What feels alot longer than a month ago I replied to a tweet saying we would be publishing a series of blog posts on how we manage our docker infrastructure and related processes here at Lancaster University Library.
The docs for how we build and manage production docker nodes and images @LancasterUniLib should be going public soon as a series of blog posts, hopefully this will be of help
— Stephen Robinson (@StephenJamesRob) May 14, 2019
Since posting that tweet and preparing the articles for publishing we have undergone a major overhaul of our infrastructure within the Digital Innovation Team, mainly prompted by the rising costs of our AWS bill and our colleagues over in Information Systems Services moving to VSAN for our on-premise VMware clusters. With this in mind I decided to delay posting the content to ensure it best reflected the practices going forward within our team.
Covering a diverse portfolio of over 30 different services hosted on a wide spectrum of infrastructure platforms from on-premise VMware Clusters to the AWS public cloud alongside SaaS vendor platforms. The Digital Innovation Team are always looking for ways we can optimise our workload and make supporting the teaching, learning and research at Lancaster University as easy as possible for us.
Currently we maange approximately 35 on-premise virtual machines totalling 80vCPUs, 225GB RAM and 5TB of storage hosted in dual on site datacenters at Lancaster University, this is complimented with a range of virtual machines and services in the AWS cloud which we have used for evaluating a range of different services. Our usage of these platforms covers all of our development, staging and production environments as well as providing tools and services for our development teams.
In reviewing our infrastructure we were looking to consolidate the range of different operating systems we were running in production alongside providing simpler ways to deploy and manage applications and services running on top of these. Having evaluated Kubernetes and docker on multiple occasions we settled on using these as the core part of our new infrastructure platform, hoping to provide a consistent experience for all our developers, whether full time staff or part time student staff. Alongside Kubernetes we plan to implement a range of support services running on dedicated virtual machines where this would aid in the deployment and performance of these.
Over the next weeks and months expect a variety of blog posts on how we manage both on premise and AWS cloud infrastructure within the team covering the variety of processes and tooling we have in place alongside the reasoning behind many of these decisions.
With all these topics we aim to show how we continually handle operational issues with these and ensure everything is managed with production in mind, No quick ‘how-to’ tutorials or summary posts. We’ll dig into the details of setting infrastructure up with day 2 operations in mind.
Our next post will provide more detailed information on our new infrastructure and how we are proposing to migrate from our current infrastructure to this.