Hyper-scaling on Openstack with Open Source tooling A use case in - - PowerPoint PPT Presentation

Hyper-scaling on Openstack with Open Source tooling

A use case in deploying hyper-scale grid computing on Open Telekom Cloud

Why Open Source?

Community Lead Contributor Model

Community Tested

Open Standards

Standards Organizations:

IEEE IEEE, ISOC ISOC, W3C 3C, IET IETF, IAB IAB, ET ETSI SI, OAS ASIS IS, NIS NIST, CSCC, CSCC, ANSI ANSI

Open Design

Cloud Design Patterns

Application Design Patterns

Distributed Systems Design Patterns

Open Choice

No-Lock in

Interchangeable Components Interoperability Reusability Invest in People not Software

Why Openstack ?

Community: The OpenStack community is nearing 10,000 members in nearly 100 countries, with foundation support by more than 200 companies. Mixed Hypervisor: OpenStack is the only solution that allows for mixed hypervisor IT environments Open Source: Open source solutions are now widely accepted and adopted after 20 years in the enterprise, paving the way for OpenStack success. Standards: provides a common, open standard through its API and allows portability between cloud environments. T-Systems Hybrid Cloud and Public Cloud based on Openstack (Same Architecture) providing interoperability and ease of migration or Burst scenarios. Community: Sharing ideas with the community allows our Developers to learn fast and deliver

• quicker. Contributing advances the entire

ecosystem Research & Development: Lowers our R&D cost and speeds time to market for new PaaS services. Open Source: Interoperability with most open source projects such a Kubernetes, Spark, Etc providing us the ability to deliver standardized services. Standards: Standardization allows us to delivery Public Cloud and Private Cloud solutions architected exactly the same. We can deliver new PaaS services within 90-120 days of requirements from customers.

User Benefits Benefits to Cloud Provider

Openstack Contribution

Source: stackalytics.com

 Adaptability and access to the source code  The open architecture design gives in-house IT

more scope to tailor the technology to specific needs

 Participating in the community enables us to

influence developments

 Community: many providers and solutions that

enhance functionality

 Establishment of multi-cloud scenarios  Hardware neutrality: connectors are available for

almost all storage, server and network components on the market

 High degree of standardization (open APIs)  Resources can be provisioned and marketed

across provider boundaries

 Data and applications can be transferred from

provider to provider

How do we contribute to the Open Source community

Zaqar Kuryr Tacker Rally Ceilometer Search Light Karbor Manila Neutron Zun Heat Tricircle

• We developed the Tricircle project which allows

Openstack to be a viable Public Cloud architecture.

• We contribute regularly to the project listed to the right as

leader or team member.

• We are on the board of governors for Openstack
• The Openstack and Open Source community benefit

from much of the Research and Development work we do for the public cloud and private cloud services.

• We have circa 5,000 engineers dedicated to the

Openstack projects for Public and Private Cloud Solutions who are leading the way on AI, ML, IOT, Kubernetes Container services.

• Corporate member of the Linux Foundation and on

governing board of CNCF, Member LF Deep Learning Foundation.

Hybrid Cloud Solution based on OpenStack

Management Link (OPS – public OTC) Management Link (OPS – Hybrid/Private Region) Data link (Hybrid/Private Region to public OTC)

E.g. Ethernet Connect or MPLS

TSI Datacenter Datacenter of choice

Small scale

O&M Team Budapest

• Public Cloud Architecture on-premise at

customer site or hosted location

• Agility and scalability built in
• Cloud Native capability
• Public Cloud based on Openstack

Deployment Tooling with Terraforms and Ansible

What is Terraform ?

• Terraform is a multi-cloud DevOps tool for

building, changing, and versioning infrastructure safely and efficiently. Available as Open Source or Enterprise License.

• Enables Infrastructure-As-Code

deployments

• Open Telekom Provider developed for

Terraform

What is Ansible ?

• Application Deployment

Tooling

• Agentless Architecture
• Integrates seamlessly with

Terraforms

Grid Computing in Finance Industry - Use Case for OTC

What is Grid Computing

Grid computing is a distributed architecture of large numbers of computers connected to solve a complex problem. In the grid computing model, servers or personal computers run independent tasks and are loosely linked by the Internet or low-speed networks. Computers may connect directly or via scheduling systems.

Finance Use Case (Grid Computing)

• Bank was looking to cloud computing to carry out their complex calculations, marking a shift from investing heavily in

in-house systems that are expensive to develop, maintain and upgrade.

• Regulatory and Business demands such as
• Fundamental review of the trading book (FRTB)
• Derivatives Valuation Adjustments (XVA)
• End-Of-Day & Monte Carlo
• Transforming IT Operations to DevOPS methodology.
• Standardized on Open Source tooling and systems such as Ansible, Terraforms, Kubernettes, Kafka, etc)
• Need for massively scalable cloud services for Grid Computing, Big Data, and Devops( Kubernetes, ML, Analytics,

Open Banking)

• Opex payment model.
• Do not want to invest in on premise servers that are not used much of the time or ever have enough capacity.
• Must speed up time to results to the business (Currently 36 to 48 hours) on FRTB and XVA calculations.

Business Requirements IT Requirements

Grid Computing Requirements and Challenges

Technical Requirements

Capacity requirements 26,0 ,000 vCPUs minimum 50,0 ,000 vCPUs Burst Grid Software: Tibco Gridserver Infrastructure deployment with Terraform Server Configuration with Ansible Elastic Compute (ECS) and Elastic Volume Service

(EVS)

Ch Chal allenge ges to

• Deliv

Delivery API Gateway overloaded – Rearchitectected Openstack API services to handle up to 10,000 API calls a second. Server provisioning to slow due to delay in image services deployment

• f images. Improved storage network capacity and latency as well

image distribution across pools). Required lots of very large hosts which were not already available. Deployed 2000 additional host to meet demand.

Improvements in Processing time

Primary Benefit to Bank – Time is money

Deploying the Grid using Terraforms

resource "opentelekomcloud_compute_instance_v2" "engine_node1" { count = "${var.vm_engine_count}" availability_zone ="eu-de-01" name = "${var.project}-node${format("%02d", count.index+1)}-az1" image_name = "Enterprise_Windows_STD_2012R2_XEN" flavor_name = "s2.medium.4" key_pair = "${openstack_compute_keypair_v2.keypair.name}" security_groups = [ "AD" ] #ANSIBLE scripts user_data = "#ps1\nGet-NetAdapter\nnet user ansible Test1234+ /add\nnet localgroup administrators ansible /add\n#ps1\nGet- NetAdapter -name 'Local Area Connection' | Set-DnsClient - ConnectionSpecificSuffix cloudgrid.bank.com - RegisterThisConnectionsAddress:$true - UseSuffixWhenRegistering:$true\nSet-NetFirewallProfile -Profile Domain,Public,Private -Enabled False "

OTC OTC API

Full Grid deployed to OTC Terraforms Plan Circa 2 Hours to deploy 15,000 vCPUs