 Well, thank y'all for coming down to the basement to watch our panel This is a good hall good size all If anyone can't hear in the back give us a thumbs up if we need to talk louder mostly for the panelists My name is Justin Shepard. I'm the CTO at the for the private cloud business at Rackspace Hopefully that's last time you know here Rackspace. I I like kind of doing these type of panels because I think it's really interesting to hear what Private cloud users and cloud users are actually doing with their tech And the challenges that they're kind of facing as they go through migrating and or operating on top of clouds And so today we've got three guests they're gonna join us and talk through kind of what they're seeing actually running on top of clouds and Hopefully share some interesting stories on the challenges that they faced how they got around them and give some good advice if anyone is starting to kind of go through and Do digital transformation with their companies or they're migrating applications to the cloud or building out net new applications on the cloud So I'll first hand off to Simone and let him kind of give an introduction of himself and what his company does And then we'll go down the road and I'll call out past that so Simone. Can you tell Spet yourself? Good evening Good evening So my name is Suman Mukhopad. I Spent over 26 years in the IT industry building production infrastructure for companies like AT&T, AOL, Yahoo, Walmart and HVOP Had the privilege of Witnessing the evolving infrastructure landscape from Dedicated server to virtualization to cloud To infrastructure as service to platform as service to software as service for anything Today working on building a cloud platform for Container orchestration and release automation Primarily Built a open-stack private cloud Which is using a commodity hardware? We didn't take the traditional approach Built 10,000 plus VCPU about 300 plus terabytes of storage Which gives 55,000 plus IOPS on a commodity platform commodity hardware platform This takes the work this workload takes the financial applications like banking Mobile banking e-banking applications and that Gives a enough resiliency and Address the security postures that is mandated by the fintech industry we partnered with RAC space for to help us obviously to build them infrastructure and The journey had been very successful so far Sergey you want to introduce yourself? sure Hey guys very hard to talk to people who are like 30 meters away, but hopefully I will manage So my name is Sergey and he can work right now CTO at Pied Drive who knows what Pied Drive does come on guys Okay, we have a supporting group here so Pied Drive is Sales to like software as a service for small medium businesses mostly focused on sales people so really know what Individual sales people need and kind of cutter for them. So my previous history. I started in banking sector worked for eight hour for eight years Developing business like internet business banking applications and then moved to Skype and I don't know how many of you know but Skype haven't been using actually any of the servers for quite some time It was running purely on peer-to-peer and After that moved To pipe drive five years ago Yeah, and I know Justin has some Questions in his pocket so pocket so I can later speak about our experiences with clouds Learn all right My name is learn tangent. I'm with G digital. I've been there for 18 months almost two years. It's been pretty interesting We're in our little segment. We do most of the operations and engineering for GE's OpenStack or private clouds Today we support a number of the GE business units and I think everybody's heard of general electric I don't know if you haven't Maybe you don't run light bulbs things like that But we're it's a pretty interesting challenge and that a lot of the business units are very large companies in themselves billion-dollar, you know, sometimes tens of billions of dollars in size and so unique requirements coming from healthcare from lighting from building locomotives to you know Making sure jet engines are running appropriately So it's been it's been an interesting ride over the last 18 months Prior to that Actually spent some time doing architecture at a company called PeopleSoft for those that have been around for a while When worked at Red Hat for a while and then spent some time with with Justin at Rackspace doing OpenStack builds and then moved to GE Thank y'all. Well, I really do appreciate y'all coming out being able to talk So we've got a nice distribution. We got United States. We've got Estonia and we've got Bangalore So we have a little bit of a mix all over Just real fast to cover all three of you have got OpenStack based clouds Perfect. Are we also adopting other technologies? So do we have other private cloud technology stacks? Yes No, and a yes public cloud technology stacks. Yes. Yes, and a yes perfect So you have the gamut of technology And so hopefully you've got some good stories on some of the challenges you face I guess the first question I would start off with for the three of y'all would be What type of challenges have you faced as you've either built applications on top of cloud infrastructure? Or you've migrated legacy applications onto a cloud platform and kind of How have you solved them? And if you're in the middle of solving them, that's okay You're in the middle of solving it. How are you kind of going about dealing with those challenges? As you're going through migrating and or building we'll start down on the right-hand side with laren and then work back this way all right It's so one of the surprises g has actually had You know been around for a while. So there's a lot of Legacy I guess is what I would call it type of infrastructure and applications that You know they've built over many many decades and as we start doing You know when I joined GE was very much down the road of digital transformation and it's They're pretty aggressive in how they're handling it and what I guess my biggest surprise was was even how aggressive they are at Moving this it's still a very big challenge. There's still a lot of applications that And you know thousands of applications that were hand-built custom-built Trying to make those cloud ready Whether it's you know an open stack based private cloud or an Amazon based public cloud You know just the concepts and misconceptions around what cloud is and how they operate and what the functionality is, you know, they most of them have experience