Whether you are making a wholesale transition of all your IT service delivery to the cloud or just one mission-critical application, you need a plan.
This is what many enterprises discover as they start on their cloud journey and for good reason.
Establishing a relationship with a cloud provider and consuming their services is just the tip of the iceberg when it comes to making sure you are maximizing cloud efficiency and scalability when moving an entire data center or even parts of it to the cloud.
As more companies embrace cloud technology, migration of IT assets and spending will follow suit.
A data center migration must start with very careful planning and a phased approach to execution, without causing a significant impact to business operations, service delivery, performance and data protection requirements.
Following these steps to strategically plan any cloud adoption will help you realize the cost, scalability, time to market, and security and compliance benefits that cloud, and in particular multicloud, can offer.
Planning and Design
This phase requires you to answer some hard questions about your existing data and application workloads, as well as what you plan to do in the future.
For example, do you need both cloud compute and storage resources for my data and applications?
How do you want to segment resources in terms of subnets, DHCP blocks, or even separate cloud billing accounts? Who do you want to access the cloud services and what will be the security and compliance policies around those users?
First, knowing where, how and by whom your IT services and applications are being used is a good place to begin. How do you want to segment resources in terms of subnets, DHCP blocks, or even separate cloud billing accounts? Who do you want to access the cloud services and what will be the security and compliance policies around those users?
First, knowing where, how and by whom your IT services and applications are being used is a good place to begin.
Many companies see cloud interconnectivity as an after-the-fact conundrum, but if you think interconnection-first, then you can build a higher performing, scalable, secure and cost-effective interconnected-cloud infrastructure.
Understanding if you are going to leverage existing MPLS or VPN corporate networks, direct, private cross connects or exchanges, or the internet for some or all of your cloud traffic should be looked at on a data and workload, case-by-case bases.
As does the proximity of your interconnection between clouds, data, applications and users to ensure the best performance. Multicloud interconnectivity is a growing requirement among enterprises, where applications are deployed across multiple clouds to decrease IT spending, increase the quality of service delivery, or for redundancy to ensure business continuity (BC) and disaster recovery (DR).
By 2020, 86% of companies will be interconnecting multiple clouds across multiple locations and over the next five years to realize a lower Capex.
Build, Categorize, Prioritize and Size
Once you have an interconnection strategy, build the cloud accounts, set up the networks and test the security. Ensure the environment is accessible, useable and available to all the systems and mechanisms you need to migrate workloads.
Be sure to organize applications and workloads into categories (e.g., by operating system, user type, data or application type, etc.) and manageable units of work.
Once you’ve completed your groupings, then you need to make some decisions about each grouping’s applicability and priority for cloud migration.
You’ll also need to evaluate your data and workloads based on a new type of usage-based sizing and scalability. What are the performance and availability requirements for this application?
What are your CPU and storage capacity requirements for each application instance? Given that you can scale on-demand in the cloud, make sure that you are managing and tracking instances carefully to that you are not overspending on instances that are no longer required or being used.
This is the “cloud” version of overprovisioning capacity.
Even this phase of cloud adoption has its own set of choices: Do you rebuild the servers, clone or containerize applications when migrating to the cloud?
Rebuilding provides the opportunity to start with a “clean slate” and is ideal for systems where deployments are already packaged or automated.
Cloning enables the migration of virtual machine images directly from existing infrastructure without need to perform complicated server rebuilds.
Also, leveraging the automation and orchestration capabilities of containers that break applications into pieces and networks between those pieces, could provide fast and flexible application deployment to the cloud.
Other considerations include understanding software and resource pricing and licensing terms, BC/DR service level agreements and workload compatibility within hybrid cloud infrastructures, just to name a few.
Article by Jim Poole, Equinix Blog Network