I have a Mac with the new M1 chip and it's not supported by MarkLogic. Could I install and run MarkLogic on an AWS WorkSpace?
I haven't used AWS WorkSpaces before but I know it supports Linux and Windows. MarkLogic has installers for Linux and Windows.
So you can run MarkLogic in an Amazon workspace, I was able to get a single node instance up and running. I chose the Power with Amazon Linux 2 (PCoIP) bundle and was able to download the RPM and follow the installation guild here: https://docs.marklogic.com/guide/installation/procedures#id_28962 (I did not comprehensively test it so your milage may vary).
My personal opinion below
I don't believe there is much official documentation around using MarkLogic in an Amazon workspace, so you may want to consider different infrastructure tools in case you run into issues down the road. For instance it might become cumbersome to setup a multi-node cluster in an Amazon workspace.
If you're ok with remote interfaces then you could consider installing Marklogic into an EC2 instance, or using the https://github.com/marklogic/cloud-enablement-aws/tree/10.0-develop repository for an easy cluster setup which you could interact with. Basically the repo above will provision however many nodes you want, and provide a load balancer you can use to interact with them across it. More docs can be found here: https://developer.marklogic.com/products/cloud/aws/
Bonus
Also Docker has M1 support. https://docs.docker.com/desktop/mac/apple-silicon/ and there do exist MarkLogic Docker Images: https://hub.docker.com/_/marklogic These may work on the M1 chip as well. However I don't believe its been tested extensively so again your milage my vary. I would still recommend launching into an EC2 instance, or better yet using the cloud enablement repo, deploying the cloufromation stack and using that.
ML should 'run' on either linux or windows workspaces but I do not recommend it for anything at all intensive. This is due to workspaces being entirely EBS storage, plus you are not in control of the EBS directly. So 100% of your IO will really be doing round trip networking. If accessed remotely 4x (2x for http or xcc request/response, 2x to 'disk' and back) Plus the pricing model for workspaces is fairly high on the high end. But 'it works'
Note: docker running on M1 in now way affects ML running on M1. M1 is a different chipset and requires apps to be compiled for that chipset (and OS). Docker is not a VM (except when it is run on top of one like on virtual box) so it doesn't make a M1 chipset into an x86m, so does not make 'ML for x86' into 'ML for M1'
ML 'works' in native x86 docker, be that docker in workspaces, docker on EC2 or docker via ECS or fargate.
Docker is particularly useful for testing cluster configurations without spending huge $ spinning up dozens of EC2.
There are CF stacks, and docker compose for ML publicly available from various sources of various quality. Recommend you always learn how these stacks work - many have not been updated in a timely way with AWS and docker.
