Before you can begin architecting your hybrid cloud, it is first necessary to assess your current situation and your future needs. The better you do at this, the better the eventual outcome is likely to be.
Do your best to assess the economics of all your IT operations - both on-premises and in the cloud. This will allow you to take the best advantage of public clouds and CSPs to satisfy your needs - even as those needs continue to evolve. My blog will help you answer the following questions:
- Which public clouds and CSPs is your business using? Which should you be using?
- Which of your applications and services are best suited for the cloud? Which applications should be on-premises?
- What new business initiatives are on your priority list, and how are those likely to translate into infrastructure, application, and service needs?
- What guidelines will you use to make hybrid cloud decisions?
- What are your primary hybrid cloud goals?
If you’re reading this report, it’s likely that your company today is not the company you want to be in the future. The whole purpose of digital transformation is to better prepare your business for success in the digital era. Because IT will play a crucial role in your success (or failure), you will need to figure out how to sustain the critical aspects of your current IT operations in a way that frees up resources to empower your business teams and enables your company to deliver new digital services to increase customer engagement, reduce business friction, and open new markets.
Assess the Current State of Your Operations
For most enterprises, assessing the entirety of your IT operations is, in itself, a daunting task. There are almost certainly activities taking place that you know little or nothing about. A regional sales team may have moved corporate data into AWS to run analytics, a skunkworks project may be using Information as a Service (IaaS) from a CSP, or your marketing team may have added new applications from multiple SaaS providers.
In some cases, the temptation not to care or get involved may be strong, but at a minimum you need to assess whether each unauthorized activity is putting important data at risk and violating regulatory requirements or corporate policies. You must also assess how much duplication and overlap exists. An obvious example is the hard-to-control spread of corporate files across services like Dropbox, Box, Google Drive, iCloud Drive, Microsoft OneDrive, and so on.
The first step is identifying all the locations where you have IT infrastructure, services, and data. This may include:
- Primary and secondary datacenters
- DR facilities
- Remote offices and branch offices
- Distribution centers, production facilities, and other edge locations
- Colocation facilities
- Major public clouds
- Smaller service providers including managed services
- SaaS providers
For each location where you own the infrastructure, you should ask:
- What resources (infrastructure and staff) do you have in the location?
- How current is the infrastructure?
- What percentage of the infrastructure is traditional/siloed? Virtualized? Private cloud?
- What’s the utilization rate of all infrastructure?
- Is physical space limited?
- How important is this location to the business?
- What part(s) of the business rely on this location?
- What does this location cost in terms of capital and operating expenses?
- How much staff time does it take to sustain this location?
For cloud providers, ask yourself a similar set of questions:
- Why are we using this provider?
- Do the services from this provider duplicate services running elsewhere?
- How important is this provider to the overall business?
- What part(s) of the business rely on this provider?
- What does this location cost in terms of operating expenses?
- How much staff time does it take to sustain this location?
With a complete list of locations in hand the next step is to identify and prioritize the important workloads running at each location. Sample questions include:
- What is the workload?
- Why is the workload running in this location?
- What resources is it consuming (computing, storage, network bandwidth)?
- How important is the workload to the business?
- Who depends on this workload?
- Who manages this workload and does it have any special or unique management or monitoring requirements?
- What other applications/services is this workload associated with/connected to?
- How is the workload protected? Is the associated data in multiple locations?
- What are the workload’s regulatory and security requirements?
- What’s the total IT budget commitment to this workload, including management costs?
Identifying and prioritizing all your important IT workloads will allow you to make smarter decisions, including any necessary trade-offs, as you plot your path forward.
Don’t Overlook Development and Test Resources
Be sure to include all IT resources dedicated to development work, including tools, repositories, build servers, and so on.
Assess Your Future Workload Needs
With a completed assessment of your current IT operations in hand, it’s time to polish up your crystal ball and start looking into the future. Your hybrid cloud strategy should address your foreseeable needs for the next three to five years, not just the applications and services that you’re already running.
Start with the low-hanging fruit:
- What new applications and services are due to come online over the next year?
- Are you planning to update, change, or expand any important business applications (ERP, CRM, SCM, and so on)?
- What’s your projected company headcount growth and how will you accommodate those new users (traditional desktops, virtual desktops, and applications)?
- Are there any seasonal workforce requirements?
- What’s happening with IT operations and development?
- Are there any changes planned to major infrastructure software like hypervisors?
- Are there any new management or security software coming online such as security information and event management (SIEM) software?
- Are there any changes/additions to development teams and tools?
Next, what new business initiatives or technology initiatives is your company engaged in or planning? It’s harder to be prescriptive here, but some likely technology initiatives include:
- Big data analytics
- Internet of Things
- Artificial intelligence
For each potential new workload, do your best to forecast:
- Is it mode 1 (traditional) or mode 2 (next-gen/cloud-native)?
