This is Part 2 of the Cost Modeling Your Cloud series. Part 1 is here.
One of my least favorite questions from potential suppliers is whether ServiceNow is CAPEX or OPEX driven, as if excelling in one forgives sins in another. My answer is always the same: we account for both.
Most of us know the old joke: lies, **** lies, and statistics. I would happily add case studies and TCO models to that list. The assumptions and data elements that feed them are contextual, and the vast majority of sales and marketing content completely ignores this context.
So how then does ServiceNow make rational decisions about what to buy when building our cloud? And how can you?
- Know your numbers
- Develop the model
- Be consistent
** Disclaimer: Math ahead. Actual costs and data points obfuscated for confidentiality. **
Develop the model
Each configuration you compare should have all of these elements. For example, if an App server has a CAPEX of $5,000 and will draw 150W, we would estimate OPEX at $4,320 (150W*$28.8). The TCO of $9,320 now allows you to compare the overall costs of devices that may be more power efficient but initially more costly or vice versa.
Then you simply divide to get the cost per performance unit or cost per customer. The $/perf for a $9,320 Server A with 5,555 TPS supporting 32 customers would be $1.68, and cost per customer would be $291.25.
Alone, these numbers are meaningless, but it becomes simple to evaluate new scenarios with this data.
- $12,000 server B with 8,888 TPS? $1.35/perf unit.
- Dropping Server A memory and power draw by 50% from 150W to 75W for Server C? TCO of $7,160 and $1.29/perf unit.
These all sound great, don't they? But now the fun begins and the model becomes more nuanced: Divide by customer.
- Server A & B both support 32 customers, so performance per customer is 174 TPS and 278 TPS (60% higher) respectively. However, cost per customer rises from $291 to $375 (29% higher), which may not be palatable for your business model.
- Server C can only support 16 customers, so performance per customer ends up at 347 TPS (roughly 2X) and $448 (54% higher).
- Server C also has the added benefit of reducing failure domain by half, which is not covered in this post.
Selections can then make informed choices depending on business priorities.
- Need to be cost optimized? Server A looks good.
- Want the best customer performance at any cost? Server C is your choice even though each server is cheaper than B and has less overall capability.
Here they are built into a simplified model:
Once you use your model a few times, you will quickly develop a sense for how a change will impact the model. Minor CAPEX increase with moderate OPEX decrease? Probably not much change. Newer parts at the same cost of the old one? Flat costs with nice performance improvements.
At ServiceNow, we optimize for the highest combined and individual performance per dollar possible, so we would probably go with Server B.
Be consistent
Next time we cover how to use this data to better manage your overall evaluation decisions.
