Dear all,
My name is Romain; I’m a researcher at ETH Zürich working with Laurent Vanbever. I have led projects on sustainable networking for about 2+ years now.
One important challenge the academic community currently faces is the lack of quality data for the energy a router or switch consumes. To address this, we work on fine-grained power modeling of network devices, and we need data to validate their accuracy “in the wild.” To facilitate that, we worked to realize the vision of “RIPE Atlas for power data” I pitched at last winter’s RIPE meeting https://ripe87.ripe.net/archives/video/1143/. And I’m happy to say that we are ready to start distributing measurement units!
Concretely, we are looking for people willing to
*
Install measurement units in series of each PSU of routers and switches. The measurement unit is an MCP39F511N https://www.microchip.com/en-us/development-tool/ADM00706 power meter combined with a Raspberry Pi for management and data collection:
We will provide the appropriate cabling. To control the unit, the Pi needs to be connected to a publicly reachable network via RJ45 (WiFi connectivity is technically possible as well, but requires additional setup)
*
Share device-specific data with us, including at minima
o Packet and/or byte counters for the active device interfaces. o Types and numbers of transceivers plugged in.
and optionally
o PSU-internal measurements o Operating system version o Fan speeds o CPU load o Average room temperature conditions *
Let the measurement run for at least one week.
*
Send us back the measurement units (we provide pre-labeled packaging for that).
In return, we can only offer our sincere gratitude and the guarantee that your effort will contribute to the progress of science! I know it s not much, but I hope that’s worth something :-)
➡️ If you are interested, please let me know by replying to this email. Thank you in advance!
Note: We want to publish the collected data (in an anonymized way), but that is *optional*. If you are willing to contribute but your organization opposes the data sharing, we won’t publish it, and we will sign NDAs if necessary.
Thank you again for reading! Looking forward to hearing from you,
Hello Romain, Laurent
Looks really cool what your trying to achieve, I think addressing this at a fundamental level would be to to include your input and work in the ARM SystemARCH group and SystemReady programs. As this is working with the actual device vendors and manufactures, e.g OEMs' ODM's It steer's what data we expect out of ACPI for example. USB-PD started to put pressure too on this topic. like CPE's, POE based devices and more carrier gear you see in the TN stacks at Telco's
I work with ARM, i;m in the technical committee for both groups above. A large majority of network gear is built on, as energy and performance is key driver for them. Just reach out directly to me
btw.. two challenges with your setup: in-line power is physically difficult, most network gear has two circults, two PSU's, and current goes up to 16A access to PDU data will give you mA over SNMP, but mostly in management networks
BR Darragh Grealish 56K.Cloud
________________________________ From: Romain Jacob via swinog swinog@lists.swinog.ch Sent: 20 June 2024 17:00 To: swinog@lists.swinog.ch swinog@lists.swinog.ch Cc: Vanbever Laurent lvanbever@ethz.ch Subject: [swinog] Research project on Sustainable Networking
Dear all,
My name is Romain; I’m a researcher at ETH Zürich working with Laurent Vanbever. I have led projects on sustainable networking for about 2+ years now.
One important challenge the academic community currently faces is the lack of quality data for the energy a router or switch consumes. To address this, we work on fine-grained power modeling of network devices, and we need data to validate their accuracy “in the wild.” To facilitate that, we worked to realize the vision of “RIPE Atlas for power data” I pitched at last winter’s RIPE meetinghttps://ripe87.ripe.net/archives/video/1143/. And I’m happy to say that we are ready to start distributing measurement units!
Concretely, we are looking for people willing to
* Install measurement units in series of each PSU of routers and switches. The measurement unit is an MCP39F511Nhttps://www.microchip.com/en-us/development-tool/ADM00706 power meter combined with a Raspberry Pi for management and data collection:
[cid:part1.Gh4kAtHz.OQ00ccNf@ethz.ch]
We will provide the appropriate cabling. To control the unit, the Pi needs to be connected to a publicly reachable network via RJ45 (WiFi connectivity is technically possible as well, but requires additional setup)
* Share device-specific data with us, including at minima
* Packet and/or byte counters for the active device interfaces. * Types and numbers of transceivers plugged in.
and optionally
* PSU-internal measurements * Operating system version * Fan speeds * CPU load * Average room temperature conditions * Let the measurement run for at least one week.
