Feb 7, 2019 from Activeeon
Within two years, 75% of businesses looking for IaaS (infrastructure as a service) and PaaS (platform as a service) solutions will be expressing the need for multi-cloud features, according to Gartner. They were not more than 30% in 2018.
Moving your infrastructure to a multi-cloud strategy is a good opportunity to innovate, as the Cloud encourages optimization, modernization and testing of new solutions. In this context, orchestration and automation tools are essential to improve agility, rethink the data flows and parallelize.
“The multi-cloud context clearly encourages a rework of the data workflows and the associated processes to better integrate new demands and leverage opportunities” explains François Tournesac, Activeeon chief sales officer.
The Cloud flows need to consider the type of resource used: VM type, computing power (CPU, FGPA, GPU, location, etc.). The goal is to improve IT processes automation by supplying optimal resources when needed.
This means that architectures must be rebuilt: “Modern schedulers need to integrate resource management to better control costs, optimize run times and select resources with specific features. It is essential to make the infrastructure elastic and scalable and allows an almost instant provisioning while controlling costs” emphasizes François Tournesac.
Another benefit of this approach is that it gives easy access to instantly available services such as managed databases.
“To move towards multi-cloud, it is needed to keep a demarcation line between the Cloud provider and the enterprise. You must stay independent and not locked in, i.e. minimize the influence of the Cloud providers”. But it is also important to take advantage of production tested solutions in real businesses and to benefit from the competition between Cloud providers.
To redesign workflows, it is better to use tools with a large range of connectors to the most popular APIs, most notably those opened to most of the Cloud-oriented offers and services such as MongoDB or PostgreSQL.
The market is moving towards a multi-cloud hybridization: “The first goal was to add some elasticity through public cloud-bursting from the internal IT, being Cloud or not. But the bandwidth and the network latency quickly limit the possibilities that orchestration solution can plan to optimize” according to François Tournesac.
A good way to reduce the dependencies is to access to multiple Clouds. This way, SNCF (the French national railways company) can distribute apps and data on different types of Cloud according to cost and performance criteria while developing its own private Cloud, a good way to learn and progress on technologies such as containerization for instance.
Public Cloud open doors, for instance, Machine Learning and Deep Learning algorithms can be quickly put in production thanks to automation and orchestration tools that maximize the use of GPU and ease the transition to production with better control, scalability and cost optimization.
In France, IT service firms offer a large range of IaaS (Infrastructure as a Software) solutions but few PaaS (Platform as a Service, reserved for developers) solutions. For instance, Microsoft Azure and Activeeon now offer a platform for “meta-scheduling” to plan and run Batch-processed automated tasks, such as parallelized computing tasks that are pipelined on a large scale of VMs. “One of the main benefits is that you only pay for what you use” explains François Tournesac as a conclusion.
Copied and edited from Silicon.fr
Apr 28, 2021 from Activeeon
Notification service allows users to subscribe to event-based notifications triggered either by workflow executions and their tasks, the scheduler or job planner....
Oct 23, 2020 from Activeeon
Deep learning algorithms are a series of (deep) neural networks that learn to recognise patterns from data. However, finding high-performance neural networks architectures for a certain type of application can demand many years of research...
Sep 3, 2020 from Activeeon
Let’s suppose that you have a large infrastructure containing several machines that have different operating systems (e.g. Microsoft Windows, Linux, MacOS) and distinct hardware configurations...