Over the past several years, cloud computing has been on the forefront of various workloads and applications across businesses. It’s not long ago that most data centers were owned on-premise, co-located, or managed by third-party vendors, and it could take weeks if not months to create new servers—this has all changed. We are now witnessing a trend inside multiple organizations when choosing multiple cloud vendors, instead of just one.
Multi-Cloud refers to a strategy where an organization uses services from multiple cloud vendors. This provides a wide variety of options to present the best-suited and effective response to any business tasks at hand. The Multi-Cloud strategy has been slowly trending in advertising and marketing expedient, and it will become a major topic in the coming months. Gartner states that “a Multi-Cloud strategy will become the common strategy for 70% of enterprises by 2019, up from less than 10% today.” 
There are multiple reasons why Multi-Cloud is beneficial for organizations. Using a variety of vendors gives access to a wider range of services provided, compared to using a single vendor. Another reason is to support geo-replication applications — —like e-commerce organizations use to serve a larger market. Depending on the developers, or the project at hand, it can be both cost- and time-effective to use a cloud vendor that is known by the specific developers.
Additionally, the strategy of not wanting to be locked into a single cloud vendor. As the cloud vendors grow and the services they provide increase with that growth, organizations are not willing to put all of their eggs into one basket. Furthermore, reducing the vendor lock-in allows flexibility to move between the cloud vendors that best suit the future of the project. Gartner expects multiprovider IaaS/PaaS strategies to become the defacto standard. 
Although there are plenty of examples of why Multi-Cloud is becoming a best practice scenario, but it also has its downsides. Managing complex Multi-Cloud environments could be very costly, as it requires a multifold of resources to maintain and understand a multi-vendor approach. A major factor would be the upkeep of employees to stay relevant on the constantly changing cloud vendors. Hiring, training, and retaining a team across multiple cloud platforms could be expensive. Also maintaining a relationship with the top cloud vendors would provide its own challenges; for e.g. number of certified solution architects and helping tier.
There is also the issue of hidden cost, as when dealing with multiple vendors. Each vendor has its own service and with each service the cost way more depending on where that service is located for the pricing model. The type of data being sent or stored can often cause an added cost.
Another caveat is the hidden VPN cost — —most applications will require a VPN connection between the Multi-Cloud strategy. AWS VPN is currently USD $0.05 per hour, well Azure is between USD $0.04–$1.25. To add on to the hidden cost is the data transfer service that depends on the provider, but can cost between USD $0.02–0.05. This in turns results in about a USD $438-–$16,000 increase when using a single VPN connection a year.
For a security standpoint, using IAM (Identity and Access Management) to manage the security within the cloud requires a brokerage intervening with the cloud vendors. Making security between the cloud vendors very difficult without a brokerage tool.
Addressing the Multi-Cloud Challenge
Having a lot of balls in the air when designing a Multi-Cloud strategy could be compared to building the Lego’s Star Wars Collector’s Millennium Falcon with 10,179 pieces. It can be done in one of two ways; efficient and inefficient. The efficient way would be to plan out what needs to be done. By organizing the vendors of choice into categories and assigning employees to those categories. While an inefficient way on the other hand, would be going with the flow, allowing for mass chaos and not sorting the pieces out before joining the Multi-Cloud strategy.
Reducing VPN and data transfer cost can be done, but it will require detailed planning. In turn this can be mitigated by using other forms of data transferring methods. For example, AWS offers VPC Endpoints that can send data to S3 buckets compared to sending it over VPN. If you used a VPC Endpoint to S3 it would only cost the price of data transfer and reducing the cost of the VPN ($USD 438.25).
This has however been done by CMT with a design approach of using ADFS and SAML 2.0.
Multi-Cloud Configuration Management Tools
With the growth predicted by Gartner, Multi-Cloud configuration management tools (CMT) will also be viable option. Configuration management tools are integrated products that incorporate self-provision system images, enable billing, and monitoring. The ability to manage the Multi-Cloud strategy with one single management tool can limit a few of the challenges that Multi-Cloud brings. An issue one might run into regarding configuration management tools is vendor lock-ins. Being locked into a specific vendors CMT is something one wants to avoid in the Multi-Cloud approach. With the growing services that each cloud vendor is producing, this makes it extremely challenging task for cloud configuration management tools to keep updated with the newest cloud services; e.g., when AWS comes out with a new service, but the CMT will still need to patch the product before it can manage the new service. This could take weeks, if not months, before that new service is available to be managed.
Cloud configuration management tools are relatively new and are popping up all over. That brings us to the question: What is the best CMT? There has been over 50 CMT released over the past five plus years. Some of them haven’t made the cut and some of them are still going strong in the market. Each tool provides different approach to connecting from on-premise to a select few cloud vendors.
The different CMTs have different unique features, and each one might fit one organization better than another—it all depends on how the organization wants to connect to your future VM’s. CMT offer two ways to get its features into the VM’s: Agent and Agentless. Picking the integration method with the CMT depends on the organization needs also when choosing the correct CMT, can make or break any project in the future or present. Researching the best CMT for the organization needs is a big investment. The current CMT leaders at the time of this article are RightScale, Ansible, Scalr, Puppet, and Chef, in no particular order.
In summary—getting the most out of agility Multi-Cloud requires that you have relationships with multi-vendors, you have the right employee resources, and have a standardized operation and practical approach to what Multi-Cloud can do for you.
To accomplish success in Multi-Cloud you should invest in the right training for the right employees. You should weigh the cost and risk of Multi-Cloud up against the benefit it could have for your organization. Using a cloud management tool could relieve the stress on learning the Multi-Cloud strategy, but this could also place you right where you don’t want to be—locked into certain CMT vendors.