Before starting
let me just not clarify
but be a little bit more specific.
So the software and our proposal that I am going to present today started a few years
ago as a research project between BSC and Lenovo in the context of energy
runtime
energy monitoring and optimization
mostly for HPC at that moment
like six
seven years ago. Okay. I'm still associate researchers and the team leader of the system
software for energy management.
And then like five years ago, we decided to create the company,
which is a spin-off of BSC and UPC.
And then right now, and this is just to give you a little bit
more of context
we have decided that there is
let's say these two lines in parallel where BSC
we deploy what we call the ER core
which is the open source part of ER with the main functionality
both in terms of monitoring and optimization.
But then we created this company to provide
professional services on top of ER. And in the last two years, we also started some
like private developments specific for the company,
mostly in the context of GPU optimization, analytics, and so on. Today, the presentation
will be a technical presentation.
I mean, I am not going to enter into business details or
these kinds of things.
Okay.
So this is how I, let's say, prepare this presentation.
So probably the
initial five minutes will be something that all of you will be aware or well known of why energy
efficiency is needed in data centers. But I will give you a little bit our view of why it's needed,
which are the stakeholders or the participants of this overall goal of having energy efficiency
data centers. And what is the ER contribution in this global picture? Because of course,
we are a system software tool
and then we cannot have too much influence
let's say
for example
in the hardware or in other parts of the system, like the cooling, but it's just to put the global
context.
So based on this initial motivation, I will go through different topics.
What is our view
or what is the full picture of the data center monitoring and management? In some of these parts,
I will mention that we are doing some active development or active contributions or at BSE
or in the context of the company.
And in some other cases, it's just the roadmap we have for a
medium term
but this may be in some other cases
we don't have yet any
let's say
Presenters
Zugänglich über
Offener Zugang
Dauer
01:04:36 Min
Aufnahmedatum
2025-10-07
Hochgeladen am
2025-10-21 16:00:06
Sprache
en-US
Speaker: Julita Corbalán Gonzalez, CTO and co-founder of EAR, associate professor at the Polytechnic University of Catalonia
Slides: On request
Abstract:
EAR is a powerful yet easy-to-use system-level software for energy monitoring and optimization. Designed to be vendor-independent, EAR runs seamlessly across any architecture, offering unmatched flexibility. It adds semantic meaning to raw metrics, reducing the need for deep expertise or domain-specific knowledge to achieve optimal performance. EAR simplifies complex energy data, making advanced energy management accessible to everyone. Its modular design allows it to operate as a full system tool—managing all aspects of energy usage—or as a lightweight user-level library. This flexibility makes EAR the ideal solution for diverse data center environments. Whether you’re optimizing a single server or managing an entire infrastructure, EAR adapts to your needs. With EAR, energy efficiency becomes intuitive, scalable, and architecture-agnostic. The presentation will introduce main EAR features, architecture, and some relevant use cases to illustrate how EAR provides practical solutions to users, administrators, and data center managers to convert your data center into an energy-efficient data center.
For a list of past and upcoming NHR PerfLab seminar events, please see: https://hpc.fau.de/research/nhr-perflab-seminar-series/