* Send us back the measurement units (we provide pre-labeled packaging for that).
In return, we can only offer our sincere gratitude and the guarantee that your effort will contribute to the progress of science! I know it s not much, but I hope that’s worth something :-)
➡️ If you are interested, please let me know by replying to this email. Thank you in advance!
Note: We want to publish the collected data (in an anonymized way), but that is optional. If you are willing to contribute but your organization opposes the data sharing, we won’t publish it, and we will sign NDAs if necessary.
Thank you again for reading! Looking forward to hearing from you,
-- Romain JACOB Postdoctoral Researcher ETH Zurich Networked Systems Group (NSG) Lead: Prof. Laurent Vanbever www.romainjacob.nethttps://www.romainjacob.net/ @RJacobPartnerhttps://twitter.com/RJacobPartner @jacobr@discuss.systemshttps://discuss.systems/@jacobr Gloriastrasse 35, ETZ G81 8092 Zurich +41 7 68 16 88 22
Hello Romain
Could you please go into more detail concerning the phrase: the Pi needs to be connected to a publicly reachable network via RJ45
Does this mean, you want to be able to remote to this pi? Or does the pi only <send> data and it would be ok to have it behind a firewall? How about router with dual power supply?
You may also contact me off list for more informations, perhaps we may share such data.
Kind regards, Urs
SBB AG Poststrasse 6, 3072 Ostermundigen
urs.bf.mueller@sbb.chmailto:urs.bf.mueller@sbb.ch / www.sbb.chhttp://www.sbb.ch/
Von: Romain Jacob via swinog swinog@lists.swinog.ch Datum: Donnerstag, 20. Juni 2024 um 17:00 An: swinog@lists.swinog.ch swinog@lists.swinog.ch Cc: Vanbever Laurent lvanbever@ethz.ch Betreff: [swinog] Research project on Sustainable Networking
Dear all,
My name is Romain; I’m a researcher at ETH Zürich working with Laurent Vanbever. I have led projects on sustainable networking for about 2+ years now.
One important challenge the academic community currently faces is the lack of quality data for the energy a router or switch consumes. To address this, we work on fine-grained power modeling of network devices, and we need data to validate their accuracy “in the wild.” To facilitate that, we worked to realize the vision of “RIPE Atlas for power data” I pitched at last winter’s RIPE meetinghttps://ripe87.ripe.net/archives/video/1143/. And I’m happy to say that we are ready to start distributing measurement units!
Concretely, we are looking for people willing to
* Install measurement units in series of each PSU of routers and switches. The measurement unit is an MCP39F511Nhttps://www.microchip.com/en-us/development-tool/ADM00706 power meter combined with a Raspberry Pi for management and data collection: [cid:part1.Gh4kAtHz.OQ00ccNf@ethz.ch]
We will provide the appropriate cabling. To control the unit, the Pi needs to be connected to a publicly reachable network via RJ45 (WiFi connectivity is technically possible as well, but requires additional setup)
* Share device-specific data with us, including at minima
* Packet and/or byte counters for the active device interfaces. * Types and numbers of transceivers plugged in.
and optionally
* PSU-internal measurements * Operating system version * Fan speeds * CPU load * Average room temperature conditions
* Let the measurement run for at least one week. * Send us back the measurement units (we provide pre-labeled packaging for that).
In return, we can only offer our sincere gratitude and the guarantee that your effort will contribute to the progress of science! I know it s not much, but I hope that’s worth something :-)
➡️ If you are interested, please let me know by replying to this email. Thank you in advance!
Note: We want to publish the collected data (in an anonymized way), but that is optional. If you are willing to contribute but your organization opposes the data sharing, we won’t publish it, and we will sign NDAs if necessary.
Thank you again for reading! Looking forward to hearing from you, -- Romain JACOB Postdoctoral Researcher ETH Zurich Networked Systems Group (NSG) Lead: Prof. Laurent Vanbever www.romainjacob.nethttps://www.romainjacob.net/ @RJacobPartnerhttps://twitter.com/RJacobPartner @jacobr@discuss.systemshttps://discuss.systems/@jacobr Gloriastrasse 35, ETZ G81 8092 Zurich +41 7 68 16 88 22
Dear Urs,
Let me keep the list on to reply to some questions, as you are not the first to ask:
* Most routers we are interested in have multiple power supply. Our strategy is to use one measurement unit per PSU. That demands extra hardware, but we wouldn't want to have two PSU connected to the same power meter anyway as this would create a potential single point of failure. * Network connectivity for the Pi is required but *does work behind a firewall*. The only requirement is that the Pi must be able to connect to the Internet.