with Other like a VMware type of cloud where they expect. Oh, you know live migrations and some of these other things and when you start talking Automation and application resiliency You know you might as well speak a foreign language to some of these folks So it's one of our biggest challenges has been just trying to get the or surprises as the Just trying to get people to understand what cloud is and how they can leverage it as they start doing Applications start to migrate applications from a traditional to a more modern platform Sergey Maybe I before to be able to relate with you who is actually working in infrastructure or on infrastructure Everyone everyone otherwise I would be surprised why you are here But yeah, so Like when I joined by drive by drive was already using private cloud rack space open stack And before that I wasn't actually I was in software development never actually run an infrastructure team and and for me it was a surprise that There is always a migration going on So maybe again a short quiz who is currently in the process of migrating from something to somewhere You can see everyone is migrating So the biggest challenge of the cloud is you have to migrate all the time To migrate from failed. I don't know hardware to a new one or maybe you have fail over there to migrate from one version of Open stack to another migrate from I don't know one public cloud to another So I guess that's the challenge of industry and How we we approach this challenge? I think it was part of our bigger challenge of a Fast-growing business like when I joined we had 5,000 customers now. We have over 80,000 customers So like just in a matter of five years We had to scale the system and You always run into certain bottlenecks, right? It can be your software Created bottleneck it can be bottleneck in open stack it can be a bottleneck in actual hardware like networking layer So you always run into some bottlenecks with your current architecture and our approach was let's not wait for the next bottleneck Let's create an architecture where we can just a copy a current setup which kind of runs well and Deploy another version So that we can run two three or five we call them data centers, but some some call them pods so we can run as many pods as we want and the port size is The size which can serve certain amount of customers without running into bottlenecks So we worked on this architecture for Almost three years now. We have two data centers one in US and over in Europe We solved some other issues like GDP a deep GDP are Coming and hitting us and then like by the time GDP are approached. We already had our Germany data center with the help of rack space What's interesting about this pods or running this multiple copies of your infrastructure is that It's not only about the architecture of how you actually Serve the customers from different boards, but it's also about how you migrate customers from one port to another and By having the solution of being able to run from multiple pods being able to migrate customers from one to another We are also simultaneously solving the problem for migration needs Okay, we need to upgrade open stack. We deploy a new pod with new version of open stack And we migrate customers to this new pod and we shut down the old one Yeah so As part of the cloud journey I started investing heavily building open stack private cloud Pretty excited that we want to go in that way and that's the way but when I looked at the drawing board realized most of my applications are Stateful none of the applications are stateless They have session affinity They have colo affinity The middle-ear application can't even handle across colo call The application is not even built Today, I'm just it's a discovery that Lot of middle-ear applications. I found the way they make the Query into the database had a very stringent Session affinity and they were not able to they're not truly scalable for the cloud environment Moreover, it's my individual opinion name one database which was built for the cloud so It was it was really Something like you are your house is ready, but people are not ready to live in we had to build a data access layer which Kind of gives a resiliency in the system that the database becomes cloud Agnostic it give it we have built multiple availability zones. We have more build multiple Redundances in the system, but the database is still sitting in one corner And it doesn't even care to what you have built in your infrastructure. So it was a great Learning the other thing is I had a Traditional approach so when I I wanted to build a open-stack cloud. I said, oh, we need ACI network and Then at the time of implementation We realized today everything is software defined you really don't need to invest heavily on the traditional technology which Nails you down to a specific set of objects And you don't have any escape route from there. So there was a very interesting Challenges that we faced and good learning Thanks, I apologize. I'm very dry Okay, so that was a good one second question And that these two kind of go hand-in-hand so as you're kind of dealing with the challenges of migrating in their building applications You've got the challenges that you're in your teams are facing, but you also have Largest surprises and I think Sergey actually may have already answered the large surprise that he found when he was doing it It presented as a challenge But Larry and I'd ask What was the biggest kind of shocking surprise? So not something that you expected it wasn't something you knew going in but As you know your teams are building applications in migrating. What was the biggest like aha? Oh my goodness I'm surprised that has caught your teams that you've had to kind of solve for I Think there's many right and I'm just I'm trying to prioritize the list of of what they are and and You know just with the the scale of it at least a GE we have a Bunch of technical Leaders right we have CTOs in each of the business units who are responsible for Applications and Architecture and those pieces and you have other organizations like the one that I work in there are responsible for hosting those And you have the same type of structure. So you have two CTOs Who are peers who may not have? The same priority list or how to how to adopt Or how to get to cloud everybody wants to cloud not all of them understand what that means not all of them understand how to get there and You know as I was talking about one of the biggest challenges is just trying to