IGF 2021 - Day 1 - Lightning Talk #44 Green data processing for sustainable growth

The following are the outputs of the captioning taken during an IGF virtual intervention. Although it is largely accurate, in some cases it may be incomplete or inaccurate due to inaudible passages or transcription errors. It is posted as an aid, but should not be treated as an authoritative record.



>> WOJCIECH STRAMSKI:  Hello, everybody, I hope you can hear me.  This is Wojciech Stramski.  I'm honored to be speaking to you today as the CEO of Beyond.pl. 

     I represent one of the most sustainable data centers in all of central Europe and I will be walking you through how green data centers services can help us drive sustainability today.

     So we are very, very lucky we are living in very good times.  Apart from COVID, we are actually living in very exciting times.  But technological transformation is undergoing an extremely rapid pace of change, and we are a witness of that today.

     The key drivers of digital transformation are technological advancement with the introduction of technologies like IoT, AI, blockchain, machine learning and 5G.  And this is all leading concurrently to a growing population of internet users.  Only 4.6 billion of the world's population uses the internet today.  That is expected to grow to two-thirds of the global population by 2023.  And that is still not enough.  We need to do more to ensure that the greater population is actually included in this digital transformation.

     And for this to happen, we need to have and ensure that the entire global population can benefit from technology.  We need to deploy more data centers, including edge facilities.  And this is actually very, very important. 

     Estimates point that thousands if not tens of thousands of edge facilities located physically close to the end user will need to be deployed to ensure that artificial intelligence, machine learning, real-time decisions are pushed to the users of the internet where they are. 

     Living in these times also means we need to take responsibility that comes from the decisions we make today and the impact of these on future generations to come.

     The global data center electricity demand today makes up about 1% of the global total electricity demand.  It is estimated that between 200 and 250 terawatt hours of electricity are drawn by the data center industry.  And actually, another 100-terawatt hours by crypto. 

     In only 10, years, and this is actually important, this is expected to grow anywhere from 4 to 20 times when global electricity demand will be about 3 to 11% taken up by the data center industry.  And who knows how much more crypto will actually take then. 

     And so when we are going through this digital transformation, we have to ask ourselves, what exactly is the cloud?  And how do we want the cloud to be in the future?  Is it going to be just another cloud of CO2 in the atmosphere?  Or is it going to be something that is actually going to be beneficial to us?

     And when we are deploying this digital infrastructure, we need to ask ourselves the question, how do we measure and ensure the energy efficiency of that infrastructure.

     And just like we deployed measurements and tools to measure other industries, we must introduce transparent and default standards for the data center industry.

     And the good news is we actually already do have this measured.  It is called PUE.  And so what is PUE?  PUE, or power usage efficiency, is actually a simple metric used to measure the ratio of total gross power drawn by a data center to power the IT infrastructure co-located in that data center. 

     And for a very easy and simple illustration example, if servers and storage equipment require 100 kilowatts to be powered, then the question is, how much additional energy is required to power the supporting infrastructure of that data center?  The cooling equipment, the lighting, the generators, the UPS's, the security infrastructure. 

     And in this very simple example, if 100 kilowatts is drawn by servers and storage and another 50 kilowatts to power the supporting infrastructure, the data center in this example has a PUE of 1.5.

     And so the lower the PUE ratio, the more energy efficient the data center.  Because the more energy goes to pool power the servers and the storage equipment co-located in that facility.  And so measuring PUE actually makes a lot of sense from an environmental perspective.  If you take a customer that has a requirement of 10 megawatt hours, that customer can save 25,000 tons of CO2 per year just by drawing from the PUE ratio of 1.1 versus a PUE ratio of 1.5. 

     PUE also makes sense from a business perspective.  A customer that requires one megawatt of IT load can save 600,000 Euros per year assuming an energy price of 20 cents Euro per kilowatt power, which is quite low when you look at where we are today on the global map of energy prices. 

     Just by drawing from a data center that has a PUE of 1.1 versus one that has a 1.5.  And PUE is actually something that is very much worth fighting for.  Total worldwide carbon emissions make up about 1%.   1% in terms of data center worldwide carbon emissions contribution.  350 million tons of CO2 per year today. 

