Adoption and Future Perspectives for the Cloud

#serverless #CloudComputing

Join Gojko Adzic, award-winning author and software delivery consultant, and Lynn Langit, cloud architect building serverless bioinformatics cloud data pipelines, to discuss the cloud ecosystem, exploring common and actual use cases and why there are different cloud adoption patterns in different regions across the globe as well as in different industries. They also touch upon what are the next cloud adopters, banks, state institutions, and where the cloud industry heading.

Intro

Preben Thorø: My name is Preben Thorø. I'm currently at GOTO Berlin. I managed to bring Gojko Adzic and Lynn Langit together here at our event. It's not the first time you meet each other. Right?

Lynn Langit: No, I think the first time was in Sydney, Australia.

Gojko Adzic: Yes, Sydney.

Lynn Langit: Yes, you were having a presentation and you had a co-presenter.

Gojko Adzic: Yes, my son was there because I kind of thought maybe it would be interesting for somebody else to do all the difficult work. So I brought him on to do a live coding demo. It was his first presentation. He was nine years old. Nobody told him that coding demo's risky on stage. So he just went and did it. I think it was quite the show.

Lynn Langit: Yes, I remember, he deployed several times using Claudia.js serverless-y, and his output went from, "I hate Vim," to, "I really hate Vim."

Gojko Adzic: Vim hatred is interesting, because I forced him to use Vim, and he didn't really like it. So yes, we'll see how he grows up.

Should every company have a cloud strategy?

Preben Thorø: We just talked about cloud today. I know cloud is a very broad term. It has been around for many, many years. It has probably changed over these years from what it used to be and what it is today. But there still is this feeling that cloud is some magic thing. Being a modern company, you should move into the cloud, whatever that means. Can every company do that? Should every company have a cloud strategy?

Lynn Langit: I think the perspective is going to depend on where you're trying to work. Most of my work is done in the United States. Until previously, it was the West Coast, United States where basically most of the currently prevalent cloud providers originated. So when I think of when the cloud movement started I was still a Microsoft employee. So in my world, as a cloud architect, I have only deployed on the cloud for every vertical for over five, seven years. Now, I have colleagues that work in different areas in Europe and the Middle East. They tell me about scenarios that are not appropriate for the cloud. But in my work, I only use cloud. How is it for you?


Gojko Adzic, Lynn Langit & Preben Thorø interview

Gojko Adzic: I'm going to step back a bit and maybe look at this from a different perspective because I think the right way to look at this is the way power was distributed a long time ago. So back in the time of boilers and steam power, every large factory had their own boiler room, they had belts and shafts and transformed steaming to mechanical movements. Then electrical energy came along. It was possible to just plug-in and get that from somebody else. I think that allowed lots of people to focus on what they're really good at. So a shoe factory could focus on shoes, rather than producing servicing a boiler, and wasting time on that. 

That gave people economies of scale, it gave both electrical producers economies of scale to produce electricity cheaper, and maybe cleaner and things like that. It also allowed other people to focus on what they're really good at.

Recommended talk: GOTO 2017 • Designing for the Serverless Age • Gojko Adzic

I think we are at the point in our industry where that shift is happening. Maybe 20 years ago, in order to really be a good software company, you had to have your own operations, you had to have your own hardware people, you had to have your own Linux admins or Microsoft admins if you were unfortunate, and you had to get all sorts of stuff there. But then we are starting to get now into a situation where you don't actually have to do that, you can focus on what you are really good at. For me, that's one of the biggest advantages. I'm currently building a product with a colleague, and there are only two of us in this whole company. We are able to compete with companies that have maybe 200 or 300 employees because we do not have to do anything that's not really our core business.

So for me, that's kind of the advantage of the cloud- thinking about plugging into an electricity network and just using that as utility, because my core competency is not going to be producing electricity. My core competency is going to be doing something else. I think maybe 10 years from now, or 20 years from now, most organizations are going to realize that as well. Some organizations are realizing that now, and I think it took probably a long time to get all the steam boilers out and get the early generators out, and we still have companies that, as a backup, have their own generators, just in case the power grid fails. For example, hospitals have that. Police probably have that. An average company might have some backup for a couple of hours.

So maybe that's an interesting thing to think about. What...if this became something like a public utility, what would you have to do to remove the risks of the public utility going down? And how much do you actually need to risk it? So the equivalent of a hospital or something like that would be maybe some highly secure thing that really operates critical infrastructure. In that case, I would expect them to have a backup plan. But for an average factory, or an average company, where they can survive an hour of electricity going down without something horrible happening, maybe there's no huge risk of going to the cloud. 

I think that's more of a risk mitigation, risk evaluation strategy. So can anybody do it? Probably. Should everybody do it? I don't know, that's really depending on the risk and how do you de-risk your infrastructure provider not being there for you when you need it. But at the same time, there are significant benefits of not wasting time on doing things that are not people's core business.

Unlikely users of the cloud

Preben Thorø: But still, you mentioned companies that don't find it appropriate to go to the cloud.