get people To understand what that is how to get there What's been nice is that? There is this mandate to get to cloud and so it's it's very much You know you were you're talking about you build it and you hope they come It's kind of that American movie right field of dreams. We we built this thing and we hope they come and what we found is that We've had to do a lot of of hand-holding and educating people about what it means to come to cloud and that it's you know just the benefits and the scale and What they can do with those those technologies and it's it's slow Or it feels slow, but we're actually making really good progress. It's just You know one of the biggest struggles the scale we still have Mainframe apps that they're like we need to go to cloud and it's like how do you go from mainframe to cloud? That's just different decades worth of technology that you have to figure out so It's been one of our our biggest hurdles and then You know the other one we're talking about migrations right getting getting those technologies to come across some of these application owners actually went out and You know the term is the lift and shift where they literally took their their old platform They virtualized it and put it on a cloud technology thinking that there's failover and ha and backups and all these different pieces and and They didn't understand that we don't do that. You know the infrastructure doesn't provide that Availability for them or the backups that it's something that they need to design in their application and that's been One of the hardest challenges for us is just setting those expectations of When you come over, you know, we don't get engaged. There's just so many applications that come to the platform They come across them. They don't engage us until they have an issue and usually it's something Somewhat catastrophic right we lost a Virtualization or a hypervisor or something like that in there. I'm down. You need to fix it And it's like well, where's your automation to go spin this up somewhere else and like what are you talking about? I don't know how to do that. So So just the I think the the different expectations that people have the Kind of the technology gap the technology debt that's there trying to get some of these legacy to the cloud has been a real challenge and then you know the Technology stack and getting customers what they need has been also difficult Just want to ask you the same question I had a Different surprise as part of the While making a choice for cloud We were looking for two things one is cost to performance and second thing is the Reliability factor it was a conscious decision to go with the open-stack cloud using commodity hardware and I was apprehensive About the performance so I was really not sure is this going to give a better or At least will it meet? match the current performance so We did a benchmarking of the performance of the CPU and memory We have AWS we have VMR and we have the open-stack private cloud using commodity hardware We took the similar instances from every cloud and What I did is I ran a NQN Program that basically where any is going to take on a fourth rates. I took the similar Instances between AWS and the VMR and the open-stack cloud on commodity hardware To my surprise I found that 260% higher performance on open-stack private cloud We got Which we never anticipated and then We did another exercise of allocating a block of 50 mb memory Running a cycle on a between AWS VMware and Open-stack again to my surprise I found The open-stack private cloud was giving me 60% higher performance compared to any one of them so It was a pleasant surprise of course but really The kind of performances that we were expecting It has it exceeded our expectation. I when I built it my apprehension was I will get probably maximum 30,000 IOPS probably I'll hit It almost gave me double almost 55,000 IOPS it started giving in terms of Storage performance. So it's a pleasant That's a good surprise most of time they're bad how many times did you end up having to recheck your benchmarks because you didn't trust the data You're getting back It's always fun Sergey I give you a chance. I know you kind of answered both of the questions already, but I would let you know I have another surprise perfect a new surprise I'm Again, it's it's I'm always amused like our Our company is creating a business software fairly simple user features and We kind of we don't want to invest in having Like invest into development of infrastructure. We don't want to invest into like creating Architecture like there are so many smart people around we just want to use already ready-made solutions and I'm continuously amazed how Like our industry is immature like you're looking for a solution and there is no solution on the market You have to go and build it yourself It's like we spend 50% of our time just building tools and building like fairly simple Usable solutions just to be able to build the business software Like continuously amused by the state of our industry It's never fun when you can't actually make a bill versus by decision. It has to be a build only All right got about 12 minutes left. So we got two more questions sketched out As you and your teams have worked through and done these migrations I want to give you an opportunity to beat your chest a little bit Tell me something that you've been very proud. So what's your biggest accomplishment that either the team has done or you've done? Something that you kind of have been really proud that you're able to kind of get done As you're either building the applications or you're migrating to cloud I Believe I will start with someone again So of course the the agility that The team built up in the cloud environment Was something which is At a cost of the if we when I compare the total cost of ownership and The agility that We have built in my team has created the environment When I compare it with other Choices it gives a inner pride just to give a sense the we did a rough mathematics We saw the TCO is 50% off When comes to Our private cloud at the same time What happens is we are in a segment where the go-to-market strategy Counts every minutes Because if you consider the fintech industry the the releases happen probably every half an hour and If you are not there It's not available. I mean you have lost the opportunity Same thing The rapid development that's happening on the AI ML market also the features gets released very quickly The responses counts every moment. So that's