Nordic WLCG Tier1

- part of the CERN worldwide calculator

Storage in Bergen, Oslo, Copenhagen, Espoo, Umeå and Linkjøping. Network provided by NORDUnet. And a lot of data coming in – around 10 Petabyte per year.

The Nordic Tier1 is one of 13 regional computing centres of the Worldwide LHC computing Grid, (WLCG), the huge international e-infrastructure that provides computing and storage for CERN, the European Organization for Nuclear Research.

As CERN does not have the resources to crunch all of the data on site, it resorts to grid computing to share the workload with computer centres around the world. The WLCG is a global collaboration of more than 170 computing centres in 42 countries, linking up national and international grid infrastructures to store, distribute and analyse the ~100 Petabytes (100 million Gigabytes) of data annually generated by the Large Hadron Collider. The grid is arranged in four layers, called Tier 0, 1, 2 and 3. Tier 0, the server farm at CERN itself, is the first point of contact between experimental data from the LHC and the Grid. From there data is distributed and processed, giving a community of over 8000 physicists near real-time access to LHC data.

Giant calculator

Or as they say at CERN: The world is our calculator.

The Scandinavian part of this giant calculator, the Nordic Tier1, developed under the auspices of NORDUnet, takes its share of the LHC data torrent. Just like the other 13 Tier1s – but yet differently. The NT1 is unique, as it is distributed between seven different locations and not concentrated in one location.

This setup was chosen out of political and financial reasons. The Nordic countries wanted a Tier1 site, but couldn´t agree on where to locate it. So, building on the Nordugrid initiative, they chose to distribute the site instead. That decision made the network aspect of the project crucially important, and so it was decided to hand over the responsibility for building and operating the site to NORDUnet.

At that time, around 2006, the other countries involved in the Worldwide LHC computing Grid met the Nordic decision to build a computing centre distributed across four countries with scepticism. The grid itself being distributed, why build a distributed Tier1 also? People shook their heads at the Northerners, but eventually that stopped, when the NT1 proved stable and dependable. And moreover, the NT1 now often is emphasized as a prime example of how you can effectively build and maintain this kind of computing centres. Among other things, some of the tailor made software developed by the Nordic programmers is now widely used in the WLCG community.

Private Nordic network

The NT1 receives, stores and pre-processes data from CERN, making it available through a “private” network to the Nordic high energy physics community of approx. 150 people, and to other researchers around the globe. The total inflow of data is around 10 Petabyte per year, 3 PB coming from CERN and the remaining 7 PB from other sites around the world and in the Nordics.

In 2013 the responsibility for the NT1 was shifted from NORDUnet to the Nordic e-infrastructure collaboration, Neic. However, NORDUnet is still a vital part of the setup, providing network between the seven NT1 locations, the 7th being Slovenia. Each location consists of two to three racks of servers and a tape robot for long time storage. Data is stored for 30 years. Furthermore, the NORDUnet network operations centre collaborates with NEICs own staff in monitoring the site 24/7.

Since launched in 2006 the Nordic Tier1 has proven its robustness and value. But still, a distributed computing and storage centre has both benefits and disadvantages. Among the benefits is the fact, that it’s always possible to store the data, as long as only one of the seven locations is operational. Among the disadvantages is the fact that having seven sites instead of one means there are seven times as many error sources.

In an ongoing discussion about pros and cons regarding a distributed vs. a centralised site, there is now an evaluation under way, to decide whether to stick with the current solution or to switch to a centralised one instead.

Highlights