small cloud team big impact supercomputing research

When science researchers at the University of Zurich need help with data analysis, a boost in computational power, run out of storage or need to optimize their workflow, they turn to the Service and Support for Science IT unit, or S3IT. The team also serves research groups and offers services for international projects.

S3IT adopts free and open-source solutions wherever possible, including Python, Linux and OpenStack. Their GitHub repositories are available at https://github.com/gc3-uzh-ch and https://github.com/uzh.

Superuser talked to Hanieh Rajabi, cloud engineer at S3IT, to find out more.

Tell us more about the team and where OpenStack comes in.

I started working on OpenStack after joining the S3IT group in April 2016.

S3IT supports University of Zurich researchers and research groups in using IT to support their research, from consultancy to app support and access to cutting-edge cloud (OpenStack-based), cluster and supercomputing systems.

Our OpenStack infrastructure is quite big — it offers 13,184 vCPU, 64 TB of ram, 4.5 PetaByte Ceph storage and 1.8 PetaByte Swift storage in raw. We’re a team of four running this infrastructure. As a cloud engineer, I run day-to-day operations on OpenStack.

I also do training for our OpenStack users. Each month we give a quick introduction session for new users joining the cloud and every three months I run Openstack training, teaching researchers how to use Horizon to manage their cloud resources.

The Barcelona Summit was your first — what’s your biggest takeaway?

It was really interesting for me. I was really impressed with how huge this community is and how many different topics and concurrent tracks there are. At the Summit it’s really easy to understand the value of Openstack, it’s amazing how enterprises are investing in a open-source software and in a community of open-source developers. When you start working with Openstack in your office, you can’t imagine how big and alive the community behind the project is.

It was really difficult to choose which talks to attend, as there was more than one interesting one in the same time slot. I really liked the troubleshooting sessions where I learned how to debug Neutron problems. The upgrades topic was important for me as I was engaged in upgrading to Mitaka at the time. Attending sessions with lots of tips about upgrades was really helpful.

Other than the routine operation, I was also interested into the Manila, Swift and Swift encryption talk using Barbican and I enjoyed the Barbican workshop a lot.

To keep myself up to date on containers, I followed a lot of sessions about Kubernetes and using it as an Openstack control plane.

Why do you think it’s important for women to get involved with OpenStack?

It is true that the number of men and women in Openstack, and in IT in general, is unbalanced. We need more women to join the community. Women need not to be afraid to show their talent for this cutting-edge technology.

I think that when the community will be more balanced, it will be easier for people of any gender to join.

What obstacles do women face when getting involved in the OpenStack community?

They are always in the minority and sometimes that just does not work. Being the only woman at your workplace sometimes makes you feel shy.

There are always debates in the OpenStack community – which one is the most important now, in your opinion ?

I think that IPv6 support for production clouds is still not completely there. This topic came up both at the round-table session and at some specific Neutron presentations. With IPv4, the network design is different, because the floating IP abstraction with NAT leads to little interaction between Neutron and the physical network gear.

With IPv6 routing, it’s absolutely necessary to have a common routing protocol between the physical network gear of the data center and the Neutron virtual routers. However, the discussion is still going on how these routing IPv6 features are going to be implemented and so far I have seen very little adoption of IPv6 in production clouds. Even people who are running IPv6 in production have to coordinate configuration on the physical network equipment of the data center and Openstack Neutron to have routing working correctly…

How do you stay on top of things and/or learn more?

I try to read the OpenStack operators mailing list to be updated with the current discussions. I also try to be active in Switzerland, attending the OpenStack meetups. It’s great to attend the meetups to learn from other OpenStack operators. Last but not least, at the university we have quite a dynamic environment where learning and running new experiments/technology is always welcome.

This post is part of the Women of OpenStack interview series to spotlight women in various roles within our community, who have helped make OpenStack successful. With each post, we learn more about each woman’s involvement in the community and how they see the future of OpenStack taking shape. If you’re interested in being featured, please email [email protected].

Tags: HPC, Supercomputing, Women of OpenStack