     That is expected to grow to 1.1 billion tons of CO2 emissions in 2025.  And actually 6 billion tons of CO2 by 2040. 

     When you think about it, the data center industry is actually growing tremendous, tremendous development and is one of the few industries, if not the industry, experiencing the greatest growth of demand in energy take-up and CO2 emissions in the next 20 years.

     And so ensuring that the industry can decrease from the levels of PUE 1.5 to 1.1 can help us save anywhere from two to three billion tons of CO2 emissions per year.  That is huge.  And we really have to stand up and fight for this.

     The good news is the work on driving sustainability in the sector has already started.  The European Commission appreciates the requirement to digitize all aspects of the economy.  However, it is key that the transformation that is happening will not endanger the overarching goal of reaching climate neutrality. 

     And the data center industry has actually come together, big global players and small players, local players have actually entered into a pact where the data center members have pledged to become climate neutral by 2030.  And this will help the EU meet its goal of becoming climate neutral by 2050. 

     And the good news is that sustainable solutions are actually possible, and they are actually possible in the least expected of places.  Poland, which is not known on the global map to be the frontrunner in green energy and applying green solutions having a power mix of 80% coal today, 20% renewable.  We can actually be very proud in Poland to have deployed one of the best energy efficiency standards for data center infrastructure in all of Europe.  We have a functioning data center that has a PUE of 1.2 versus the norm on the geographical latitude of 1.4 to 1.6. 

     And how have we done this is Beyond.  We are today an 8 and a half megawatt campus located in Poland.  We have commissioned the infrastructure in 2016.  We are actually one of the three, only three data centers in the entire European Union, to have the best rating in terms of rated for certification from the agency on CTO 9.2 which allows us to deliver services at the best quality. 

     There is only three such facilities in all of Europe.  Currently the infrastructure is powered 100% by green energy.  And it's actually delivering a PUE of 1.2 versus the norm of 1.4 to 1.6.  And how we have done this is we've actually invested up front significant funds to make sure that the infrastructure is sustainable.  We've introduced technologies like adding back cooling technology which helps decrease the amount of energy we draw to cool the servers co-located in our facility. 

     And we have the heat that is generated in the chambers to heat the office facilities, the logistics buildings.  We rolled out a data center management system to manage energy efficiency.  And we have actually deployed E24 cloud which is the first Polish cloud solution on the market which allows our customers to better leverage the assets that are co-located here. 

     And probably something that we are most proud of the last two years, we've actually pushed our vendors and procured best in class service and storage equipment that is also energy efficient.  Not only do we require less energy to power the data center infrastructure, we also require less energy to actually power the servers and the storage equipment themselves. 

     And today we are working on an expansion project to grow this campus to 42 megawatts.  And alongside the expansion project we are working alongside stakeholders to ensure that this significant heat that will be generated from the data center will not be lost and will be recycled to the residential and commercial real estate surrounding the facility to provide heating of water and heating of facilities during the winter period.

     Another consideration to take is can we actually push digital transformation to be deployed in an Airbnb type of model.  If you look at the success of Uber and Airbnb, all of these technologies were deployed under the foundation that these asset classes like cars, apartments, hotels, office buildings, were not utilized to the best of their capacity.

     And so you need to think how can we do the same in IT?  Today there are millions of data centers which are deployed and owned adjacent to factories, within office buildings.  And we just have to as a society come together and understand this is not an efficient model.  There is no way that these facilities have PUE standards below 1.2.  Because it just doesn't make sense to make these investments.

     And so we need to go to these off-premise data centers where the mass of the customers or the users in these off-premise data centers can actually leverage the purchasing power of these customers and allow us to deploy these energy efficient solutions.

     Another thing to think about is actually what we are witnessing now with the cloud transformation.  It's allowing us to transition from a buy-to-own model versus a service model around leveraging access to servers and storage equipment.  Today the majority of servers and storage equipment in a buy-to-own model are utilized at 25 to 30, maybe 35% capacity.  So that there is ample room to leverage off that and that is why you are seeing this transition into cloud.

