Jonathan Bryce (OpenStack Foundation) moderates Trove Day 2015’s final panel discussion featuring Anni Lai (Huawei), Andre Bearfield (IBM- Blue Box), Vino Alex (Wipro Technologies), and Frank Rego (SUSE Linux).
—
—
Transcript of Session:
Jonathan: We have a number of panelists here form several different companies who all have different takes and different positions in the OpenStack ecosystem .I think it’s going to be interesting to hear from the different viewpoints as we talk about what really is I think the question that I hear from organizations all the time. Do I need a private cloud? How much can I move to the public cloud? Is hybrid cloud real? Can I actually do that? Hopefully these fine panelists today are going to help us get all of the answers for that. To start with I would like to just go down the line and we can do some quick introductions and then jump into some questions. Along the way, happy to take audience questions as we go along I think we have a mike for the audience questions as well. To start with Frank, why don’t you introduce yourself and just let everybody know kind of where you fit in the ecosystem.
Frank: My name is Frank Rego. I work for SUSE as a business development manager. SUSE has been producing Linux for 22-23 years, so not new to Linux. We have an OpenStack distribution, which I think our first release was Essex, so we’ve been doing OpenStack for a number of years.
Jonathan: That’s a long time. All right. Anni.
Anni: Hi, my name’s Anni Lai. I’m with Huawei. As many of you know Huawei has been a CT company for last 20 plus years. Actually for the last 30 years we’ve been investing very heavily in the IT space. We do have our own service, storage, network and cloud platform, which is OpenStack based. We have been involved in OpenStack since Grizzly days. In OpenStack world we name the era by the release. Our mission is to enable our telecommunication customers as well as enterprise and service providers to stand up their private cloud/public cloud as well as hybrid cloud.
Jonathan: Thank you. And Andre?
Andre: Hello everyone, my name is Andre Bearfield. I work with Bluebox. I work in the product organization. I’ve worked specifically on OpenStack. The Bluebox cloud product, which is a private cloud, is a product from its instantiation, until ultimately the organization’s acquirement by IBM a couple of months ago.
Jonathan: Congratulations.
Andre: Thank you so very much. We’re primarily active in the private cloud arena.
Jonathan: Okay. Vino.
Vino: Hi. My name is Vino. I’m from the organization Wipro. We are primarily doing consultancy and system integration of all the open cloud technologies. We start with CloudStack then go to OpenStack. We have a couple of products in also.
Jonathan: Okay. Great. Thank you guys for those introductions. The first thing I want to start off with is, this panel is really, the way it was phrased was public versus private, versus hybrid, versus AWS. When I talk to companies, I’m fortunate enough to get to go talk to lots of users, lots of organizations, what I hear is that they’re often doing a mixture of those and trying to figure out how they navigate over the long term. My question for you is, which models do you support and do you think that private clouds, this is something we hear sometimes; private clouds are really just kind of a temporary step on the way to everyone putting everything in a public cloud. Anyone wants to take that?
Anni: I can. At Huawei we are the enabler. We enable our customers to implement their private cloud or to operate their public cloud and then sometimes they need hybrid cloud. I also talk to a lot of customers. I’ve travelled to like 30 plus countries working for Huawei. The majority of the customers I talk to are Telco end service providers. Sometimes they have public louds and they sell to their local markets. I think really it depends on the customer.
SNBs obviously they don’t want to own IT anymore, so they will resort to public cloud, but for enterprises they also realize that there’s agility and scalability benefit the public cloud and also there’s this shadow IT going on .They realize that they really have to incorporate public cloud into their IT environment. They’re moving towards more of a hybrid cloud environment. They are not going to give up their private cloud for sure because they have a lot of data centers; they have a lot of investments. Giving up their own private cloud or their own IT is I think, for the next ten years probably not going to happen, but they all are moving towards a public cloud environment, I mean hybrid cloud environment.
Jonathan: You’ve said that you’ve been to 30 plus countries and working with Telco’s and a lot of them are implementing public clouds. Are you seeing any trends by region in terms of public cloud adoption?
