Cluster Room history
When I got to the lab in December 2010, the computing resources was only composed by 25-30 desktop machines, and the computing room by a few very old servers and about 25 unused desktop machines (with different CPU technologies and generations); the whole completely partitioned and without centralized management service.
After a few months, some PIs and I decided to think about a new computing architecture to significantly improve the computing and storage capabilities and performance. This was the origin of the cluster named Hades (2011-2012), closely followed by Lucifer (2013-2014) and Baal (initiated in 2015) ones.
Hereafter, the various budget sources that enabled this breakthrough:
- ERC ABIOS (Guillaume Stirnemann)
- ERC Thermos (Fabio Sterpone)
- ANR Exaviz (Marc Baaden)
- IUF (Philippe Derreumaux)
- ANR PEPS (Marc Baaden)
- ANR NicoChimera (Marc Baaden)
- ANR BioPac (Marc Baaden)
- ANR MAPPING (Sophie Sacquin-Mora)
- ANR COSMOS (Jérôme Hénin)
- LABEX DYNAMO
- LBT laboratory's funds
- …
On the next illustration, you can see the current computing room organigram.
Thereby, in a few years and only concerning the LBT's computation resources, we have:
- harmonized, homogenized and centralized the technical computing architecture
- multiplied by dozens -and even many more with GPGPU technology- the global computing power
- grown up the storage system from 0 to roughly 720 TB raw, scratch volumes excluded
That said, we have to keep in mind that the needed investment budget (for computing room rehabilitation and modernization) has been mainly apportioned for the first 4-5 years and is currently of around 900k-1M€ (hardware only).