 Okay, thank you for sitting patiently. I hope you've had a good day so far. It's talking about multi-cloud. It's about having options. It's a talk for people who like to have options. And it is not about Cloud Foundry, or about Bosh really, from the technical point of view. What I'm going to focus on is what it means for you to have those things, and why it's more important than ever for you to have that capability. In your organisation. And to do this, I'm going to look back at the last ten years of Cloud Buying Decisions in public cloud, particularly. And then I'm going to have a little look at some of the current trends and some of the things that are happening around us day-to-day, which are changing the way we see things. And then I'm going to look forward to how, if you take all that into account, and you have things like Bosh and Cloud Foundry in your life, how you can make better decisions for your business and better economic choices. But before we do all that, we're going to have a little story. So it's a story about another Dan. So I didn't introduce myself. My name's Dan. I work for Engineer Better. My business associate, my partner, he's also called Dan. And Dan has a moustache. Now, he wasn't born with that moustache. He has it and it remains on his face by virtue of the decisions that he makes in his life. And every day he gets up and he looks in the mirror and he has to contemplate whether he's going to keep it or not. He's holding a razor in his hand and he decides, no, I'm going to keep the moustache. And he goes away and he comes back and then he thinks again, he looks in the mirror and he goes, no, I'm going to keep the moustache. But the longer that cycle goes on, the more everything in his life becomes dependent on it being there. So his identity, the ability of his friends and his co-workers to even recognise him, his identity in a professional sense and a business sense, the marketing of his fledgling business. And even his family are completely dependent on it being there because the first time he got rid of it, his girlfriend burst into tears. So you could say his switching costs away from having the moustache to being clean shaven are very, very high. He's locked into it. Now, we kind of have a similar situation in business and we often feel quite uneasy in the decisions that we make in business. And we have lived through, or many people in this room will have lived through a couple of decades of Microsoft and VMware and Oracle and these buying decisions that we make. And sometimes this begins to feel very familiar when we're looking at cloud as well. That's disappointing that we have to have that feeling. And it translates into a more tangible effect as well in the business. So we have a study here by Wakefield Research that spoke to 400 IT decision makers. 78% of those people believe that the fears of lock-in were preventing their business from really taking advantage of cloud. So it feels bad and it is bad for the business and we aren't really getting the most out of things. So what exactly is it? What is lock-in? So I went and had a look at some research on this and some ideas about how to conceptualise lock-in. And it turned out it's not a binary state. So you're not either locked in or not locked in. It's a cycle that you go through and this cycle progressively escalates your switching costs. So you come to have to make a decision and you're choosing a vendor and you choose that vendor and then you sample that technology and you go, this is pretty good stuff. And then you begin to use it in your production environment or over else and you begin to be locked into that. Because by the time you come around to your next technology decision the world looks different by the time you've been around this cycle once. And every time you go around the cycle choosing the same vendor again and again you start to have fewer and fewer options available to you when you get back to your next technology choice. So if you take that idea and then you map it onto the last 10 years of cloud decision making, we're going out and we're buying public cloud. We've got a situation where we've had a really sort of pretty much a one horse race, I'd say. It's fair to say, you know, had a decade of AWS. They've created a market. You could say it's almost a captive market rather than a competitive market. It's a situation where you can see that they've been very, very good at persuading you to go around that decision cycle as fast as possible without really thinking about it. And you have very compelling economics so you say, oh, I can't afford not to do this. Everyone else is doing it. I've got to get on the train. Let's go around the cycle again and you think, well, there's no one to switch to anyway so I'll just have to keep buying from them. And it's not that I'm picking on AWS but this is just what I've perceived if I look at this locking cycle and apply it to the last 10 years. And there's been no real way of switching easily. So there's been very few tools available for you to do this. So you just keep going around the cycle. So what? Who cares? AWS works, right? It's a good service. Well, it is often that you look back at infrastructural technologies throughout history and you realise that they sort of evolved to become shared. So you look at roads and railways and electricity services and utility services and it's only through sharing stuff that we get true value out of it as a society. But the more things become shared, the greater the risk when they actually start to fail on us. And there's this quote that I picked out from the CloudWatch 2 consortium and had a chat with the CEO of Strategic Blue who had written a report for the European Commission on how we're kind of getting to the