Anni: Yeah. Some regions that Amazon and Google and Microsoft are there, so they obviously compete against those, even they are there because of the data sovereignty issues and stuff. Telco becomes the de facto public cloud service provider because they own the customer, they have the government support, they have the network, so they become the public cloud service providers. They’re still, when it comes to cloud adoption, they’re still compared to U.S. market and Western European market; they are still a little bit behind. They definitely see the benefits. Also the benefit of those regions, especially in Africa or Latin America, a lot of them they didn’t have IT to begin with, so they didn’t have to go through this transitioning or migration process. They can just start from scratch.
Jonathan: Frank. SUSE has I know private cloud customers, some of them have come and spoken at our summits before. What do you think you see in terms of the trends around deployment models?
Frank: As an operating system vendor, I think we’re especially vulnerable to the new cloud ecosystem. A lot of our customers, in fact probably most of them, 95 % are deploying to the public cloud. We have a really good presence now in the public cloud on the big cloud providers, both the global providers and regional providers. At the same time a lot of our customers are now looking at supplementing their private cloud. When I say supplementing I mean most of our customers are using VMware already. We rarely go into a customer who hasn’t already virtualized 90% of what they’re doing. They’re looking to not necessarily move off of VMware, but maybe supplement what they’ve done in a proprietary way with a more open standard around cloud. I think we’re seeing both ends of the spectrum. I think we’re seeing customers move much quicker into public clouds. We’re also seeing customers who have deployed in the public cloud but are now moving some of those workloads back into the private cloud. For us, the OpenStack business is growing It’s a lot of dev and test now. A lot of kicking the tires. With every new release that we do we get more customers that are moving into production. But for sure there is a mass migration to public cloud and now private cloud just starting to take off.
Jonathan: Okay. Andre, at Bluebox you have kind of a blended model with hosted private cloud where you allow people to have dedicated resources but in a fully operationalized model. One of the biggest hurdles that we’ve seen people run into when adopting OpenStack is finding the expertise to operate those environments. Bluebox tried to productize and answer to that. Can you tell us what you’re seeing and especially are there any particular workloads that you see. It’s kind of a service provider model still. Why would they go to hosted private instead of just going straight to public cloud?
Andre: To public cloud?
Jonathan: Yes.
Andre: Okay. I think that to back all the way to the beginning, the initial question, Anni is probably the most right here right. The thing is, every customer is going to have a different need. Every customer is going to have a different use case. That detail is probably not going to change. The reason why we determine the focus on private cloud as a service, this hosted model is because cloud, especially OpenStack, especially in Havana and previous to Havana, was very difficult to deploy, very difficult to manage. Customers not only would have to build a data center, buy a bunch of hardware figure out what software to use, but also figure out how to operate the hardware and the software that lives on top.
I guess your question is why would a customer do any of that and why not just use public cloud. I think the reason is a bunch of workloads have security and privacy requirements. Some customers require that their data is on a machine that they can point to or in a data center that they could feasibly go into visit. I think that these workloads kind of necessitate kind of a variety in this market at the very beginning here. In addition to that I think, Bluebox built private cloud as a service. Obviously public cloud, the question of AWS, the elephant in the room is huge and amazing to think about what they’ve done and it really drives a lot of what’s happening in the private cloud space and in the hosted private cloud space as these entities start to ramp up.
I think that some of the question around whether public cloud will take over, there are a couple reasons. I think the first reason is because in the public cloud the developer can drive. It’s not an IT organization necessarily. It’s not a CIO question. It is a developer who needs to do something right now. They will drive adoption. I think that’s what has led us in this direction where public cloud is obvious. It’s the first place to go. It’s the easiest place to get because the developer has a need, but we still have these organizations that have these huge IT organizations and they have 12 or 15 or 20 years of background doing this thing and they’re trying to figure out how to the next thing. I think public cloud is a difficult transition for them. I think that’s the first place that private cloud starts.
Then the easiest way to step your way into cloud is probably avoiding purchasing the hardware, but building a data center and then figuring out what software to buy and then operating that software. I think private cloud has a place there. I think private hosted cloud has a place there. I think public cloud is the natural leader and will help us understand what the developer needs. At the end of the day infrastructure as a service is just a foundation on which a bunch of other things needs to happen. At the end of the day, we’re empowering developers to do what they need as quickly as possible. I think that’s what we want to do.