     Another consideration for the industry is how can we push digital transformation in a circular economy type of a model.  I mentioned what we are doing with Beyond working with the stakeholders where the 42 megawatts used to cool the data center infrastructure, the servers, the heat generated by the servers in turn will be used to heat the residential and commercial facilities that are located adjacent to the facility.

     Can we actually push -- and this is something very important as well -- can we actually push for more recycling in the economy?  All the servers that we are building and rolling out today, all of the storage equipment is actually a huge contributor to the minerals and mining space when we need to mine various resources to continue to build these.  We need to recycle the existing infrastructure and recoup and recover the minerals to minimize the footprint on the mining side.

     So just to conclude here with a call to action.  We all have a role to play to ensure that digital transformation is carried out in a sustainable way.  The data center operators and vendors, we need to be transparent.  And we need to deploy sustainable solutions.  We should stop focusing on short-term gains and look at the big picture and push innovations in this space to deliver climate neutrality. 

     Consumers and end customers need to be positioned to make informed purchase decisions.  They need to be made aware of their carbon footprint and in turn be demanding of solution providers to deploy solutions in the most sustainable way. 

     To the government and regulator representatives, we need to drive legislation in the industry to force sustainability.  And we need to encourage consumers and clients of the industry to purchase sustainable.  And for banks and financial institutions, this is actually pretty easy, we just need to stop providing financing and capital to projects which cannot demonstrate sustainability.

     And last, but not least, digital transformation is not possible without all of the innovators globally taking part in this transformation.  And the innovators, as the drivers of this digital transformation, they need to make sure that when they are choosing the solution how it will be rolled out, where it will be hosted, that they do this with sustainability in the back of their mind.

     Thank you very much for your attention.  And I'm happy to take any questions you may have.

     >> AUDIENCE: Thank you very much.  I have a question like can you explain in more detailed way what is the important action or actions to achieve PUE below 1.2?  You mentioned some ways, but if it is possible to explain like in more detailed way what are the actions?

     >> WOJCIECH STRAMSKI: Sure.  I think you have to look at it from -- you have to start when you actually design the data center.  That's the place where you start. 

     And you have to envision where you want this data center to be, and you have to think about where you are in the geographical coordinates on the global map.

     If you are deploying a data center in a desert, you will need more energy to cool the infrastructure during the day and potentially less during the night when it gets cold.  Similarly, if you are in the north in the Scandinavian market you may utilize the natural resources available to you to provide the cooling. 

     One of the biggest issues in the data center is to ensure that the temperature in the chambers and server rooms is maintained at an optimal level to ensure that the servers continue to run.

     And so you have to minimize the overheating.  And then there are multiple solutions that you can deploy from introducing obviously the source of the energy into the data center, whether it is green or not green, that's one.  Even if it is green 100% and your PUE output is 1.5 or 1.6, that is not efficient.  What is available to you in terms of the natural resources available.  What is beyond the cooling.  That is nothing more than utilizing the natural circulation of air to provide cooling to the data centers. 

     When you provide this natural enabler we use less energy to draw on the data center -- to actually power the cooling equipment.  We don't need as much air conditioning to cool the equipment.

     The other element is what I mentioned is this heat that is being generated in these chambers.  So when you think about, you know, hundreds of servers running in a room, it really gets warm in there.  So if you can actually utilize the heat somehow and reuse that heat in other areas of the facility.  So recycle that heat.  I mentioned to you the campus.  We have another data center in the city of Poznan, a small edge facility.  And that is the future of the world, more edge facilities will come. 

     We deploy the heat generated by the data center to warm the shopping center, the stores and the common area so they use less energy to heat the facilities.

     So if you think about what you can do to utilize the natural resources that are available to you whether, you know, it is all this interesting technologies coming up that pumps the air-to-air cooling ground water to air different pumps cooling, deploying heat, reusing, recycling that heat in the infrastructure, that all will drive down the footprint of the data center itself.  I don't know if that answered the question.

     >> AUDIENCE: I have a question about the number you mentioned during your presentation. 

     You said that in 2040 data centers will be responsible for I think 14% of global electricity consumption.  Is this correct?

     >> WOJCIECH STRAMSKI: 14%, yes.  14% actually of emissions.