Thanks again to those who already reached out!
Best,
Am 2024-06-21 16:48, schrieb Romain Jacob via swinog:
Dear Urs,
Let me keep the list on to reply to some questions, as you are not the first to ask:
- Most routers we are interested in have multiple power supply. Our
strategy is to use one measurement unit per PSU. That demands extra hardware, but we wouldn't want to have two PSU connected to the same power meter anyway as this would create a potential single point of failure.
- Network connectivity for the Pi is required but DOES WORK BEHIND A
FIREWALL. The only requirement is that the Pi must be able to connect to the Internet.
Thanks again to those who already reached out!
"the Internet" - you mean Port 443, any IP?
Dear Rainer,
* Network connectivity for the Pi is required but DOES WORK BEHIND A FIREWALL. The only requirement is that the Pi must be able to connect to the Internet.
Thanks again to those who already reached out!
"the Internet" - you mean Port 443, any IP?
Actually, no. We use a custom port to establish an gRPC connection from the Pi to our server back at ETH. So, if your network has tight security rules, the only thing we really need in one open port open towards one IP address.
Best,
Hi Romain again
How would you like to have the desired counters? Say, you would receive power data for a defined timeslot and we have to watch for the counters in that same timeslot? Or a constant reporting of the data on the network device?
Regards, Urs
Von: Romain Jacob jacobr@ethz.ch Datum: Freitag, 21. Juni 2024 um 16:47 An: Müller Urs (IT-PTR-CEN1-SL4) urs.bf.mueller@sbb.ch Cc: Vanbever Laurent lvanbever@ethz.ch, swinog@lists.swinog.ch swinog@lists.swinog.ch Betreff: Re: AW: [swinog] Research project on Sustainable Networking
Dear Urs, Let me keep the list on to reply to some questions, as you are not the first to ask:
* Most routers we are interested in have multiple power supply. Our strategy is to use one measurement unit per PSU. That demands extra hardware, but we wouldn't want to have two PSU connected to the same power meter anyway as this would create a potential single point of failure. * Network connectivity for the Pi is required but does work behind a firewall. The only requirement is that the Pi must be able to connect to the Internet. Thanks again to those who already reached out! Best,
Hello again Urs,
How would you like to have the desired counters?
Say, you would receive power data for a defined timeslot and we have to watch for the counters in that same timeslot? Or a constant reporting of the data on the network device?
Yes, we need to match the traffic counters with the power values we measure over the same time interval. We do not need the counters live though, it's no problem if we get them after the measurement is completed.
So far, we worked with counters collected every 5 minutes (the SNMP default as far as I know). The frequency of the counter data does not matter much, but the finer-grained the better for us. If you are already collecting this data somehow, it is most likely enough for our needs.
Best,
Hi Romain,
and are you interested in some kind of environment? different type of devices? vendors? etc? you can ping me to discuss more in details.
thanks & regards Martin
Sent from Outlook for iOShttps://aka.ms/o0ukef ________________________________ From: Romain Jacob via swinog swinog@lists.swinog.ch Sent: Friday, June 21, 2024 8:17:59 PM To: Mueller Urs SBB CFF FFS urs.bf.mueller@sbb.ch Cc: Vanbever Laurent lvanbever@ethz.ch; swinog@lists.swinog.ch swinog@lists.swinog.ch Subject: [swinog] Re: Research project on Sustainable Networking
Hello again Urs,
How would you like to have the desired counters?
Say, you would receive power data for a defined timeslot and we have to watch for the counters in that same timeslot? Or a constant reporting of the data on the network device?
Yes, we need to match the traffic counters with the power values we measure over the same time interval. We do not need the counters live though, it's no problem if we get them after the measurement is completed.
So far, we worked with counters collected every 5 minutes (the SNMP default as far as I know). The frequency of the counter data does not matter much, but the finer-grained the better for us. If you are already collecting this data somehow, it is most likely enough for our needs.
