PoV On Cloud Optimisation Approach
The global public cloud infrastructure is expected to grow to 35% or $120 billion in 2021. While this is a positive development that will benefit businesses, it also brings with it a set of new challenges. Cost optimization tops the list. Cloud brings with it tremendous opportunities for businesses, such as increased agility, lower total cost of ownership, accelerated innovation, among others. But we often hear that switching to a cloud-based architecture leads to businesses overspending in the cloud, with a double figure percentage of investment on unused services, and on more resources than required.
Organizations shifting from on-prem/private cloud to public cloud providers can increase workload utilization rates while simultaneously reducing their overall costs. One way is leverage pay-as-you-go model more effectively that provides advantages over traditional up-front hardware purchases. Cloud providers can help businesses deliver value only when used correctly.
Here we present the NexTurn’s point of view for running a cost-efficient workload on cloud.
Our Cloud Optimization Pillars
Right sizing is the most effective way to control cloud costs. Right sizing involves understanding the true performance needs of the resources and sizing them to be the most cost effective. Right sizing can be achieved by continuously analyzing instance performance and usage needs and patterns. Key factors to consider
- Understanding the application workload model (Traffic patterns, Concurrency, Seasonality) Application type (Web request, Database/Table, Batch Process)
- What are the system requirements? (CPU threads, Memory Constraints, Disk & Network) Availability requirements? Avoid single point of failure
- Production vs Nonproduction workloads
Elasticity involves being able to scale workloads to match the demand. This is important since it ensures that you are paying for only what is used. Cloud providers offers several services that offer elasticity such as autoscaling.
One other option to considered as part of the system design is to use managed services wherever possible.
- Managed services operate at cloud scale, they can offer a lower cost per transaction or service
- It removes the operational burden of maintaining servers for tasks such as sending email or managing databases
The third pillar involves leveraging a right pricing model for specific usecases. Most workloads start off as on-demand workloads, meaning you pay for what you need at a moment in time. There are a couple other pricing constructs like reserved instances and Spot that can give you more discounts based on how well you know or design your environment.
4. RIGHT STORAGE CLASSStorage is the second or third biggest component on most of the customers cloud bills. Cloud providers offer many types of storage based on your application’s needs, but it is important to leverage the right class of storage to be the most cost efficient.
5. GOVERNANCEUnderstanding your business’s cost drivers is critical to managing your expenditure and identifying cost-reduction opportunities. Businesses typically operate multiple systems managed by multiple teams. These teams can each have its own revenue stream. The capability to attribute resource costs to the systems, individual business units or product owners drives efficient usage behavior and helps reduce waste.
These are the key factors to consider for improving your usage and expenditure awareness:
- Tagging
- Visibility
- Resource lifecycle tracking
- Identifying stakeholders and attributing cost.
Tagging also has other benefits as mentioned below.
- Automation (autoscaling, scheduling)
- Control & compliance (IAM policies)
- Cost allocation (reporting & chargebacks)
It’s very important that organizations have deeper visibility into their cloud costs which will help them identify opportunities for savings.
Cost optimization requires a granular understanding of the breakdown in spend, Ability to model and forecast future spend, having sufficient mechanisms in place to align cost and usage to business
objectives, cloud providers provide a suite of reports and tools estimate, monitor, plan, notify, report on, and analyze your cloud spend.
We can save 100% cost by not running services that you don’t need. Don’t stifle innovation by discouraging developers from experimenting but enforce hygiene for shutting unused resources off. Use monitoring, tooling, and tags to clearly identify workloads by dev/tribe and set/enforce policies for resource utilization. Bottomline is don’t leave things running that should be turned off.
Explore using serverless first for new workloads, then fall back to managed services, then containers on compute. The overhead you save in managing infrastructure can be used to innovate better products for your customers or extend your runway.
For any existing workloads running on cloud instances, explore architectural improvements first and ensure instances are right-sized. If no fixes are viable or near-term, then pursue reservations. However, don’t wait 5-6mo to pursue Savings Plans/RI’s; you’ll break even on most reservations in that time period.
Once architectural optimizations have been made and reservations are in place, then pursue private pricing where it makes sense. Private pricing is confidential and contracted on a case-by-case basis. It can be contracted at the service level or at the payer level. Private pricing requires a usage-based or spend-based commitment to one or more services for 1yr or more.