Pre-Migration Azure Contract Optimisation 

Contract optimisation is part of the FinOps Optimise Phase. It is typically where FinOps and Technical Experts work together with Procurement.  

In this case, Azure contract optimisation was part of a customer’s migration from on-premise to Microsoft Azure. 

It is important to seriously look at potential impacts of cloud optimisation scenarios before committing to long term tie-ins that commercial models such as Reserved Instances provide. What looks like a no-brainer on first glance may not be the right thing to do with some considerations.  

 Our experience is that in many cases vendor proposals attempt to over simplify a nuanced and complex issue and force customers to commit to a Reserved Instance strategy that appears the cheapest option at the outset but often leave customers committing to resources that may not be needed in the medium or longer term.  

 Classic Lift & Shift 

We engaged with a customer who was planning a typical lift and shift migration from on-premise to Microsoft Azure. This is a fairly common start for many enterprise cloud journeys.  

 The customer had an on-premise estate of 500 mostly Windows based Virtual Machines. A Microsoft Enterprise Agreement (‘EA’) proposal was received. The value of this proposal was based on a vanilla, like for like migration.  

We were asked to deploy step 5 of our Cloud Cost Saving Approach: Contract Optimisation. For this Azure Contract optimisation we first looked at the four options Microsoft offered the customer.  

 Four Enterprise Agreement Options 

 Microsoft offered four technical – commercial option combinations for the EA to the customer: 

1. Technical Like for Like Migration with Commercial OnDemand Model; 
2. Technical Right Sized Migration with Commercial OnDemand Model; 
3. Technical Right Sized Migration with Commercial OnDemand Model with Azure Hybrid Benefits as described below; 
4. Technical Right Sized Migration with Commercial 3–year Reserved Instance Model, again with Azure Hybrid Benefits. 

The customer was a heavy user of Windows Server and had multiple SQL Server workloads. Options 3 and 4 listed above highlighted the savings available should they utilise the Azure Hybrid Benefit model. Azure Hybrid Benefit is a license discounting mechanism for Microsoft and certain Linux products allowing customers to utilise their existing on-premise licenses on Azure. This can avoid customers duplicating their license spend in some cases. This license discounting model is unique to Azure and, if a customer is a heavy user of Microsoft server and database products, can make Azure an attractive financial option compared to other hyper scale cloud vendors.  

 Microsoft had structured the EA offering to highlight that option 4 was the most cost-effective and appropriate solution. As part of due diligence, the customer requested FinAnts to review the presented options. 

 Best Option? 

 We reviewed the migration approach and EA. We provided insight and recommendations on how best to engage with Microsoft and optimise the technical migration as well as the commercial model so as to optimise the economics of cloud adoption. 

 On paper and from an IaaS Cost perspective, Option 4 is far and away the most cost effective. It provides a discount in excess of 80% over Option 1 as well as a cost benefit of approximately 90% over on-premise deployment.  

 Cloud Benefits Not Considered 

 Firstly, we found that some key cloud technical behavioural benefits were not considered: 

1. Benefits of Server Consolidation Not Considered – A limited Workload analysis of existing VM’s had been performed. Server Utilization rates were reviewed and a significant number of VM’s with low utilization rates with CPU utilisations under 30% were identified. Consolidation of these machines was not considered in the proposed EA. 
2. Benefits of Scheduled Downtime Not Considered – There was little to no understanding of the VM’s roles. Cloud benefits from a different behaviour for Production vs. Non-Production were not taken into account. Scope power off VMs when not needed were not considered. Often, operating a VM on an OnDemand partially uptime model can be more cost effective than securing it with a Reserved Instance for 1 or 3 years. 
3. Benefits of Autoscaling Not Considered – Similar to Scheduled Downtime, Cloud benefits from auto scaling the system to meet workload peaks were not considered. Operating a number of smaller VM’s on demand and auto scaled can be more cost effective than securing it with an RI for 1 or 3 years. 

Other Cost Saving Scenarios Not Considered 

 Secondly, we found that some other key cost saving scenarios were not considered: 

1. Cost Arbitrage Between Regions Not Considered – Different cloud regions attract different cost for the same workloads. An analysis of the most cost-effective Azure region (or regions) to host the VM’s while still achieving technical criteria was not performed.
2. Containerisation Not Considered – Containerisation is an important consideration when developing a 3-year strategy and prior to committing to a long-term lock-in. Is the application that is hosted on the VM a candidate for containerisation and if so, then virtual machines can be retired and consolidated onto either Azure Kubernetes Service (AKS) or self-managed Kubernetes cluster, changing the demand pattern for RIs. 
3. Re-platforming Not Considered – It is important to consider the core architecture of your platforms when embarking on a migration. It is a good moment in time to make changes that are not otherwise cost justified. No such analysis, considering whether the VM is a candidate for re-platforming, was executed. Questions such as ‘can we utilize the migration to Azure as an opportunity move off Windows and onto Linux’  or ‘off SQL server to MySQL’ are important ones. 

 In Conclusion 

 We re-evaluated the proposal for a migration of the on-premise VM estate based on the above key findings and provided a clear and concise set of recommendations. This pre-migration Azure contract optimisation enabled the customer to better understand it’s Azure needs and to negotiate a more meaningful, more cost-efficient Enterprise Agreement with Microsoft.  

 This also put them on a solid cloud financial management footing from day one and ensures that FinOps principles are embedded within their culture as they embarked on their cloud journey. 

 Our experience is that in many cases vendor proposals attempt to overly simplify a nuanced and complex issue and force customers to commit to a Reserved Instance strategy that appears the cheapest option at the outset, but often leave customers committing to resources that may not be needed in the medium or longer term.  

 A factor driving the customers Azure migration was its need to respond more efficiently to opportunities the market offers; over subscribing at the outset is a classic case of Cloud Wastage that in this instance was prevented. 

 Interested in exploring how this can work for you? Contact us: 

• For Americas – [email protected] 
• For rest of world – [email protected]