Best, -- Romain
Regards, Urs
Von: Romain Jacob jacobr@ethz.chmailto:jacobr@ethz.ch Datum: Freitag, 21. Juni 2024 um 16:47 An: Müller Urs (IT-PTR-CEN1-SL4) urs.bf.mueller@sbb.chmailto:urs.bf.mueller@sbb.ch Cc: Vanbever Laurent lvanbever@ethz.chmailto:lvanbever@ethz.ch, swinog@lists.swinog.chmailto:swinog@lists.swinog.ch swinog@lists.swinog.chmailto:swinog@lists.swinog.ch Betreff: Re: AW: [swinog] Research project on Sustainable Networking
Dear Urs,
Let me keep the list on to reply to some questions, as you are not the first to ask:
* Most routers we are interested in have multiple power supply. Our strategy is to use one measurement unit per PSU. That demands extra hardware, but we wouldn't want to have two PSU connected to the same power meter anyway as this would create a potential single point of failure. * Network connectivity for the Pi is required but does work behind a firewall. The only requirement is that the Pi must be able to connect to the Internet.
Thanks again to those who already reached out!
Best, -- Romain
On 21/06/2024 16:24, Mueller Urs SBB CFF FFS wrote:
Hello Romain
Could you please go into more detail concerning the phrase:
the Pi needs to be connected to a publicly reachable network via RJ45
Does this mean, you want to be able to remote to this pi? Or does the pi only <send> data and it would be ok to have it behind a firewall?
How about router with dual power supply?
You may also contact me off list for more informations, perhaps we may share such data.
Kind regards, Urs
SBB AG
Poststrasse 6, 3072 Ostermundigen
urs.bf.mueller@sbb.chmailto:urs.bf.mueller@sbb.ch / www.sbb.chhttp://www.sbb.ch/
Von: Romain Jacob via swinog swinog@lists.swinog.chmailto:swinog@lists.swinog.ch Datum: Donnerstag, 20. Juni 2024 um 17:00 An: swinog@lists.swinog.chmailto:swinog@lists.swinog.ch swinog@lists.swinog.chmailto:swinog@lists.swinog.ch Cc: Vanbever Laurent lvanbever@ethz.chmailto:lvanbever@ethz.ch Betreff: [swinog] Research project on Sustainable Networking
Dear all,
My name is Romain; I’m a researcher at ETH Zürich working with Laurent Vanbever. I have led projects on sustainable networking for about 2+ years now.
One important challenge the academic community currently faces is the lack of quality data for the energy a router or switch consumes. To address this, we work on fine-grained power modeling of network devices, and we need data to validate their accuracy “in the wild.” To facilitate that, we worked to realize the vision of “RIPE Atlas for power data” I pitched at last winter’s RIPE meetinghttps://ripe87.ripe.net/archives/video/1143/. And I’m happy to say that we are ready to start distributing measurement units!
Concretely, we are looking for people willing to
1. Install measurement units in series of each PSU of routers and switches. The measurement unit is an MCP39F511Nhttps://www.microchip.com/en-us/development-tool/ADM00706 power meter combined with a Raspberry Pi for management and data collection:
[cid:part1.MCgTQf2B.gRlit8Sc@ethz.ch]
We will provide the appropriate cabling. To control the unit, the Pi needs to be connected to a publicly reachable network via RJ45 (WiFi connectivity is technically possible as well, but requires additional setup)
1. Share device-specific data with us, including at minima
* Packet and/or byte counters for the active device interfaces. * Types and numbers of transceivers plugged in.
and optionally
* PSU-internal measurements * Operating system version * Fan speeds * CPU load * Average room temperature conditions
1. Let the measurement run for at least one week. 2. Send us back the measurement units (we provide pre-labeled packaging for that).
In return, we can only offer our sincere gratitude and the guarantee that your effort will contribute to the progress of science! I know it s not much, but I hope that’s worth something :-)
➡️ If you are interested, please let me know by replying to this email. Thank you in advance!
Note: We want to publish the collected data (in an anonymized way), but that is optional. If you are willing to contribute but your organization opposes the data sharing, we won’t publish it, and we will sign NDAs if necessary.