I think public cloud will continue to lead, but there are workloads that probably will never need to go there. I think what we should be targeting as organizations is how can we provide a single simple interface that allows customers to choose, based on workload and data, whether or not their services and their workloads are on their premises, whether they could be on private cloud as a service, on someone else’s premise, but on a dedicated machinery. Then finally what’s the easy, low risk work load that I need to get up as soon as possible in public. That’s my thought.
Jonathan: Okay. Vino, you provide services across a variety of these different deployment models. Have you seen, Andre was talking about picking the right model for the right workload, have you seen any patterns in terms of which workloads people are moving to public or private or any of these different models across the customers that you’ve worked with.
Vino: What the model we are normally finding is the traditional workloads our customer more prefer to move to a public cloud, it’s more of a random. It’s nothing to do with the cloud like workloads; it’s more of a move couple of workloads in a very traditional fashion move towards the public cloud. Many of our customers are also having the real cloud-like workloads. They are setting up private clouds.
This trend, our system integration partner for many of our customers. What we are doing in this place is something like, we are bringing in almost 50 plus cloud technology partnerships under one umbrella to test benchmarking, many of the cloud use cases and bringing that value to the customer. We are trying to build up a public like private environment. The trend is more of a, I can specifically tell that I know many of the customers; they’ve already started using public cloud. They are asking us to set up a proof of concept, well within their environment on how seamlessly the workload can be moved from a public to private and how it bring value to them. We are seeing a lot of movement from public towards private.
Jonathan: Actually, one of the things that I found to be interesting is just in the last year we have seen several OpenStack users, not just big users like PayPal and Wal-Mart, who have thousands of machines, but several OpenStack users who have set up private clouds for cost cutting reasons because, Tapjoy is one that comes to mind, which is a mobile analytics platform, they were a huge public cloud user and as they went through and looked at their business model they realized that they were pouring tons of money into the infrastructure and they worked with a partner to kind of design and architect a private cloud environment. They were able to get 10x the resources for the same price in a private cloud by making sure that they built something that was specific for their need, which was data science and analytics. André you were taking about public cloud just kind of blows all those barriers away for the developer.
Andre: Yup.
Jonathan: They have no barriers any more between idea and on the internet or idea in front of customers. Over the long term as some of these companies try to really find their long term business model and squeeze that profitability out it seems like private does become an option. This is Trove day. Do any of you have Trove specific products or services that you’re offering? Any kind of Trove connection.
Anni: We just started our partnership with Tesora a couple of months ago. Like I said, our mission is to enable our care and customer service providers, enterprise customers to stand up a private cloud or to be a public cloud provider. We understand that DBaaS, we call it DBaaS is very popular especially like I said earlier, a lot of SMBs they really don’t want to own IT anymore. Anything as a service especially database is such a very common requirement definitely is a very appreciated. We just partner with Tesora and we’re going through technical due diligence and doing a lot of testing work and hopefully in the future we’ll be able to offer Tesora DBaaS on top of Huawei OpenStack platform.
Jonathan: Okay.
Andre: Bluebox also partnered with Tesora.
Jonathan: Well done Tesora.
Andre: We got pretty far down the road, building up proof of concept etcetera and then big blue showed up and now we’re trying to figure out exactly how we continue to move this thing forward. I was reading an article a few weeks ago, 451 research was talking about how database as a service is responsible for about 150 million dollars in revenue in the year 2012, but it’s expected to be as much as 1.8 billion by next year. I think that there’s real movement in the broader market, desire for database as a service. I actually thing that specifically private cloud has lacked there because I think that what we’re seeing is the traditional IT buyer who has a tendency to error on the side of physical machines for their databases. We just determined the kind of push this thing forward because I think partially it’s because they don’t have the experience with public cloud to recognize the value of database as a service. We think it’s vitally important to move forward there.
Jonathan: Yeah. It’s interesting that you mention that. We do a user survey every six months and the number two workload I think for the last three cycles that we‘ve done the user survey has actually been databases. The Trove service is not the number three service. That’s [unknown 18:35] though. I think that kind of points to what you’re saying, where even if they move it into a virtualized environment or into an OpenStack environment, they’re still thinking about it as, this is a database that runs on a server of some sort and so there’s a lot of opportunity. Obviously lots of databases running in these clouds and putting more tooling round it is a big opportunity.
