In cloud services conversations , AWS is often cited as a leader. But for most companies, suppliers is just an entry. More and more, they are adopting multicloud strategies that combine elements of public cloud and private cloud . The interest is to fight against a captive commitment with a single supplier, to diversify the available services, to weigh on the disparities of tariffs and / or to keep control on confidential information.

The use of several cloud computing platforms therefore seems a good strategy. But it can cause application management problems, API management and configuration management .

Kết quả hình ảnh cho công nghệ 4.0

Managing applications and infrastructure configurations on different clouds that do not share the same API and rely on very different services and billing models can be a very complex task. But it's not really a drag, because a number of mature software and SaaS options can automate deployments across different clouds. However, all automation tools are based on the same conceptual framework: it is about treating cloud resources as abstract objects that can be configured, executed and managed like code. This encroaches on DevOps organization methods and models .

The imperatives of multicloud

Do you think that the use of several cloud computing platforms is reserved for startups or multinationals born in the Cloud and hyper growth? Think again !


According to a RightScale study , 58% of the companies surveyed use a combination of public and private clouds. In addition, 14% of them deploy a strategy combining several public clouds, while 55% are moving towards a hybrid public / private environment. According to Forrester, 52% of large companies already use multiple public cloud providers, and one third use at least three cloud computing platforms

The requirements of multicloud are quite simple. No company wants to see a critical infrastructure depend on a single supplier, as important and reliable as it is. Indeed, without an appropriate architecture including multiple availability zones, the risks of failures are real as could be seen by Amazon, AirB2B, Tinder, Reddit and others during the breakdown that hit Northern Virginia for several hours.

And yet, as Forrester points out, the heterogeneity of clouds is a source of concern for IT professionals. Hybrid multivendor cloud models require rigorous management across multiple platforms to provide a consistent experience for developers and application users.

Management interfaces and disparate control interfaces are a real source of frustration. And indeed, a quarter of the people interviewed by RightScale find that the management of several cloud platforms "is a challenge".

If you look deeper, Forrester's data reveal the main difficulties of multicloud management:

homogeneity of services between suppliers;


migrate workloads between clouds;


Consolidated management across multiple clouds


support for multiple Cloud user portals.


But in all these cases, a cloud-independent deployment software can solve the problem.

Independent Multicloud Management Options

There are dozens of software and SaaS products designed to automate infrastructure management and application management across multiple clouds.

Some are focused on specific needs or usage scenarios. Cloudyn , for example, is designed for asset and cost management. It includes a workload optimizer to identify the deployment option that has the best cost-performance ratio for a given workload.

As for CSC , which uses the former ServiceMesh product, it focuses on cloud governance, security and lifecycle management.

Others, like Cliqr or ElasticBox (Century Link), are taking an application-oriented cloud automation approach.

The most popular Multicloud products are typically used by companies that adopt a DevOps approach in the cloud , a tactic that extends application programming to the field of infrastructure configuration and management. The programming language is indeed an important differentiator of the tools.

RightScale is generally one of the products chosen for cloud automation. But, according to their own survey, the most used infrastructure DevOps tools are Chef, Puppet, Ansible and SaltStack .

Chef , who sets the culinary metaphor, turns the configuration, deployment and management of infrastructure into a set of recipes that can be interpreted by all systems using the Chef customer. Behind the scenes, everything is not so simple on the server side, of course, but Chef can manage all the elements of the cloud application deployment and can be run directly in the cloud (in other words, the Chef server, the developer workstations the system nodes and the analytics engine can all be run as IaaS instances). Chef supports leading cloud services including AWS, Azure, Google, VMware (vCloud Air) and IBM Softlayer, as well as SmartCloud Orchestrator and Rackspace / OpenStack.

Puppet , which is considered the ancestor of orchestration software, is both mature and widely supported. Puppet uses a class-based language, inspired by the Nagios configuration file format, which is close to JSON. Although equipped with a Web user interface, it requires a passage through the programming and use of CLI for advanced configurations. A new tool, Puppet Razor , enables automatic infrastructure discovery and inventory and dynamic selection of a preferred system image for naked system provisioning.

Ansible is an agentless open source platform, which commercial version was purchased by Red Hat . It works entirely through secure shell connections. Its configuration playbooks, based on YAML, are used for configuration, deployment and orchestration of systems.

SaltStack is a fairly new platform, focused on speed and scalability. It is offered both as open source code and enterprise edition. Salt uses the YAML language to describe system states, but the platform includes a complex set of components that imposes a steep learning curve, especially for novices on automation platforms.

recommendations

All major automation platforms described here work on both private infrastructure and large public clouds. But their integration characteristics present great differences.

The choice of product should be dictated by the level of evolution and scope of your infrastructure and by the expertise of the IT / DevOps team.

Embedded SaaS products such as RightScale or Scalr are the easiest to deploy and use because they all have a complete web interface and prebuilt templates and integrations for leading cloud services.

Still, their connection to the internal infrastructure can be more complicated. Puppet is the most mature general-purpose tool, making it a favorite in big business. But Ansil, with its agentless design and simple YAML syntax, is arguably the easiest tool to implement and master.