- What IT resources will it require (infrastructure and staff)?
- What data and services will it depend on?
- Is it going to be on-premises, in the cloud, or a combination?
For example, for industrial IoT initiatives, sensors are likely to be on-premises (or at least not in the cloud). You may need to process the data near the point of collection, with some data transmitted to a corporate datacenter or the cloud for further analysis.
Creating a Workload Decision Matrix
With the information from your assessment in hand, you’ll begin to get a pretty good picture of what resources you have and where gaps and opportunities exist. This information all feeds into the process of architecting your hybrid cloud, as discussed in the next two chapters.
At this point in the assessment process, you have everything you need to create a well-informed decision matrix to help you determine where to locate each workload going forward and how to assess anticipated costs. Some sample questions to ask for each application include:
- Who uses this application? Is it customer-facing or used by company employees?
- Does the application depend on an ecosystem of other applications?
- Does the application have stringent compliance requirements?
- What is the impact to your business if the application is down or unreachable?
- What level of data protection and disaster recovery does it require?
- Does the application consume and release resources as needed, or is it persistent?
- Does the application consume more and more resources over time in a way that you can’t control or limit?
- Is the same or a similar service available in a SaaS model?
- Does the application scale vertically or horizontally?
- Can the application be containerized?
- What’s the I/O pattern? Steady or fluctuating? High or low?
- These factors tend to be interrelated. High I/O requirements or the need to run as part of an ecosystem will affect the cost of running an application.
Once you have the right cloud decision matrix for your organization, you can evaluate individual workloads and applications against it. Some applications will rank as prime candidates to move to the cloud if they aren’t there already; some applications will clearly not be well-suited to the cloud and should be sited on-premises; others may require a little work to make them cloud ready.
As you plan, keep in mind that on-premises and private cloud environments tend to be a better choice for the most critical and performance-sensitive applications, while the public cloud and cloud service providers are a good choice to meet the needs of cloud-native applications that scale elastically. (See Figure 1.)
Ultimately, your goal is to optimize application placement so you can focus time and financial resources on delivering the best application stack for each application or service to ensure the best customer experience.
Two Classes of Applications That Belong in the Public Cloud
As a rule, very few existing enterprise applications were engineered to be well-suited to the public cloud. It may be years before legacy applications evolve to be public cloud–ready, if they ever do. If you move an application that isn’t ready to the public cloud, you’ll likely find you’re burning money and not meeting your business needs.
The applications that do belong in the cloud often fall into two categories:
- Highly elastic applications
- New applications where you don’t know the resource demand
Applications that have a low ecosystem requirement and are very elastic (have highly variable resource requirements) are often perfect for the public cloud. They can get all the resources they need when they need them and release them when they don’t. Hosting a highly elastic application on-premises might mean having to provision a large amount of expensive infrastructure to accommodate occasional activity spikes.
When you deploy a new application, it can be hard to predict what the demand or the resource requirements are upfront. It often makes sense to put these applications in the cloud initially, so they can get the resources they need. Once resource needs are understood, applications can be moved on-premises if that makes practical sense based on resource demands and application design.
Many companies are initiating large numbers of development projects to create new applications and services as part of their digital transformation efforts. Only a handful of such projects are likely to succeed and deliver a high return on the initial investment. By developing and deploying those applications in the cloud, you can minimize upfront capital outlays. Applications that succeed may eventually be reevaluated to determine where they fit best; applications that don’t do well can simply fade away.
Figure 1. This figure shows some simple guidelines that often help when deciding whether a workload runs on-premises or in the cloud. Creating a decision matrix will help you make the best decisions about what to run on-premises and what to run in the cloud.
When assessing your current and future needs, it’s imperative to involve all internal IT customers and stakeholders. Your goal in bringing IT and business teams together is to get greater visibility and buy-in so you can move forward to create a hybrid cloud that meets the broadest range of foreseeable needs, while providing the flexibility to adapt quickly to changing market demands and unforeseen events.
Establish Your High-Level Hybrid Cloud Goals
As part of your assessment process, consider establishing three to five high-level, quantifiable objectives that you want your hybrid cloud to meet. These can help guide your planning process. Some examples might include:
- Consolidate less important mode 1 applications and reduce operating costs by 20%.
- Reduce provisioning time for virtual machines below 1 hour.
- Decrease time to market for new applications by 40%.
- Establish online disaster recovery for 95% of tier 2 applications.
- Containerize 50% of mode 2 applications.
It will be harder to determine if your hybrid cloud is a success if you don’t start with a clear set of objectives you want to achieve.
This blog provides detailed guidelines and suggestions for assessing your current and future needs and explains the benefits of creating a decision matrix to guide the process of siting applications going forward.
- Identify how much on-premises infrastructure you need.
- Determine which datacenters and other facilities are targets for modernization.
- Identify target applications and services for public clouds, CSPs, and SaaS providers.