches (C2 VMs), both burning down the same Flexible CUD commitment even though they’re in different project and utilize different CPU families. Both teams are happy, and I’ll also become the chief sustainability officer’s golden-eyed boy!

Thanks Genie, this is brilliant!”

The Genie beamed. “Just a few points to note.

First: The Standard and Flexible CUDs co-exist. You can continue to use Standard CUDs for workloads where you don’t foresee a change in the machine families.

Second: There’s an order of application for discounts. First, Standard CUDs will be applied. The resources which spill over the Standard CUD commitment, will qualify for Flexible CUDs. After the application of Flexible CUDs, if there are further resources which spill over, then they’ll be billed at the hourly on demand rate, and qualify for Sustained Use Discounts (SUDs).

Third: Flexible CUDs are not available for all types of resources. Only specific resource types are eligible for Flexible CUDs.

Also, do take a look at discount exclusions once.”

“Gotcha! Now let me run down and tell everybody of this exciting new feature” exclaimed Zach!

“Off you go! Bye” the Genie waved and minimized himself as Zach excitedly rushed out of his office!