Barry Whittle - November 06, 2019

Optimizing Cloud Costs when Migrating Microsoft Workloads

The business case for migrating Microsoft workloads to the cloud is undercut by indefensible migration costs. Build a business case, and measure success, with defensibility.

Moving Microsoft workloads to Azure seems a like-for-like move. Sister companies should, theoretically at least, be a path of least resistance for cloud migration. But migration is simply a tactic—one to be executed well or poorly.

Estimating cloud spend before and tracking and course-correcting cloud spend after migration are the key activities that optimize Microsoft workloads in the cloud.

 

Before migrating Microsoft workloads to cloud solutions

Best practice: Estimate workload spend by baselining targets.

Fail to plan, plan to fail.  Cloud promises agility and flexibility but, as spend is consumption-based, there is always some doubt on the size of the first cloud. IT leadership could hide behind that uncertainty, but that doesn’t absolve them from coming up with at least some level of an accurate estimate.

In the new IT operating model, IT leadership are entrusted with spend that could balloon with aggressive adoption.

IT leadership is on the hook to calculate and communicate those estimates.

Optimize VM spend by applying the pricing structure that reflects the use of cloud services. Azure optimizes its infrastructure by aligning customer workloads to the most appropriate service offering—apply the same for your own migration.

Define your operational footprint and estimated cloud bill by inventorying your assets including servers, VMs, databases, and storage.

For estimated cloud compute costs, identify operating systems, licensing that can stay current with software assurance, reserved instances applicability, VM uptime, location, and currency settings.

For estimated storage costs, aggregating the storage costs of all VMs by calculating the monthly storage cost for all disks attached to specific machines.

Right-size VMs: Adopt a pricing structure that reflects your usage

Cloud providers optimize their own internal infrastructure by aligning customer workloads to the most appropriate service offering—optimization feeds cloud provider profits. Leverage their internal optimization efforts by right-sizing your VMs.  

For Azure, map your VM usage to the appropriate Azure VM:

Type

Details

Use

General purpose

(B1S)

Balanced CPU-to-memory.

Good for testing and development, small- to medium-size databases, low- to medium-volume traffic web servers.

Compute-optimized (Fsv2)

High CPU-to-memory.

Good for a medium-volume traffic web server, network appliances, batch processes, app servers.

Memory-optimized

(Ev3)

High memory-to-CPU.

Good for relational databases, medium- to large-size cache, in-memory analytics.

Storage optimized

High disk throughput and IO.

Good for big data, SQL and NoSQL databases.

GPU optimized

Specialized VMs. Single or multiple GPUs.

Heavy graphics and video editing.

High performance

Fastest and most powerful CPU. VMs with optional high-throughput network interfaces (RDMA)

Critical high-performance apps.

 

For AWS, map your VMs to the appropriate EC2 instance type:

Type

Details

Use

General purpose

Balanced CPU-to-memory

Applications that use these resources in equal proportions such as web servers and code repositories

Compute-optimized

High CPU-to-memory.

Compute bound applications that benefit from high-performance processors

Memory-optimized

High memory-to-CPU.

Fast performance for workloads that process large data sets in memory

Accelerated Computing

General purpose GPU instances

Floating point number calculations, graphics processing, or data pattern matching

Storage optimized

High disk throughput and IO.

High, sequential read and write access to very large data sets on local storage

 

Select the right storage: Align services to your usage

Every interaction with Azure and AWS storage has a cost. Not all touch-points are charged (and most are nominal). Azure Blob Storage service is tiered (Hot, Cool, Archive) defined by redundancy (locally redundant, zone-redundant, geo-redundant), and available with different storage accounts (GP v2, GP, v1, Blob). AWS S3 has similar storage selections.

Find the right combination of AWS or Azure service by understanding how you manage storage.

Lock in Hybrid licensing benefits: Software assurance (SA) carries over to Azure

The migration costs of Windows workloads to Azure do not include licensing purchases when covered under SA. Which seems like a win.  However, this assumes that organizations will not wish to upgrade legacy on-prem solutions once they start using cloud services.  Not surprisingly, AWS identifies this as a reason to pause on migrating to Azure—the benefits of SA diminish if you know you are going to upgrade from SQL Server 2008 to SQL Server 2017. 

Purchase reserved VM instances: Buy in bulk and save

Buy in bulk and save. Prepayment locks in discounts to reduce the cost of your VMs, SQL databases, and other services. Of course, a discount only comes to pass when pre-purchased resources are fully used—buy, and consume, smartly.

Drive accountability: Aggregate cloud spend across subscriptions

Cloud procurement doesn’t have to be centralized. Corporate IT has a lock for on-prem IT spend when purchasing and provisioning has to go through them. Cloud solutions have only a financial stage-gate—one which anyone with a corporate credit card can get through. This freedom silos cloud spend. When anyone can buy a cloud solution no one is held accountable for the aggregated spend. 

View cloud spend across departments and resource groups with Azure Cost Management, AWS Cost Management,  or a purpose-built cloud cost management solution.

 

READ: Why you need a cloud spend and optimization solution

 

 

After migrating Microsoft workloads to cloud solutions

 

Monitor workloads: Build a baseline of costs post-migration

An annual budget for cloud must be based on more than estimates. Captured cloud spend is the data point that ties budget planning to reality—and validates the business case for moving to Azure or AWS.

Monitor resource utilization: Move migrated resources to different workloads

Tolerable in an on-prem world may be prohibitively expensive with Azure or AWS—which is a function of behavior over anything nefarious with price gouging.

VM and EC2 billing happens when it's allocated and stops after its deallocated. That gives many opportunities to optimize spend through operational decisions (e.g., downgrade resources outside peak times with autoscaling).

Shift to another type over time, based on usage, costs, and shifting business requirements.

 

Summary

Applying cloud migration best practices doesn’t remove the complexity of cloud adoption.

Taking recommended steps before and after a migration increases your odds of success—but “best practices” do not offer a frictionless path to the cloud. What lands in one organization will suffer organ rejection in another.

Succeed with your Microsoft workload migration by following a playbook of migration best practice, while acknowledging the tactics that will work in your organization’s culture. 

 

Download: 5 Steps to Prepare Your Cloud Migration

Our site uses cookies. By continuing to use our site, you are agreeing to our cookie policy.

Accept & Close