As cloud adoption rates have increased and cloud models for enterprise IT mature, multicloud deployments have become more and more popular. They happen for a variety of reasons: some cloud platforms are better suited for specific applications, others may have security or compliance measures that are necessary. They might be located in different physical sites, fostering failover and disaster recovery or serving satellite markets. For many users, avoiding being locked in with a single vendor is huge for negotiation and data sovereignty.
Going multicloud isn’t a simple task, however, especially if you want to manage everything with a simple workflow. Here are the biggest stumbling blocks companies are facing when implementing multicloud.
Cloud sprawl can be a real problem even if you’re only using a single vendor. Once you add multiple IaaS and SaaS providers, costs can quickly spiral out of control. Creating a strict cloud purchasing and management process will help this; regular systems auditing by a dedicated staff member can go further.
We’ve even heard reports of one application or type of application being recreated in disparate clouds, resulting in overlapping service delivery and significant cost overrun. Make sure you have a cloud migration and management plan in place as well as managerial oversight for your entire cloud team before you start your multicloud experiment.
Every cloud provider has different ways to login and manage your virtual machines, along with their own set of processes for ordering, configuring storage/networking, setting up access control and identity management, and so forth. There are now a number of “single pane of glass” dashboard vendors, who offer a platform that can manage across multiple cloud providers. They aren’t yet perfect so you are likely to need supplemental and/or manual management tools for some processes like patching, security monitoring, or resource alerts.
Costs can pile up if you aren't using AWS for the right scenarios. Download this white paper to learn more.
Going hand-in-hand with different management processes, many organizations cite a lack of in-house cloud expertise as one of the biggest challenges when going multicloud. Because each cloud is different, it can take hours and hours of study to become certified and knowledgeable in their intricacies and advanced configurations.
MSPs can be a good way to fill in your staffing gaps for specific cloud platforms and services. A solid service provider can also help guide your workload planning to put applications and data on the cloud platform that maximizes performance and minimizes costs.
These are two of the primary drivers of multicloud as well as remaining two of its top obstacles. Legacy applications often work better with one cloud provider over another. You might need a provider that has optimized for your SAP or Oracle workloads, for example, or you might need to move from on-prem VMware to hosted VMware. New applications might be architected specifically for hyperscale, on the other hand. You need your entire environment to work in concert while still limiting the impact of migration if possible.
Meanwhile, your standards and policies for security, access, data archiving, eDiscovery, and architecture must be universally applied across various cloud providers. While steps have been taken towards interoperability — VMware has established itself on AWS, Azure can run containers, and you can export VMs to Open Virtualization Format — many challenges remain and manual enforcement of global policies might still be required.
Ensure all of your cloud and service providers meet your compliance standards before engaging with them. Sign your Business Associate agreements ahead of time. Different cloud providers take different approaches to securing the infrastructure stack, so be sure to obtain a detailed explanation of who is responsible for which piece of the infrastructure. Ask about compensation for a breach if the source falls under the provider’s purview.
The above problems with multicloud all trickle down to security, as well. Different portals and processes mean you may have limited visibility into the overall cloud environment unless you find and configure a universal monitoring solution (which may still have some holes). A coordinated response to a security breach can be difficult to push through as you struggle to work with a variety of platforms while under pressure.
Ultimately, many of these obstacles to multicloud adoption boil down to an increasing Total Cost of Ownership due to administrative overhead (more man hours spread across different platforms), new software required, and training.
It’s easy to get caught in a reactive IT management mode rather than staying proactive and searching for vulnerabilities, performance improvements, or new business initiatives.
One way to avoid staying on your heels is to have individuals in charge of each cloud platform, working under a cloud service manager who coordinates the overall multicloud approach. Another way is of course to contract with a knowledgable service provider who can architect and manage different cloud platforms on your behalf. In either case, careful planning is vital to avoid cost overruns, security holes, and management headaches.