Lynn Langit: Yes, I have a colleague that works in the Middle East, and he works for the government of the Middle East. After doing some evaluation, I believe, they went with the Oracle Cloud, because it's deployable in a data center with more control. So it was really an interesting conversation that I had with him about this because, in my world, the way that we have redundancy is we have multiple cloud providers. But that's not available in all parts of the world. So for his situation, to have redundancy with some part in the public cloud and some part in a private cloud, they went with this blended solution, which is again something that hasn't been important for my customers. It really depends on which providers are available and where I think. That's been my situation.

Gojko Adzic: That's something that will change over the next 10, 20 years. The example you mentioned is a large government, whether it's better for the taxpayers of that government to pay Unix admins or Microsoft admins, and hardware and replace hardware every couple of years, or whether it's better for them to rent it from someone else that really depends. At some scale, it's justifiable for companies to operate their own electricity. I assume for very, very large factory complexes, it's probably cheaper to make their own electricity, either from solar power or from, reverse or something like that. It's some kind of small or medium scale that makes no sense at all. So I guess there will be a shift there with very, very large banks, there was a report, for example, from Bank of America a couple of days ago, where they claimed that for them, it's $2 billion a year more economical to operate their own kind of flexible infrastructure than rent it from somebody else. Maybe. I don't know. At some scale, that will make sense.

At the same time, I expect that people who specialize in providing IT hosting, or hardware hosting, software hosting, can iterate on that and reduce the costs quite faster than a bank would. So although it might be that today it's cheaper for somebody at the scale of Bank of America to run their own stuff. In five years time, that might change because these cloud providers have their own patents on data center cooling, they're investing in dumping data centers into lakes, and frozen rivers and things like that, where they can get to that because they have economies of massive, massive scale, which they can provide. 

Recommended talk: GOTO 2019 • Serverless: Five Key Things you need to Know • Gojko Adzic

I think the critical thing that lots of people still feel unconfident about, uncomfortable about is the security of data, and encryption of data, and how much you want to kind of share the data. Again, there are some examples where, for example, Amazon got so big now that their competitors are uncomfortable giving them their business data.

And there are examples, I don't know, Walmart in the U.S. insisting that none of their suppliers have data in Amazon, because Walmart doesn't want Amazon to be able to access anything there. So I think the challenges that people are facing migrating to the cloud are no longer technical, they're kinder security-oriented or privacy-oriented. And that's a completely different game. That's more about where and how we manage and store the data.

Preben Thorø: But given that, do you have examples of companies that you wouldn't think we’re in the cloud, but still are?

Lynn Langit: Yes, I have a lot. We were just watching the Amazon conference because they're going to put unexpected examples in their keynotes. Capital One is one of the big banks in the United States, and they're all in on the cloud for quite a long time. Now, quite embarrassingly, though, they had a big security breach, not because of an Amazon problem. But because of what Gojko was talking about, which is one of the blockers to cloud adoption is not only not trusting the security of the provider, it's not understanding the security paradigm. One of the things I do in addition to being an architect is I make courses on LinkedIn Learning. And one of the areas that I've really focused on, the state of security because as an architect, I find that there's just not enough focus put on this area. So you have first this trust problem, and then you have a lack of knowledge and sort of understanding, and it confounds the problem. So, it's kind of a side answer to your question. There are lots of infrastructure companies that I've seen from the U.S., like banks on the cloud, but it's not without challenges.

Global footprint if everything moved to the Cloud

Preben Thorø: Do you have any idea about what the global footprint would be if everything could be moved into the cloud?

Gojko Adzic: Footprint in what sense? Like climate or...?

Preben Thorø: Yes, climate.

Gojko Adzic: Well, again, my assumption is that it's such a big economy of scale, cloud providers can spend significantly less energy per unit of delivered output, whatever that means, or is measured, because they just can have much cheaper cooling, many kinds of more efficient energy processes and things like that, where they can compare to, especially smaller organizations or even midsize organizations that would have to buy commercial, off the shelf hardware, these people can do custom stuff that's specialized for what they need. I don't have any insight or data on how these people are doing it. There are lots of press releases now because climate change is a big topic, people claiming that they're 50% renewable or 100% renewable. There's no way for me to verify that. But I think it's going to be much easier for large providers to do that than for smaller organizations that have to buy electricity. It's a kind of retail price and buy hardware at retail prices and buy commodity hardware, rather than doing specialized things. I'm really not an expert in that. But there's a reasonable assumption that they can do it more effectively.

Gojko Adzic, Lynn Langit & Preben Thorø interview

Lynn Langit: So it's been really interesting. I was pretty ignorant in this area. But recently, I discovered ex-Amazon employee, Paul Johnson, who now has taken up the cause of educating people about the climate impact of the cloud data center vendors, and he's started to do talks about it. 

