It's Not One and Done

Cloud Cost and Rightsizing
1

I’m thinking that the combination of changes in loading for scale up and scale down, plus frequent updates or changes to cloud provider offerings mean that cloud infrastructure cost optimization and rightsizing aren’t a one time thing - It needs to be done continuously to keep costs in check and avoid the “end-of-the-month” surprise.

The provider offerings instance and storage offerings change frequently. AWS, Azure, Google and other cloud infrastructure providers are continually updating and adjusting charges for sizing, services and reservations. Costs vary by location too. That seems a good reason to use CloudWisdom’s tools to get timely reports on cost changes, and perform deep “what-if” analysis to that tunes instance I/O, memory and network to get instance sizing “just right” for your organizations risk and performance goals.

The other axis for cost and rightsizing seems to be scaling. Will you need reservations for additional instances based on historical usage patterns? Are there random events that spike demand? If you can plan in advance, then reservations will save money. But done wrong, you end up spending money needlessly for capacity you’ll never use.

Similarly if you know you’ll need the capacity, but not when, you’ll have some decisions to make. Will you compromise performance or user experience in favor of cost and stick with what you’ll need 95% of the time? Should you increase instance size to handle the expected, but mostly unused capacity and eat the additional cost? Or will additional reserved instances be less expensive?

With CloudWisdom’s tools, get the answers you need to find the right solution for your organizations risk profile and budgets.

Then as you learn more about your workload usage patterns, and usage grows, you’ll need to periodically re-run your scenarios and see what works best.

In both cases, “One and Done” gets you a quick hit for optimizing your costs, but it’s only for a “snapshot” in time. It doesn’t solve the long term problems you’ll encounter over time.

What do you think?

1 Like
4

Hey Andy - I agree, and its part of why we recommend a staggered approach to right-sizing and purchase planning. It can be very difficult to right-size your entire infrastructure all in one go, different workloads have different constraints, or even ability to be changed.

5

Additionally, consistent cyclical review and refining of the migrated workloads is key. Along those lines I like the point mentioned in the Virtana Cloud Migration for Dummies guide that states: "Migration is an area where you shouldn’t be overly optimistic
about your team’s capabilities. Good team members can come
up on cloud skills, but you should bring in talent — employees,
consultants, and partners — to fill in the gaps if you want your
migration to move smoothly.
Consider building a matrix of responsibilities and fill in who on
your team will do each task. Any gaps should be filled before you
start migrating."

The Cloud journey is an opportunity for individual and team growth. Not just cost optimization through reduced utilization in compute resources (an area we often focus), but also human cycles, which can be more costly-especially at the onset.

2 Likes