Component Soft’s main consultation references in cloud and cloud native topics
Planning, Building and Supporting OpenStack and Kubernetes systems
- Requirement analysis
- Planning the hardware and software elements of your OpenStack and Kubernetes system
- Building your OpenStack on bare-metal with Openstack Ansible or Triple-O
- Building your Kubernetes on OpenStack with Magnum or on bare-metal with kubeadm or kubespray
- High Availability and security features
Building, testing and supporting the mission critical OpenStack based NFV (Network Function Virtualization) system of the Hungarian Telekom
Hungarian Telekom, part of the Deutsche Telekom Group in Budapest Hungary
OpenStack, Ceph, Ansible
Building, testing and supporting the OpenStack based NFV (Network Function Virtualization) system, called TelcoCloud, of the Hungarian Telekom, part of the Deutsche Telekom Group. This is a relatively large system, currently with some 200 physical hosts in 3 OpenStack regions in 3 data centers and it is absolutely mission-critical.
In 2015 we were tasked with performing an independent compliance test of the company’s internally designed and built OpenStack region – part of its NFV system. The test revealed some functional and performance problems which were promptly fixed by our engineers together with the internal staff.
Based on these results we were contracted by Hungarian Telekom in January 2016 for yearly third level support and additional consultation hours for the Linux, OpenStack and Ceph parts of their system. The contract was renewed for the following years as well. During this period as well as solving several support issues, we also completed another independent compliance test on the second OpenStack region of this system in a second data center in 2016, designed and built again by internal engineers.
Based on our success in these fields we were tasked in late 2017 to actually design and build the third OpenStack region of this system in a third data center using OpenStack Queens. This required the rewriting of the install scripts from Chef to Ansible. The installation of this third OpenStack region was finished in Q3 2018 and we now help the onboarding of several mission critical telco applications from bare metal environments to the OpenStack cloud.
After finishing the onboarding, we will empty the first and second region of applications one by one, reinstall them with OpenStack Queens and populate them with the applications again. At the end of the upgrade project the TelcoCloud will have 3 redundancy regions all running up-to-date OpenStack and Ceph versions.
Designing and building an OpenStack + Ceph based cloud system to be used at the development and testing of a monitoring software for OpenStack based NFV (Network Function Virtualization) systems
SevOne (www.sevone.com) in Boston, USA
OpenStack, Ceph, Kubernetes, Istio
According to its web site, Boston-based SevOne “provides modern monitoring and analytics solutions that organizations need to monitor their networks today, tomorrow and beyond.” One of their new products under development is a solution to monitor OpenStack based NFV (Network Function Virtualization) systems at telco companies. While SevOne is an expert in monitoring networks it didn’t have the needed OpenStack and Ceph knowledge at the beginning of the project to design and implement the cloud system based on these technologies, to be used at the development and testing of this software. That’s why they turned to Component Soft in the spring of 2018 for help.
During a few online discussions with the SevOne’s sys-admins who work at their development lab datacenter in June 2018, we discussed their needs and goals with his system. Component Soft has designed and implemented the cloud system used by the development and testing team in the USA and India in July. Later on we provided support for the system in American East-coast working hours for the SevOne team.
The beta version of the software was finished by March 2019. Our new assignment at SevOne is to educate the developer, testing and technical support teams in the USA, India and Europe for Kubernetes and Istio in July and support them at the development of several new cloud-based solutions under development.
Migrating a cloud-based mission critical mobile advertisement platform from a Docker based solution in AWS to a Docker-Kubernetes based solution in the datacenter of Glispa while keeping up its reliability, security and response time.
Docker, Kubernetes, Calico, HAProxy EE, Envoy/Gloo, Cilium
Mobfox is a popular mobile advertisement solution, “the smarter platform for managing your media and optimizing results, to implement, monitor, optimize, and get paid, all in the same place”, according to their web site. It’s absolutely mission-critical for their advertising customers and has a strict expected response time given in milliseconds in which time frame it identifies the user of a mobile web page or app, determines the suitable ads according to the user’s characteristics and several pre-defined rules and display the chosen ads on the client’s mobile.
Mobfox has been running in Docker containers on AWS for a few years till now but the high price of AWS forced the company to investigate and design its migration to a Docker-Kubernetes environment running in the datacenter of Glispa while keeping up its reliability, security and response time. This migration also required to find open-source software to implement some auxiliary features of the AWS environment, e.g. networking, web proxy, API gateway and container aware application level firewalling which are seamlessly provided by AWS in the current setup.
Component Soft helped Mobfox and Glispa in late March and April 2019 to design the architecture of the new Docker-Kubernetes environment running in the corporate datacenter, also investigating several potential open-source software for the above features and choosing the suitable ones for the specific needs of the application. Then it built up the on-premise Kubernetes system with the selected additional software modules in the second half of April and May according to the above design. Our client is currently migrating the application to the Kubernetes cluster and we are negotiating new possible assignments.