stage now when we're ready to think about just what impact we see from having this global service. Because we're always unprepared. So society is notoriously bad at anticipating unforeseen events, these black swan events. And we're very bad at pricing into our decision-making. So we don't really have a way of understanding the economic cost of random extreme events that will affect us in our lives. And here's a question. So what other utility service can you think of apart from computer utilities where a single massive outage or a critical failure of that business affects the entire world? Can you think of any other utility service you have, like gas or water or electricity? All of those things are regionally bounded. GPS, but that's not... Is that a utility? Could be. That's a question, think about that one. But I would say that if you compare the commodity-like utilities that we deal with every day, you compare that with computer utility, i cloud, then we've got maybe a much bigger problem to deal with there when we think about failure. Because there's no real predictable outcome. So these systems are complex. Global systems are complex. So by definition we have no predictable link between cause and effect. We can think of things like imagine if countries in Europe started to vote to leave the EU. Imagine if that was to happen. Can we foresee that happening and can we imagine what sort of data legislation might spin out of that and how that's going to affect us and how that's going to affect the regions of our cloud providers? Stuff like that. Brexit and Trump and even the banking crisis. These things are complex events and we aren't able to really foresee the outcomes of these things. So that's something to think about. So that's at a macro level, but meanwhile, what's been going on in the last ten years and much more recently in the trends that are happening and why would we need to take this into account as well? Well, while everyone's been going round that decision cycle of the lock-in, so is everyone else. So you come back to a stage where the competitive advantage from choosing AWS, for example, has started to erode because it's become much more ubiquitous and the skills have started to diffuse a bit more throughout the industry and everyone else has got the same capabilities as you, more or less, access to the same skills and the same capabilities. So your competitive advantage that you were so keen to get in the beginning has started to erode. And we just keep on abstracting. This is another trend. Liz Goff here, who was the first American woman to get a PhD in computer science for her work in computer languages. So her work led to the development of object-oriented programming. She said this about abstraction. Modulality through abstraction is the way things get done, which is very, very true. So we just keep on abstracting things. We keep on abstracting away from the complexity behind, and we use those things as building blocks to become more productive. So why would we stop doing that just because we've got different silos of different cloud providers? If we have the means to abstract beyond that even, why would we not do it if it makes our businesses and our lives better? And of course, we keep on re-architecting. So these trends to have stateless 12-factor applications and to implement microservices so that we can scale our teams and we can scale our applications better, these are starting to help us decouple the idea of the application away from the infrastructure. So again, there's a level of abstraction there that is really helpful to us, and we see this in Cloud Foundry, the app as the currency. So all that's going on, and meanwhile we've got this really sort of pent-up unmet demand for portability. So as I said, Amazon are really, really good at playing that game of the locking loop. So everyone's going around and buying Amazon and they've even had the audacity for Andy Jassy to stand up and say, hey, everyone kind of wants choice, but then they try and they give up. So he was quoted as saying that that was the case and I thought that was quite brazen. So there's also survey data that suggests that that is the case. So people are locked in, but they really would like to have those applications available to them and to have that choice of portability. Of course, public-to-public portability is one area, but the private-to-public hybrid world is definitely a thing. So this is the step that all organisations are making from owning things to not owning things, and you could, if you turned it the right way, call that multi-cloud. So this was the study that RightScale did, 82% of that survey resulted in activity you could call multi-cloud. So it is really important to have that portability. That's what the customers think. What do the analysts think about this? I wanted to find some evidence that suggested that things were changing in the marketplace that would support the idea that multi-cloud is a really good idea and it's becoming more important. Well, Gartner, of course, had a magic quadrant. So what I did was I took three consecutive years of the magic quadrant, put them all in one slide so you could see the changes in the marketplace from their perspective. And the bottom left square, as you can see here, is what I would describe as a shrinking island. So the market's becoming completely polarised. The only three players in the public-cloud market that aren't going backwards are Google, Amazon and Microsoft. Microsoft have even closed the gap on Amazon a little bit. So that's quite interesting. You can look at the visualisation there of the data and say, OK, well, things are changing. There's fewer competitors, but there's only three horses really in this race now. But that wasn't