     >> AUDIENCE: Okay.  And I guess that emissions will go up year by year until 2040.

     >> WOJCIECH STRAMSKI: This is correct, yes.

     >> AUDIENCE: So I guess because of the efficiencies you are trying to achieve will be slower like the increase will be slower than it is, but it is still not enough, right? 

     So I guess because you say you talk a lot about from your perspective obviously that 1.1 of PUE is a very good result.  But maybe, and I don't know if the industry looks in it in this way, maybe we need also to look at the data you are processing in the data centers.  Maybe there should be something.


     >> AUDIENCE: That should encourage your users, you know, to use less data, less space and less carbon emissions.

     >> WOJCIECH STRAMSKI: 100% I agree.  And this comes down to I think making the users actually aware of the CO2 emissions. 

     If you introduce a standard way to actually measure the emissions, if you are a user of Facebook or Netflix or whatever it is that you are actually a user of, do you know how much CO2 you are contributing to as an individual?

     So I think this comes down to, first, making sure there is a standard which is actually deployed and measured by the industry.

     Okay.  Making them the consumers and the customers aware so whether it is a company, or whether it is an end user, an individual so that they understand how much they are contributing to that carbon footprint.  And I completely 100% agree with you. 

     Getting to 1.0 in terms of data industry is probably not enough because, like you said, the CO2 will continue to grow at an alarming rate.  I think the other side of the equation is how much benefit are we providing to society by deploying digital solutions. 

     So if you look at e-learning, if you look at medicine, if you look at the pharmaceutical industry, if you look at financial industry, if you look at developing countries and providing users access to mobile and giving them access to the internet and to many, many solutions which were not available to them, you kind of have to take the view that you are probably making their life better. 

     So that digital transformation is not something we're going to go away from.  That increase in CO2 emissions is probably not something we are going to go away from unless we define the standards what we require in the industry.  And we can do more to minimize and to go back to your question, to make people aware of that.  What solution they are using and how much they are contributing individually, as a collective, as a society by measuring a few emissions and cars and airlines and any other industry that is being measured.  I think I see a question from Wathagi.

     >> WATHAGI NDUNGU: Yes, thank you very much.  Like should this be something that the specific industry consider when coming up with the smart city hubs?  And how do they ensure that they reduce and at the same time ensure that the security measures are in place because it is most of the biggest problem with the EU I saw that so many security measures are required to be, especially in the regulations like the GDPR.  So how do you think of this concept would apply in the privacy side?

     >> WOJCIECH STRAMSKI: Yeah, so I think that is a very good question. 

     I think this comes back to the concept of where we are today.  If you look at the map of data centers, the majority of the data center infrastructure the large infrastructure today that's been built out is actually regional data centers.  Now from the regional data centers, you cannot actually deploy efficient smart city solutions.  If you think before, the emergency response you need the data centers in the city so the AI and all working to power the emergency response system are in real time and they cannot be affected from a latency cap.  Requiring the concepts will require the majority of these cities to have edge data center infrastructure.

     Now, if this edge data center infrastructure knowing that you have to deploy this edge data infrastructure, you need to ensure that this is done, this investment is carried out to ensure sustainability.

     Okay.  So you cannot be deploying data centers and not allow for whoever is rolling it out, the private sector or public sector to allow data centers that are not -- depending on where you are on the global map.  But you need to do this.  You need to do this.

     And I think in terms of security, and access, obviously this is another important component.  You know, how will it edge data center facility look like?  One item is physical access in securing or limiting physical access.  The other element is obviously the cybersecurity element which today is becoming more and more important.  All that, even cybersecurity that is a consumer energy.  These solutions consume energy.

     So edge data center infrastructure will be critical to support smart city deployments and we need to make sure that that edge data center infrastructure is sustainable.  And if you are taking business decisions around this infrastructure today, we have to think 10-year, 50-year horizon.  So we're in 2020 today, so whatever infrastructure you are putting up today is going to be there for the next 10-15 years.  So you need to take that decision today to make sure that it is at the best PUE levels possible.  I don't know if that answered the question.

     >> WATHAGI NDUNGU: Yes, thank you very much.