where the agility matters a lot so This particular cloud environment it created it matched the rhythm of the Agility expectation in terms of the go-to-market strategy at the same time the The bottom line is that the TT TCO has significantly Contained it's it's not ballooning which all the CIO's dreams is that they keep the bottom line Domination so I think those two are very proud moments for for the team and myself as well okay, as I mentioned one one of the Moments of pride is definitely being able to deploy our system into multiple locations multiple clouds We have here in in the room people who are actually Responsible for that. I mean I'm very happy to have them But it's it's To understand the complexity as you also mentioned the the like we have Seven or eight infrastructure engineers and we have hundred fifty developers this hundred fifty developers Deploying their changes all the time. We have like fifty deploys a day And we have eighty thousand customers who are using our application twenty four seven and Under this cost constant change and constant Load to be able to migrate Customers from one location to another Christian migrated over two last weeks How many thirty thirty thousand customers? From from one location to another like just be able to live in this ever-changing world is It's complicated, but it's possible. That's awesome awesome awesome accomplishment learned yeah, um You know surprisingly one of the the biggest benefits that we've gotten back from some of the other business units Cios CTOs is that When they come to the platform, it's actually the first time that they get to see what their true spend is in a single single place like they understand What their infrastructure cost is what all those pieces that go into running their application for the first time they actually get a lot of visibility into that and what their consumption is and it's allowed them to create a lot of You know because of the platform a lot of automation to help them control costs That's not something that they've been able to do before it's not you know easily done when you have a bunch of Mainframe or physical servers be like oh, I need to scale down or I need to write size In this environment, they've actually been able to do it and That's been a very positive piece of feedback that we've gotten from the Business unit leaders that they've been able to control these costs The other piece is it's actually been very stable. You know internally this is You know one of our biggest KPIs that we measure against is is API availability, right? So we may have hardware failures or maybe individual failures and some of the other teams within GE digital measure against those failures Whereas you know when you're looking at cloud adoption or cloud technologies having that API availability And if you do lose a piece of your workload, you're able to go spin that up and have that availability there you know we do use we do partner with Rackspace and You know, I think for the last probably two or three quarters. We've Very nearly hit a hundred percent of API availability Availability per quarter and so one of the surprises when we go and talk to These these business unit leaders and about the availability of cloud and what cloud means you know a surprising piece to them is We label it as stability right the ability to Put your workloads out on on these platforms and have it be resilient if it's designed correctly and then when you couple that with the Transparency of what they're using and being able to control those costs We've actually it's been a really surprising conversation surprising the positive conversation to have with those leaders about the platform So we've got about four minutes left I want to give you all one minute opportunity this kind of any advice that you'd give for any companies that are just Beginning this or any application teams that are just starting to do a migration or a brand-new build out of an application On to a cloud platform Sergey, so just be prepared to migrate I would say think crazy Don't be traditional This is the biggest learning that I had is I went with a mindset of the traditional infrastructure Trying to build a modern cloud and define that everything is softer defined I don't have to really carry on lots of baggy baggages to build a private cloud. So be disruptive think crazy learn I Think the the biggest piece of advice that I would have is if you're starting your journey or if you're even just a little bit down the journey is potentially Go in and seek help, right? And I mean that by get advice from others who have done it talk about some of like this panel the struggles that they've had and help try to find help in Figuring out your journey to the cloud. There's a lot more Or the digital transformation journey. There's a lot more to it than just being able to modernize an application and put it Out on the cloud. There's You know, we've talked about some of the the resiliency and you know, you start talking about pipelines and some of those other pieces You're usually most people are very tight closely to the application. It's hard to To think that bigger picture and get the broader view one of the successes that we've had within GE is Being able to consult with those teams and help them with that bigger concept of Not only how do they modernize their application, but also how do they create that resiliency? How do they create the automation? Potentially going to like a CICD pipeline, you know, these are all mind-blowing concepts for them and it's What I would say is if you're if you're leading that Try to get some outside help Bring in some new ideas Make sure that you're getting the full picture before you you get too far down and you're really kind of Locked into where you're at with your your transformation Well guys, I'd like to really tell you I appreciate you getting up here It takes a lot to kind of get up and talk about some of the stories You never know if anything's interesting So it's always a little bit nerve-wracking to you know Talk about some of the things that you're experiencing and no one wants to kind of admit problems So it does show quite a bit of vulnerability to get up and share those Trials and tribulations you go through so I'd really like to thank y'all and I'd ask do we get a nice round of applause for all of our panelists Thank you very much two minutes Now we can go ahead beers. Oh wait our beers at five or six Dang it. Sorry you got one more hour to get beers Thank y'all thanks guys. Thanks