An Interview with CERN

Learn how the IT department of CERN laboratory saves time and money for greater performance, reliability and security on their massive Elasticsearch deployment

May 11, 2018
What follows is an email interview with the lead of ROR adoption project at CERN in 2018.

Hi Ulrich! Who are you and what’s your role at CERN?

My name is Ulrich Schwickerath. Although I was an experimental physicist who did his PhD in Higgs boson searches at the LEP collider, I’m actually now working in the CERN IT department, so I’m not very involved in physics data analysis these days. One of my various tasks is to provide an easy-to-use and managed Elasticsearch service to interested user communities at CERN.
— All our Elasticsearch clusters are using ReadonlyREST. – Ulrich Schwickerath (CERN)

Give us some numbers! How much data flows through ReadonlyREST every day at CERN?

Nowadays, we use ReadonlyREST to provide index level security on shared Elasticsearch clusters. We have many small use cases which can share the same hardware resources. Our largest shared cluster at this time (Aug. 2017) consolidates about 12 different use cases on the same resources at a low cost, and we see http access rates of up to 90Hz on all clusters.

What is the origin of this data?

It depends on the use case and the users. A large portion comes from log files, like service or systemlogs. On top of that, another source is monitoring data, like service status or usage.
— Our largest shared cluster at this time (Nov 2017) consolidates about 17 different use cases on the same hardware, lowering the total cost

Can you give us a list of the departments/experiments whose data flows through ReadonlyREST?

All of our Elasticsearch clusters are using ReadonlyREST. Our customers include the four biggest LHC experiments, the IT department and the accelerators and technology sector.

What do you expect of the ReadonlyREST project in the future?

ReadonlyREST provides the functionality we need. We’re very happy with the support given by the community and the response time to our feedback and inputs. ReadonlyREST has become a critical component of our centralized Elasticsearch installation, and we rely on a continued good collaboration with the ReadonlyREST community.
— ReadonlyREST has become a critical component of our centralised Elasticsearch installation.
 

Related content: ReadonlyREST talk at CERN

https://cds.cern.ch/record/2261999

HOW CERN SAVES MONEY WITH READONLYREST

This year, CERN (The European Organization for Nuclear Research) optimized the usage of computing resources by consolidating 30+ Elasticsearch clusters into a handful of multi-user clusters.
Watch the presentation CERN organized to understand the guiding principles behind ReadonlyREST.