No. 3: Cost Of Migration
Whether you're planning to migrate existing workloads to the cloud or use the cloud for net-new environments, there's a human cost in setting things up.
It's not trivial to move applications from your data center to the cloud. Analyze your IT landscape for suitable migration candidates; prepare those applications by either refactoring or rebuilding them; load those applications into the new environment; integrate them with the on-premises infrastructure; run both environments in parallel for a validation period; and sunset the on-premises environment. Each of those steps involves a number of cross-functional teams, so coordination is critical.
Even if you don't plan to move any existing applications to the cloud, you must still extend and migrate your existing architecture to the cloud. Consider identity management. New (internal-facing) systems must be aware of the user accessing the system without requiring an entirely new authentication scheme. This means that you will want to extend your identity infrastructure to the cloud to create a seamless experience for end users. To have a truly integrated portfolio -- regardless of where the application is hosted -- you must extend your infrastructure perimeter to the cloud. Your IT operations team will have to spend a fair amount of time planning and implementing this integration layer.
Make migration and integration a core part of your planning discussions. Look for obvious migration candidates, including lightly modified commercial packages such as Microsoft Exchange and SharePoint, service-oriented Web applications and applications with bursty, unpredictable usage. Don't waste time trying to retrofit monolithic commercial software, or systems with a web of connections to internal systems.
Establish a cohesive plan for how your core infrastructure components -- identity, networks, data and applications -- will be exposed to the cloud. Choose a non-mission-critical application as a trial balloon.
Look for a provider with a software catalog that lets you easily load your virtual machines and custom applications onto cloud servers. Look for guidance on all the ways you can (and can't!) create integration points between the cloud and your own data center. Work with a professional services group to plan the cutover procedures and minimize disruption to end users. All of these activities will reduce the toll on your staff while preventing trial-and-error migrations.
No. 4: Cost Of Maintenance
Estimates show that at least 70% of IT budgets go to maintenance of existing systems. That percentage may not change dramatically by using cloud technologies, and it could even go up if you don't have the automation to handle the influx of new resources. A successful cloud program will lead to requests for more environments (see point No. 1 above) and will support the construction of new types of applications. But can IT handle that?
What would happen if your organization doubled its server footprint tomorrow? Surveys show that server-to-admin ratios range from 50:1 to 300:1 in a typical enterprise data center. Management of those servers includes installing software, patching, performing security scans and integrating with networks and other systems.
Management becomes more daunting as servers get added -- and deleted -- each hour. By adding cloud servers, IT pros now have to maintain server templates, keep configuration management systems up to date and keep an elastic pool of servers secure and running smoothly. Given that server patching is still one of the most painful and time-consuming activities (because of testing and the inevitable reboots), adding more servers can cripple an organization that doesn't embrace automation.
The only way to truly succeed in the cloud on a large scale is to aggressively identify ways to automate server provisioning, scaling, patching, updating and retiring. Use commercial tools and scripting engines to eliminate manual tasks wherever possible.
Cloud providers offer a range of solutions. Some let you set global security, monitoring and usage policies that cascade to all users. Look for clouds that make it easy to scale servers (automatically) based on utilization, thus saving you the human effort of monitoring and manually resizing servers. Find a provider that makes it easy to schedule maintenance and perform bulk actions against sets of servers. See if you can offload time-consuming aspects of server management -- like patching -- to the cloud provider's managed services team. A cloud provider that embraces automation is a cloud provider that will keep your human cost under control.
I'll be presenting as part of the Platform-As-A-Service track at Cloud Connect in Chicago Oct. 21 to 23. I look forward to seeing you there.