Recommended talk: GOTO 2020 • Visualizing Cloud Systems • Lynn Langit

What's really been fascinating to me, I knew from reading some of the company press releases that Google was really putting more of a focus on this and purchasing carbon offsets. But you always have a bit of skepticism when the company puts it out. So here's this guy, Paul. And he's now been doing keynote talks at technical conferences, and really looking at the data that's available, and analyzing it. It appears that Amazon has work to do in this area, based on his information, which I'm just really reporting. I haven't dug into it myself. But I applaud efforts like his. I hope that he encourages not only more people to look into the data and to talk about it, but also the companies to be more responsible. But again, I would agree with Gojko that, so long as we can get the companies focused on this through public pressure, that it's sort of logical that these large companies would probably be more effective than individual companies trying to have their own data centers. I'm actually pretty pleased with what I'm seeing from Google, which, given some of the other activities of Google, makes me happy. They're doing something good.

What comes after the Cloud?

Preben Thorø: Let's look into the future. What comes after cloud?

Gojko Adzic: Well, it depends how you define cloud because cloud started as renting virtual machines from somebody, and then it became actually renting more infrastructure services like storage or databases, and now we're seeing pretty much like application hosting and things like that. If I look at what Salesforce was selling as cloud in in 2001, 2002, I think that's where the word cloud originated. Salesforce was selling something like, "We'll give you this business capability, we'll give you a customer relationship management system. Don't worry about where, how, we will run it for you." 

We're now seeing higher and higher-level applications being launched by these cloud providers. I've noticed that Amazon launched their own contact center application, Microsoft is launching higher and higher-level services that you can rent from them. So it looks like we're kind of coming back full circle where cloud started from, "Here's this business capability, rent it from us," to, "Hey, this is very, very low level technical, rent it from us." And then kind of high-level technical. Now we're getting to the stage where, again, "Here's this business capability, rent it from us."

That's an interesting cycle. So I think, if anything, our industry moves in cycles. I would expect that after this, we start getting another cycle of, "Here's some more crazy technical capability, rent it from us." 

Then that's based on some other way. I don't know what that is. But I would assume that that's kind of the cycle that we get, being able to rent different stuff, like rent GPUs, rent... I know you're doing a lot with machine learning, but rent machine learning platforms from us or something like that. That would be an interesting thing to see.

Preben Thorø: Couldn't it get crazier: the quantum computing in the cloud?

Lynn Langit: Yes, things are interesting. In the world of quantum, I actually had an opportunity, just through kind of random chance, one of the big quantum computer companies based in Vancouver, Canada, and it was in the airport. I was with a very aggressive programmer friend, and we saw the D-Wave people, and we engaged with them. Actually, many years ago, we got access to the quantum simulator. So that's when I got interested in quantum computing as a potential solution. I've tried to educate myself subsequently about the reality versus the hype, which is quite hard. Because there are very strong opinions on do we actually have a real quantum computer that is able to process cubits or not, is it some fake thing, right? What are the categories of problems that are solvable in quantum?

It's interesting, there was a paper a 17-year-old girl put out recently, and maybe it was Google, I can't remember the company, but her goal was to prove the applicability of quantum algorithms. What she actually proved was the reverse that for the class of problems she was working on, you could just optimize classic algorithms, which is a very fascinating space. Of course, this week, there was big news from Google that there was a big breakthrough on an algorithm that I haven't really dug into yet. The challenging thing I find is the level of mathematics that you have to have to be able to validate what's publish is quite advanced. So you read the paper, and then you go look at the math, and you try to figure out is this real thing, for those of us who don't work in this day in and day out. It is interesting, though, that one of the primary investors in D-Wave is Jeff Bezos. It goes to Gojko 's comment about will we have quantum as a service on Amazon, right. Like, that's the logical thing.

Recommended interview: Quantum Computing in Practice Murray Thom

That's one crazy idea, but I have one that goes even further. So I've been doing a lot of work with research groups that are working on cataloging genomic sequencing variants for creating personalized treatments for human diseases. 

Around that ecosystem and learning, it is interesting to look at some of the early efforts around using DNA as a storage mechanism, which can be quite frightening to think about, the matrix. Right? But, again, I think it's critically important that we follow the technological advances. One of the things that I would commend you for is that you had this woman here at GOTO speaking about China because of course, we all know China generally has a different ethical base than we work from, particularly in these newer technologies. I think it's actually extremely important to look globally at what's happening. So in terms of quantum and in terms of using  DNA as a storage mechanism, I'm looking to the world but particularly China, to see what's coming so that I can understand it and figure out what's going to be the impact on my work.

Preben Thorø: I think that's a wonderful statement on computer discussion here. Thanks a lot for your time

Gojko Adzic: Thank you very much. 

Lynn Langit: Thank you.

Preben Thorø: Enjoy the conference.

Recommended talks

GOTO 2014 • Adaptive Planning Beyond User Stories • Gojko Adzic

GOTO 2018 • Experience Report - Teaching Kids to Code for 10 Years • Lynn Langit & Jessica Ellis


Related Posts