One of the things that I’d be interested in hearing from somebody who worked across models how do customers handle data in a hybrid deployment. How do they handle the issues of data gravity and databases across public and private use cases? Are they keeping those separate? What are they replicating? Are they using NoSQL databases? What have you guys seen as any of the common scenarios that people are making use of?
Frank: I think that’s one of the challenges, is where is the data. That’s what we’re seeing now. We honestly don’t have a lot of customers that are doing hybrid deployments.
Jonathan: Okay.
Frank: Databases happen to be I think the most popular workload that exists on a SUSE Linux server on a customer today. We have customers who are using databases on SUSE in the enterprise. We have customers that are using SUSE in the public cloud. I haven’t seen a lot of hybrid usage yet. I think precisely because,
Jonathan: Because it’s hard.
Frank: Right. It’s hard. Where the data is the data and the workload running in the same place. I suspect that will get easier over time, but maybe you guys have a better answer.
Vino: We practice around the workloads. To promote the service practice, we actually crated a video, our marketing team, they are sitting here, they created a video at a switch of a button the data is flowing between private to public, public to private. First time I see the video I was like, so. This is something we are evaluating. Primarily we are working with Tesora precisely for that reason. Where we are seeing that? Why we many of our customers data, we are seeing a potential chance to bring in database as a service layered to the DBB. Currently many of our customers are running the up tier on OpenStack and the data tier on typical virtual machines or even on physical boxes on an active passive type of with a shared storage and it will talk to the up tier on OpenStack.
Jonathan: So, Tesora, Tesora, Tesora, what other tools are you seeing in common that people are using across OpenStack environments. OpenStack itself is a set of services that provide those really fundamental building blocks around compute storage and networking. As you were saying what’s really important is all of those workloads that get put on top. What are people using for monitoring for platform a as service, anything with containers. What are the trends that you’re seeing? In terms of the tooling ecosystem.
Frank: I think for us, really a lot of it is homegrown today. Much in the same way that people like to start with upstream OpenStack and then they figure out it’s really hard to deploy, so we get a lot of customers that want to do a distribution because that’s a lot of the benefit. A lot of the tooling is homegrown. We’re seeing a lot of usage of Horizon from a management dashboard standpoint. The thing that I would say drives a lot of decisions is the network and the STN tools. The plethora of STN tools that area out there. Getting the network right is very important for our customers. They have a lot of choices. There are nuances to each of them.
Jonathan: What are some of the nuances? Going to put you on the spot here.
Frank: Yeah. It’s depending on what layer of the network that these vendors address. Some of them address only layer three, so layers one, two and three. Some of them address all layers four through seven. It depends on what the use case is for the customer. We have partnerships with three or four network providers that all kind of do different things. The choice of the Open Stack vendor is almost secondary because they’re keying in on the network. It’s important as an OpenStack provider to build an ecosystem of partners that can plug in and work with your stuff as well. That’s what we’ve ended up doing.
Anni: My observation is I think we’re going through this management tool sprawl era. In fact that’s one of the biggest challenges for a lot of large IT shops because over the years they have inherited a lot of management tools and vendors also they have their own management tools. A lot of vendors even claim that their management tools work in a hybrid cloud environment. You have so many choices, but then which one is one to go. That’s why I think OpenStack is good news for everyone because if everyone is OpenStack compliant then the interoperability will be there and the integration will be a lot easier.
Jonathan: Andre?
Andre: Yeah, I kind of alluded to this earlier. I think about OpenStack as infrastructure as a service I don’t think we have to figure out a lot of these answers because I think public cloud is leading in these ways. We can think about kind of how they’ve empowered the developers, and look at the catalogs that they’ve developed and kind of try to understand how many customers are consuming these services and why are they consuming these services. I think ultimately when I think about infrastructure as a service, it still requires people to build the servers and install operating systems on machines. Ultimately that’s probably not something developers want to do either. You mention PaaS. And I think that platform as a service is an incredible opportunity for us to take one step on top of this IS level and really simplify a bunch of the things that, this whole concept of DevOps has been created to solve this gap here. The PaaS layer really empowers developers. We’ve seen from Heroku’s success, etcetera. Really just having a single, simple dashboard and empowering developers to do what they need to do and just being able to lay that on top of OpenStack is I think really empowering.