     >> PIOTR KULAGA: I see a question from Johnathan.  Which regions will have the most growth in internet users and which regions will see the most energy use increase for data centers?  I see there isn't time. 

     I think you see the growth in internet users in African nations, in South America and Asia where actually access to the internet is being deployed via access to mobile networks.  So the biggest growth is in those regions and obviously there you will see it's the largest increase of data center energy use. 

     And I think just to conclude, I think one important view that we have to take as a global society is countries which are not the frontrunners of green and digital transformation, allocating into these centers and these regions where the country is not currently green will obviously not help drive sustainable transformation in the digital space. 

     So if you look at Asia, East Asia, big consumers of coal, the growth of internet users in these countries and obviously data centers that will predominantly be using coal energy to power themselves, I think that is the biggest risk we have.

     So thank you very much.  I hope -- I don't know if there is any other questions.  Probably an interesting topic we could speak quite long about.

     >> AUDIENCE: Hi.  Like you really concentrated on increasing the power efficiency by lowering the PUE. 

     So you emphasize on lowering the power consumption as well.  Have you ever planned or thought of using captive clean fuel generators such as fuel cells and all and use the cooling efficiencies by the heat shield heater because there are lots of industries have started using clean fuel generators, captive generators in their -- on the premises?  Thank you.

     >> WOJCIECH STRAMSKI: Thank you very much for that question.  I think so I think this is the call to action on the last slide.  Bringing innovation to the data center space.

     And I definitely am a firm believer that we just need to do more as an industry and there are technologies that we are watching. 

     I think for Beyond specifically, today we are still a relatively small mega center in terms of the 8 megawatts and going to 42.  That is something that we are looking at to look at in terms of what other technologies are available to us to ensure that we can be self-sustaining and green.

     We are 100% green already.  I would push this cycle of solutions as you mentioned to geographical locations again where deploying green data centers is not possible and to have them look into these kinds of solutions.  Definitely, I think we are in the very, very early stages of transforming this industry and we need to transform.  If there are no other questions --

     >> AUDIENCE: I have additional question.


     >> AUDIENCE: About how can you use AI solutions and some kind of algorithms to lower the usage of energy in terms of efficiency of some processes and also potentially in terms of energy usage?  So energy generation and how efficient is the process of analyzing the data?


     >> AUDIENCE: What would you think about like AI in this field and how -- what can be the role of AI?

     >> WOJCIECH STRAMSKI: I think there is actually -- thanks, that's a good question. 

     I think in the data center industry in particular there is actually quite a lot of players who are deploying AI solutions to help you monitor energy efficiency. 

     And this is on two fronts.  And I think one thing is, first of all, IoT is allowing you to actually measure the energy draw from -- you know, on a particular equipment level. 

     And then you can analyze the power draw on a particular equipment level, individual equipment level.  And obviously then you can run your AI algorithms to identify trends when the power drive is leading and what is happening is something that is happening internally in the data center.  Outside the temperature getting warm or the summer coming, and you are just drawing more energy to cool the infrastructure.

     But also on an individual, you know, server rack level, that is also possible now.  So we actually have a data center management system which is allowing us to analyze data power draw on a rack-by-rack level.  Obviously where the racks are physically placed in the chamber, how close to the cooling equipment.  Are they optimally placed?  If you have two or three racks that are very energy -- big energy consumers and then 20 which are not energy consumers do you have the flow of air in the chamber optimized for that?

     So it is about providing you a lot of information on a very, very pinpoint granular level to optimize floor planning, layout, air circulation, et cetera, et cetera.  So there is quite a lot of solutions already made available in the industry.  We do some solutions and then AI obviously the more data you gather the better you are at predicting the feature and driving our hopefully future investment decisions, right. 

     So obviously we know the data center space is growing, more chambers will be coming online.  Maybe we will then need to use this data, and I hope we do use this data to better design the facility of the future.  

     Okay.  I see the moderator is asking us to close.  I will be happy to take any questions.  There is an e-mail on the presentation as well.  So if there is any questions, I would be happy to connect.  This is something I'm passionate about so hopefully we can do more together as societies and different stakeholders and different roles that we play to make sure that the digital transformation minimizes the carbon footprint.  Thank you very much.