Thank you again for reading! Looking forward to hearing from you,
--
Romain JACOB
Postdoctoral Researcher
ETH Zurich
Networked Systems Group (NSG)
Lead: Prof. Laurent Vanbever
www.romainjacob.nethttps://www.romainjacob.net/ @RJacobPartnerhttps://twitter.com/RJacobPartner @jacobr@discuss.systemshttps://discuss.systems/@jacobr
Gloriastrasse 35, ETZ G81 8092 Zurich +41 7 68 16 88 22
--
Romain JACOB
Postdoctoral Researcher
ETH Zurich
Networked Systems Group (NSG)
Lead: Prof. Laurent Vanbever
www.romainjacob.nethttps://www.romainjacob.net/ @RJacobPartnerhttps://twitter.com/RJacobPartner @jacobr@discuss.systemshttps://discuss.systems/@jacobr
Gloriastrasse 35, ETZ G81 8092 Zurich +41 7 68 16 88 22
-- Romain JACOB Postdoctoral Researcher ETH Zurich Networked Systems Group (NSG) Lead: Prof. Laurent Vanbever www.romainjacob.nethttps://www.romainjacob.net/ @RJacobPartnerhttps://twitter.com/RJacobPartner @jacobr@discuss.systemshttps://discuss.systems/@jacobr Gloriastrasse 35, ETZ G81 8092 Zurich +41 7 68 16 88 22
Hi Romain,
I just looked at the power data our PDUs show (Riedo uPDU, Swiss product) and didn't find any correlation (<1%) to the bandwith consumed. So I guess, the consumption is mostly dependent to the equipment installed and not by the bandwidth consumed.
Regards,
Claudius
PS: Contact off-list for more.
On 6/20/24 17:00, Romain Jacob via swinog wrote:
Dear all,
My name is Romain; I’m a researcher at ETH Zürich working with Laurent Vanbever. I have led projects on sustainable networking for about 2+ years now.
One important challenge the academic community currently faces is the lack of quality data for the energy a router or switch consumes. To address this, we work on fine-grained power modeling of network devices, and we need data to validate their accuracy “in the wild.” To facilitate that, we worked to realize the vision of “RIPE Atlas for power data” I pitched at last winter’s RIPE meeting https://ripe87.ripe.net/archives/video/1143/. And I’m happy to say that we are ready to start distributing measurement units!
Concretely, we are looking for people willing to
Install measurement units in series of each PSU of routers and switches. The measurement unit is an MCP39F511N https://www.microchip.com/en-us/development-tool/ADM00706 power meter combined with a Raspberry Pi for management and data collection:
We will provide the appropriate cabling. To control the unit, the Pi needs to be connected to a publicly reachable network via RJ45 (WiFi connectivity is technically possible as well, but requires additional setup)
Share device-specific data with us, including at minima
o Packet and/or byte counters for the active device interfaces. o Types and numbers of transceivers plugged in.
and optionally
o PSU-internal measurements o Operating system version o Fan speeds o CPU load o Average room temperature conditions
Let the measurement run for at least one week.
Send us back the measurement units (we provide pre-labeled packaging for that).
In return, we can only offer our sincere gratitude and the guarantee that your effort will contribute to the progress of science! I know it s not much, but I hope that’s worth something :-)
➡️ If you are interested, please let me know by replying to this email. Thank you in advance!
Note: We want to publish the collected data (in an anonymized way), but that is *optional*. If you are willing to contribute but your organization opposes the data sharing, we won’t publish it, and we will sign NDAs if necessary.
Thank you again for reading! Looking forward to hearing from you,
-- Romain JACOB Postdoctoral Researcher ETH Zurich Networked Systems Group (NSG) Lead: Prof. Laurent Vanbever www.romainjacob.net https://www.romainjacob.net/ @RJacobPartner https://twitter.com/RJacobPartner @jacobr@discuss.systems https://discuss.systems/@jacobr Gloriastrasse 35, ETZ G81 8092 Zurich +41 7 68 16 88 22
swinog mailing list -- swinog@lists.swinog.ch To unsubscribe send an email to swinog-leave@lists.swinog.ch
-
Claudius Zingerli -
Darragh Grealish -
Kostial Martin -
Mueller Urs SBB CFF FFS -
rainer＠ultra-secure.de -
Romain Jacob