the most remarkable thing about the report because if you dig inside it and you find this sentence about Azure, you find that they're describing Azure as good enough so you should consider it good enough for your organisation and base your decisions on things other than the technical characteristics. So that's good enough is something you talk about when you're talking about commodity-like markets. So that's actually quite interesting, the fact that they start to use that sort of language. And then Ripmunk went away and they said, OK, well, let's look at how fast the pricing is changing in the cloud market. And they said that it was slowing down. So the race to the bottom is starting to level out now. The rate at which pricing war is taking place is slowing down and starting to level out a bit, which is sort of consistent with it becoming a commodity market or a commodity-like market. OK, so where are we so far? Perasif, single supplier. So we've got this situation with 10 years of us buying services from one provider. That's a big risk to us. We're very bad at managing our switching costs in that world. Unexpected things do happen and we're awful at predicting those events. We keep on abstracting. We've also got ubiquity in AWS. So we've kind of lost our competitive advantage by using that as a single supplier. We keep on abstracting away. So there's a trend for us as human beings using computers to keep wanting to abstract. And we wish we had options. So there's evidence to suggest that we know it feels wrong to just keep doing the same thing over and over again. We want to have more options in the future. And the market's becoming more like a commodity market. So that's the trend. And this is where open source steps into the picture. So there's an amazing structure in my mind around the way that Cloud Foundry and all of the assets of the Cloud Foundry source code and the governance structure around the Cloud Foundry foundation really helps us. And what it's effectively doing is allowing us to create a competitive market space in cloud. So where we've had a very captive market space in the early days of AWS, it's much more in keeping with Sam Ramsey's idea of this positive sum game to have a competitive marketplace by using things like Cloud Foundry Certified, where everyone's put on a more level playing field, and having things like Bosch and Cloud Foundry together, wrapped in this idea that you have not just one company in control of things, but over 60 different companies. And the governance model around that is that because the Cloud Foundry foundation is a trade organisation, it's a 501C6, I think, it means that it is illegally impossible for the source code assets of Cloud Foundry to be transferred to any single commercial entity. So you are very well protected when you choose this technology. You have a lot of the risks of locking, mitigated for you by choosing this technology. Of course, Dave's can go anywhere when they use it, so it applies not just to operators, but also to Dave's, of course, because that's where it all started. So you can see a push to anywhere, and you've got a certified provider from anywhere, and you know that your app is just going to work in exactly the same way on all these different providers. And the operators have Bosch, and Bosch I describe as this travel adapter for cloud. So it's basically the, it allows, just in the same way that I can take my laptop from here to any other country in the world, and as long as I've got the right travel adapter, the right CPI in Bosch terms, I can take that anywhere, and I just know that I will be able to run exactly the same deployment in the same releases on a different cloud. And I just put some of these CPIs up here that I know that are in common use. There are obviously others. I would assume that most people in this room have probably come into contact with Bosch by now. If you haven't, please go away and read a lot about it. The beauty of it, as I just said, is that modularity. So you can swap out a CPI completely and keep everything else in your deployment in exactly the same, and you have that nice separation, and that separation of concerns between the IS and what comes above it. So who's paying attention to this? Well, we've got governments. We've got the British government, we've got the Australian government, and banks. So these are organisations that have been early adopters of Cloud Foundry, and it's really because the stakes are so high in these organisations. So they do care about switching costs. Governments are not going to want to take on the risks that I've been talking about a few minutes ago. And also working with a very big bank at the moment for whom that ability to deploy into public cloud but then repatriate that into their data centres is a huge issue for them. So Cloud Foundry was the natural choice for that to meet those requirements. The best example I've heard from the community around multi-cloud is NE9s. So they've actually done this twice. They've already moved from VMware to OpenStack, and they did that mainly as a cost-saving exercise. I think they cut their licensing costs or their operating costs in half by doing that. And they did that on a production platform that had 120 apps. They moved between these two CPIs. They did it with just two to three people in a couple of weeks, and it only had a 30-minute downtime for their production-hosted CF platform. And then they did it all again from OpenStack to AWS because they were having stability issues on OpenStack and hardware problems. So this is a real-life example of somebody taking a service which from the customer's point of view remains exactly the same, between the three different clouds, which is remarkable, I think. And the good