Are you planning to deploy SAP HANA on public cloud? Are you looking to unbox the full potential of public cloud capabilities to deploy ERP Applications? Have you considered all the flexibility public cloud has to offer?
Enterprises have embraced public cloud for deploying critical applications like ERP with availability commitment up to 99.9%. Public cloud already brings down customers’ total cost of ownership. But following certain design principles can bring down your cost further.
Here we will discuss a few cloud cost optimization strategies for public cloud for SAP HANA deployment.
One size does not fit all: Public cloud offers wide range of HANA hardware starting from as small as 128 GB HANA Box to as much as 24 TB. But it is very important to do the sizing activity before deploying SAP HANA workload on public cloud. There are a lot of sizing and SAP HANA optimization guidelines available from public cloud providers and SAP to choose right size HANA Box based on User, Transaction etc. Just to accommodate a few hours spike of uses during month end or yearend is not a good idea. With the flexibility of public cloud it is possible to upgrade your HANA boxes to accommodate the spike uses.
Choose your hosting region cautiously: Public cloud providers like AWS and Azure have widespread availability across globe for HANA workloads. But cost of the services differs from one region to another. Putting your workload strategically in a region considering all other factors like latency, bandwidth etc. can reduce the cost significantly. So choosing a region is also strategic to bring down your cost.
Right buying options: Public cloud now comes in different buying options such as
- Reserve (1 year and 3 Year)
- Convertible (1 Year and 3 year)
Different options provide a lot of flexibility. If you are confident about your sizing and assumptions then buying reserve will save a lot of money for you. If you are using the instance for the PoC or Sandbox then doing a commitment for a lock-in period does not sound like a good plan. In that case, an organization can use on-demand instances.
Scale-up when in need: You are deploying SAP HANA but before fully moving to production you want to do a PoC to evaluate the business case and then eventually move to full production via route realization, testing and then deployment. So, it is not logical to provision full environment at the starting. Environment provision on need basis according to the deployment plan will allow full utilization of environments and also cut down the cost of redundant environment.
Choose new available hardware series rather than old series: New available series under same family in public cloud inventory is usually cheaper than older ones. Like the new series r5a.XXX available in the AWS is bit cheaper than its predecessor.
Stop Instances when not using: We don’t use non-Prod instances all the time. Assuming 100% utilization of systems and running them 24*7 is not an intelligent utilization of resources. These systems should be stopped after a certain time period or by end of the usage.
Trim the storage flab: Due to higher lead-time in procurement of hardware for on-premise setup, it was a practice to have some buffer during storage calculation. But in case of public cloud, resources are very easily scalable and can be bought on-demand to meet the need. Also in public cloud there are different types of storage class. Like AWS has Block Storage and Object Storage. Object storage is further divided into different classes. Classifying your data into categories and using correct type of storage for the data is key to reducing the cost.
Eye on data transfer: Data transfer on public cloud is a chargeable event. Where cloud vendors do not charge for ingress data, they certainly charge for egress and inter-region transfers. If data transfers can be measured previously and controlled by using monitoring mechanism, it helps in reducing the public cloud bills.
DR Strategy: Based upon your RPO/RTO need, there are a few possible scenarios of disaster recovery that can be designed for SAP HANA Workloads on public Cloud like Cold DR, Lean DR, Optimized DR, and Warm DR. It will bring down the cost significantly than choosing a traditional DR.
Once you have completely assessed your current environment and followed the design principles stated above, public cloud cost optimization is easily achievable in public cloud. All these principles are developed understanding the current cost and the changes that are critical in both estimating and delivering real cost saving. In future when you are onboarding public cloud, keep these points in mind to achieve expected benefits.