Tech Trailblazers Showcase: Bamboo Systems flash talk

For the first year ever, the 2020 Tech Trailblazers Awards teamed up with the London Enterprise Tech Meetup to host an event where some of the top of the crop of entrants could showcase their award-winning businesses. One of those who gave a flash talk at the event was Tony Craythorne the CEO of Bamboo Systems, the firm which won the Cloud Award this year.

Tony gives an introduction to the firm and provides a quick explanation of how Bamboo Systems is aiming to revolutionise data centres by bringing Arm servers into the x86-dominated market, offering a 50% reduction in acquisition cost and a 75% energy saving whilst offering better performance than AWS or Intel.

The host is Ian Ellis for London Tech Enterprise Meetup and the session is moderated by our very own Chief Trailblazer, Rose Ross.

YouTube:

Also available as audio only on:

Interview transcript

Ian Ellis: I think what we’ll do now, in the spirit of moving forth, is let’s move on to our next demo which is going to be Tony who’s going to present from Bamboo Systems. So, Tony, are you with us?

Tony Craythorne: Good afternoon, good evening. My name’s Tony Craythorne, I’m CEO of Bamboo Systems. We’re actually headquartered in Cambridge, UK, although I’m based in the US.

Let’s just go on to exactly what it is we’re doing. We last year launched a server that is unlike any other out there right now, because it’s based-on Arm technologies, and I’ll talk about the market in a second. But why are we doing this?

The server market is dominated by Intel obviously, an x86 Architecture. I heard people talking about being in the 90s, I actually started selling PCs in 1987, so I don’t know if that makes me the oldest person on here, I hope not. But it’s a huge space, it’s an $80 billion market, as I say dominated by x86. That architecture and that technology really hasn’t changed. What has changed is the applications and the uses that people are using them for now. Machine learning, AI are obviously dominating everything. Everything used to be single threaded workloads, programmed in Java and C++, now they’re highly containerised in Go and Python, etc. Kubernetes is dominating everything.

So, what are people doing? They’re throwing more and more compute at the problem, and we were talking about Green-tech earlier, today over four percent of the worlds power is taken up by data centres, and people are just throwing more and more compute at it, it’s just becoming a bigger problem. So, we think there’s a better way of doing it.

Everybody talks about Arm and Arm in the data centre; our company was formed in 2015 and probably came out too early. The market wasn’t ready for it, the ecosystem wasn’t ready, but if you look at what has happened in the last 18 months, AWS moving to Arm, Apple have moved to Arm, Nvidia have bought Arm, Microsoft and VMWare are now developing for Arm, and then look at what Gartner believe in the hype cycle, they believe as well that Arm is coming to the data centre. It’s definitely an opportunity to really disrupt what is an $80 billion market, and that’s just the on-prem market, that doesn’t include the hyper-scalers in cloud, that’s just the on-prem market is an $80 billion market we’re addressing.

So, this is our product, it’s called the B1000, there’s a couple of things unique about it and different that I’ll talk about. We’ve gone for a pretty unique brand as you can see. Yes, that is real bamboo on the front, and no it doesn’t catch fire – we get asked that a lot, believe it or not. But what makes us completely different is there are quite a few other Arm servers out there, and all everybody else has done is literally just stuck an Arm chip into an x86 motherboard, and you get some of the benefits of Arm on that with power and cooling etc., but apart from that you’re not getting a lot else.

What we have done is completely different. We’ve re-architected the entire motherboard, the entire server, which is our IP, and where you see a 1U server in Intel or any other Arm server you get one server, that’s it. Inside that 1U there are eight fully functional servers that can run Linux, Kubernetes, etc. and we’ll come onto some of the use cases in a bit, and a little bit more extra-secret source in there, the way that we separate out IO processing from application processing. But the numbers that we’ve got on the right are very real numbers, compared to x86 and other Arm servers, because we get so much from such a small footprint. We save on average about 50 percent of the acquisition cost, compared to our competition.

We’ve only measured, and we’ve only been able to measure right now, 75 percent saving on energy consumption. What we believe is, as we scale up and we deploy racks and racks of Bamboo Servers, that we’ll be able to save even more than that, because we can spin up and spin down based on the use of the workload. And then because we get so much compute in a very small footprint, we can save a huge amount of rack space, which again is key for just about every organisation right now; be it in data centre or edge, people want to get more and more out to the edge now, we can get a lot more compute. Each of those eight servers can run independently, or we can cluster them together to really get an incredibly high throughput.

I won’t go into the speeds and feats, but you can get a sense that we’ve got a fair amount of compute, a lot of memory. Because of the way our architecture works we actually run at native NVMe speeds, so we’re running at 24 gig per second (24 GB/s), so it’s incredibly fast inside the system, and we continue to test various workloads as we run up to launch.

Here’s a couple quickly. This is just us comparing ourselves to a 3 Xeon cluster and a couple of AWS M4 large instances. This isn’t a two-blade system, we ship our system with either one blade or two blades, we just tested it with one blade, so four servers, and you can see on a fairly straightforward Elasticsearch benchmark test that we are out-performing both Intel and AWS. And then everybody’s talking about cloud, I don’t think many people are going 100 percent to cloud, especially not large organisations, but either a hybrid cloud is absolutely a thing and is going to continue to grow. So we took our friends over at AWS, their own ‘total cost of ownership’ calculator and used that. They use this to bash on prem. So we used it, did another comparison again between a typical Intel configuration, AWS, and us – and again using their metrics you can see that the total cost of ownership of running a Bamboo Server comes in considerably less than both Intel and AWS.