news is that Bosch just keeps getting better, and it's, you know, with the good work of people at Google and Microsoft here. I'll come back to that in a minute. So we've had the CPIs getting better and better, and Bosch, as a technology, is maturing. So it's much easier to use. All of the complexity and the nastiness of all the yamol manifest is beginning to be sort of moved out into different areas, and there's going to be much better tool sets available. So I think it's a great time to start investing in that, and it's bringing down the marginal cost of us using Bosch in our production world. And I saw a great talk earlier from Colleen and your colleague about how Google has been working on service brokers and the CPI as well, which really helps us when you combine that with the fact that the commodity market is evolving in cloud. It means that those two things coming together make it a great time to start implementing us in multi-cloud. So at the end of the day, I think this is what you begin to see when you look forward. So we've had the last 10 years. This could be the next five to 10 years of us making much better choices in our businesses. And by doing this, what we're doing is we're managing our switching costs. So because we know it's a cycle, it's not a binary state, so it means that we can control things, whereas we haven't had that control before. So I'm working in a team at the moment, and when we have our IPM or when we are evaluating stories throughout the week, every now and again, someone will say, is there a Bosch release for that? And when they say that, I know that we are unconsciously managing our switching costs. And that is the side effect of having that capability and we're equipped with that tool. We have a choice to either use an inherent service or, sorry, a native service in AWS. If that's going to accelerate us, then we have to weigh up that choice against actually having a portable service. And as I said, for one of our customers, that's really important. So the reminder here is that you can manage your switching costs in exactly the same way that you manage your technical debt. I think that's the key issue here. It's not a binary state. You're not born locked in. You're not born locked in by the decisions that you make. So to wrap up, open source is the enable for all this. That's what enables us to have those competitive markets, which is really important. We get portability for devs with CF and we get portability for ops with Bosch. And we've got that lower marginal cost. So it's getting easier and easier to use Bosch. So you should definitely start using it. And the benefit of switching is that you've got viable alternatives to AWS now if you're in public cloud. And of course, you also have the benefit of more easily being able to move between private and public clouds. So that's a wrap. If you've got any questions, now's the time. Okay. Oh, there's one here. So, whenever you're migrating, you're moving things over and it says something like a green blue deployment that you envision for moving over to a different cloud or something like that. Just using Bosch releases, I've just deployed the same setup and then I have to do the switching manually and stuff like that. So there should be something above that layer. I guess in broad sense, DNS is probably something you need to manipulate quite cleverly there. If you're moving between clouds and you would need to factor that in. DNS is something that you should probably have outside of the cloud provider that you're investing in heavily. At some point in the bar in the last few days, I had a conversation with somebody who said okay, on something to deal with going Bosch, two different CPIs and moving volumes across as well. Maybe I'll make enough. I'm sure something. We'll think about it and get back to you. We'll take your details. Anyone else? Any other questions? Oh yes? Do you want to take the mic? How do you see the predominant role of single companies in an environment like Cloud Foundry? So we have this situation. For example, yesterday I counted the number of talks. I think there have been 32 talks and 16 of them, Pivotal was involved in some way or the other. He told us about not a single vendor dominating a project. How does this play together? There's no denying that the contributions to Cloud Foundry are heavily weighted towards Pivotal and they buy more influence in the direction of Cloud Foundry because of that. By the virtue of the fact that the foundation has and they're still working on this openly admits that it's not perfect. The structure's in place to make sure that everyone has a voice in that group of member companies is really important. Also Pivotal themselves don't want that because they know that it doesn't look massively healthy to have an 85% ratio towards them committing. But I think it's going to get better and I think that you can see the rate of growth in that foundation membership. All we can do is open more dojos and encourage more people to become committers and try and change it. That shrinking island of a second tier IaaS providers that you mentioned in the Gartner diagrams. Do you find that any people are starting to look more closely at those via the mechanism of the Bosch abstraction so that you can put them more easily into your sampling part of the cycle? If they do offer the services that you want. Does the Gartner diagram become less important now that you've got that Bosch abstraction? A good question. I think you could do that. It would be a good way of making niche providers much more relevant and giving them a chance to offer things which are consistent with public clouds. Any other questions? Good. Thanks for coming. Enjoy the rest of the day and have a safe journey home.