Well, most people know what public and private clouds are by now. A hybrid cloud is a mix of a private and a public cloud where part is located in a dedicated, private area and part is accessible via public systems.
A multi‑cloud is where several cloud providers are used in parallel. The use of these providers is more or less transparent for both users and administrators. So users shouldn’t notice any difference whatever provider is hosting the services used. And suitable tools should be used to abstract administrative processes sufficiently so that administrators can use the same tools for all providers.
Advantages of multi‑cloud
But why use it? Using a range of cloud providers brings advantages. It allows you to choose the best solution from among your cloud providers’ portfolios. AWS might have an AI service that fits your requirements perfectly right now, but you prefer the cost structure of Google’s messaging service and buy an OpenShift platform from Red Hat. This best‑of‑breed approach can reduce development times and has the potential to cut costs.
Deploying identical services across several providers also increases uptime. It is rare for large‑scale outages to affect two cloud providers simultaneously. These types of faults do occur, as shown by the outage of large parts of the AWS cloud in March 2017, when an incorrect entry by a technician paralyzed 54 of the 100 largest online retailers in the US for several hours. A multi‑cloud approach would have been able to prevent this. Your organization is probably already using a multi‑cloud. Presumably you already get your infrastructure services from Azure, and your sales team works with Salesforce. Since Salesforce is a software as a service solution, you are operating two clouds. And voilà – you have a multi‑cloud.
The abovementioned advantages come at a price. And this starts with procurement. A separate contract is required for each cloud provider. These have to be negotiated, understood and aligned with company guidelines. The compliance department has to be sure that the various providers meet regulatory requirements. All these conditions are dynamic because terms and conditions as well as regulations are in a constant state of flux and are continually being adjusted by both sides.
Special attention should be paid to data residency. In sensitive sectors it is imperative that you know at all times where the relevant data are stored and processed and via which routes they are transported. Ideally your cloud strategy should set out clearly which data are suited to which cloud.
In addition to the legal and regulatory aspects, the various services have to be integrated on a technical level. Communication channels have to be set up, secured and maintained between on‑premise installations and each individual cloud service. To these must be added any links between the cloud services so that the many connections can be brought together as appropriate for the topology and your requirements.
User and rights management presents a particular challenge. Anyone with experience of Identity and Access Management (IAM) projects will know how difficult it is to define and implement functioning processes in IAM. Synchronization and delegation between different providers makes this even more complicated. At the application level, protocol adapters may be required to enable data exchange between different providers’ services. These need to be developed, tested and provided.
One frequent source of problems is latency, which is by definition much greater between services supplied by different providers than within a single data center. With current microservice architectures, latency can mushroom into a serious performance issue, even with planned workload scenarios, if this aspect is not taken into account in planning and thoroughly tested during the implementation stage.
Orchestration and management
If application mobility is assured by several cloud providers, abstractions must be defined for the interfaces and orchestration. After all, porting an application from Cloud A to Cloud B needs to be as easy as possible. Guidelines on this subject are complex and call for extensive experience if serious errors are to be avoided. Specialized tools are available to assist with the management and operation of multi‑cloud environments. Parts of services can also be outsourced to service providers. The important thing is to have a fully automated delivery pipeline across all providers in order to keep operating costs low while ensuring service quality remains high.
The pivotal issue for any strategy that envisages the use of a multi‑cloud is expertise. Ultimately, expertise must cover both the technical interfaces and cloud providers’ toolsets and product portfolios. This hurdle gets harder to clear with each additional partner while the technical demands placed on IT staff spiral.
The adoption of a multi‑cloud strategy should be given careful consideration. The ti&m multi‑cloud strategy is designed to help with this: Our diverse projects in the banking, insurance, public and transportation sectors mean that our experts are familiar both with the challenges specific to individual sectors and with cloud providers’ strengths and weaknesses. ti&m is currently partnered with AWS, Google, Microsoft and Red Hat, and is therefore able to take a vendor‑neutral approach to finding you the optimal solution. And with the ti&m swiss banking cloud, we can offer you the option of a private cloud based in Switzerland to meet any special requirements you might have.