In terms of our use cases, we were talking about DevOps earlier; everything is moving to containers and Kubernetes etc. We’ve tested all of these applications, basically anything that will run on Open Source Linux will run on a Bamboo Server. Some will need a little bit of recompiling but we’re targeting to make it as simple as possible, but again with Nvidia buying Arm and all the other push around Arm right now, and AWS have made a huge investment on their Graviton2 instance which is based on an Arm. The applications in the eco-system are now definitely there for an Arm server to become more and more prevalent.

So that’s it. I flew through that pretty quickly. We are getting an incredible amount of attention in the press. We’ve had, I think, over 120 articles written about us, won multiple awards. For a little tiny start-up, we’re 22 people right now, I’ve got a new sales team starting literally next week as we go to market. We’ve got an incredible book of business already; it really has been quite an incredible journey that we’ve been on.

So, with that thank you for the opportunity. Any questions?

Rose Ross: Fantastic Tony, thank you. Nice to meet you, we’ve met virtually, and it’s snowing in the UK and in Nevada, it’s all over the place.

Tony Craythorne: Whoever’s in Seattle, we’ve got all the various weathers going on at the moment, because I know it’s cold over there in Seattle, it always rains anyway, and now we’ve got snow here, so it’s crazy.

Rose Ross: Jennifer’s been basking in sunshine this morning. Who knows what’s going on!

I’ve got a couple of questions here for you Tony, if you could take those. Hamed said can we get the slides, I’m sure we can get that organised, I will leave that in Ian’s capable hands.

It’s Ollie, hi Ollie, who is asking, ‘How easy is it to use this platform for my existing applications? How painful is it to be built?’

Tony Craythorne: It depends obviously on the application, but I think now there are over 200,000 software products that run on them natively now, but as long as it sits on them; so we run on CentOS, Ubuntu…I’m meeting with Red Hat this week; they’ve actually reached out to us. So we’re going to continue to test, we’re going to continue to publish on the website, but if it’s something you want to try then get in touch with us, we will help you through. We’re still at a very early stage, and we’ve got a fantastic team of engineers that will help any customer port over to it, if work needs to be done. But all of those applications that I show work natively, and there are hundreds and hundreds of them, so the ecosystem is growing every day.

Rose Ross: Fantastic, so hopefully that answers your question Ollie, I know it’s a little bit difficult when you’re typing in.

I’ll move onto Gillian Levy. Hi Gillian. ‘Have you benchmarked against Graviton2?’

Tony Craythorne: Not yet, but we’re going to. We’ve got a little re-spin we’re doing on the blade right now, and then we’re going to. Graviton2, there’s two different ways you can deploy an Arm chip. You can do one of the big, huge chips like on Ampere or something like that; we didn’t do that, we’ve clustered together smaller chips because they’re much more efficient, much less power-hungry, because even some of the large chips right now are pretty power hungry. So that’s why we’ve gone with our configuration of having still a lot of cores but in smaller chip configurations that are clustered together. So we’re going to do that and, well, if it comes out good we’ll put it on our website, but if it doesn’t, we’re going to tune ourselves up even more. But yeah, we will be doing that, we’ll be testing things.

Graviton2 looks remarkably similar to our architecture. AWS have done it right, I’ve got the patent on it, we got the patent late last year, but I don’t think we’re going to be going up against AWS on that one unfortunately!

Rose Ross: Fantastic. Well we’ve got two more questions, so we’ll try to get through those quickly, and I think Charles is going for comedian of the event here, but firstly Scott, ‘Who is your ideal customer and why?’

Tony Craythorne: We have been contacted by universities, by healthcare. We’re working right now under incredibly strict NDAs so we can’t talk about it, with one of the biggest names in IT who are launching a Kubernetes rack that they want to deploy on every single customer’s site, we’re talking tens of thousands, and we’re in prime position to be the Arm server of choice for them.

We’re working with a company who make military systems for aircraft and they’re not buying the system; they’re just buying the blade.

What else? General IT, general purpose server. Kubernetes is a huge one, we’re getting contacted a lot because if you think about it, eight servers in a 1U, you can deploy a lot of Kubernetes instances on that. So, it’s right the way across the board, large corporation, small-medium organisation, OEM, and just it’s right the way across the board. A little company called Google reached out to us last week, and we met with them last Friday as well.

Rose Ross: No name dropping here at all from Tony!

Tony Craythorne: No.

Rose Ross: No-no-no, no spoilers! No spoilers! Watch the next episode of Bamboo Systems! Thank you.

Howard, ‘do you think chips of Intel or AMD will be outdated against agile processors like Arm, NM1 chips in servers?’

Tony Craythorne: The reason why we’ve launched the 1U simple device, simple to sell, simple to install, simple to understand is nobody’s going to throw out their Intel, they’re going to hug their Intel. It’s everywhere, right? So, nobody’s going to throw it out. I think there will be a place for both. I do think – I don’t know if anybody from Intel’s listening in here – but I do think Intel have taken their eye off the ball, which is why Arm is really coming forward across multiple areas. If you think about it, Arm runs everything apart from data centre, everything you touch from your iPads etc. have got a low power chip that’s based on Arm technology. So I personally think the two will become complementary, but Arm does give a lot more benefits in terms of power cooling, etc. which is such a hot topic right now, that it’s going to take a slice, but it’s going to take time.

Our financial group projections, we think Arm will take about 10 percent of the market in the next five years. So that’s what, $8 billion, and whoever owns the space – and we are currently being seen as the leader in the space, which is dead nuts because we’re tiny, we hope that we’ll take a double-digit share and it will happen very, very quickly.

Ian Ellis: Thank you, Tony, and thanks Rose for a moderating the Q&A.

Tony Craythorne: Thank you Rose.

Ian Ellise: A very insightful presentation.