Jonathan: I can ask you this because I started one of the first PaaS companies ten years ago. We ran half a million applications dot net and LAMP. We were collectively using more bandwidth than YouTube at the time. I was an early believe in PaaS. However, do you think that PaaS means anything? I’m an old disillusioned PaaS provider now because what I’ve realized eventually people outgrow it. They have a custom need. This was when I was doing what became the RackSpace cloud, at the time we were able to offer them dedicated servers or cloud servers or whatever but do you think that there’s going to be a PaaS model that can meet a broad enough set of needs.
Andre: Yeah, I think that’s a really interesting question. We needed you to have the experience you had to understand we know what this is. I think when you’re talking about net new developers getting on to a platform, it’s probably primarily a lead generation deal right. Where you also have the capacity of offering infrastructure as a service and maybe you add this cataloging feature which allows customers to pick and choose the things they want to utilize after they have feasibly outgrown the PaaS. That’s a public perspective. Maybe that’s a hosted perspective. On the private perspective you have customers who, they’re IT shops, they manage servers. They do this stuff that public cloud is doing and abstracting from customers. But there are these entities that still do this. What they want to do is just simplify this whole lead flow from the developer to IT. I think that there’s a tremendous benefit there. I don’t know if there is a PaaS only model for like a revenue model. I don’t know about that, but I do think that the technology is tremendously valuable. I think it does fill this air gap here between IT and engineering. I think there’s clearly some space there that we need to figure out how to create a buffer for.
Jonathan: Okay. If there’s VMware then that also implies that there’s a lot of Windows in addition to Linux workloads inside of these environments. What are you guys seeing with mixed operating systems and especially Windows specifically?
Andre: We have governed a tight ship at Blue box so far so we have really restricted what customers can do because we’ve been opinionated about it, but more and more we hear the desire to have Hyper-V. We have not dug into that quite yet.
Frank: We have.
Anni: Yeah.
Andre: Tell us about it.
Frank: It’s great. We support Hyper-V. I honestly don’t know how many customers today are taking advantage of running their Windows workloads on OpenStack in SUSE OpenStack cloud, but it’s there and it’s supported. Just having multiple hypervisor support off of this is just a good checkbox for the customer, but that’s more around ESX I think. There are more ESX workloads running on OpenStack for us than there are Hyper-V.
Jonathan: Yeah. Back to the tools discussion. You mentioned that you had, I think you said 50 something partners around the cloud practice that you have. What are the tools, what are the things are being added into OpenStack environments, so these cloud environments to kind of get them to that level where they are fully useful.
Vino: One of the layer from our experience, clearly lacking the feature what customer expecting is the something like the hybrid infrastructure management layer. Having said that, multiple players are there in that area. Many of the early players in this area, whether they were calling as a multi cloud manager or a hybrid cloud manager or a hybrid infrastructure manager, what all it is. The problem with those components are they are pretty much monolithic. It is not a, so take the example Tesora. Once we allowed Tesora to many or our early implementation of OpenStack, it is very tied to integrate the Tesora layer to say, a hybrid cloud management layer. We had to wait for three or four version down the line to have that to reach that hybrid cloud management layer.
This is one of the areas clearly lacking currently because I still believe there is a wide open opportunity to build a hybrid cloud which is more than enough. Also when OpenStack onboarding projects, the products can be plugged to that layer and still that layer can manage OpenStack, the CloudStack, forget about CloudStack, but at least VSphere infrastructure or AWS. That layer should be, it should be a flexible model than a monolithic model. Whatever currently available applications which has claimed to be managed multiple infrastructures, it’s not flexible simply.
Jonathan: Yes. Okay. I think that fills up our time. I just want to thank our panelists here, Frank, Anni, Andre, and Vino. Thank you guys very much for joining us. Sounds like we live in a very exciting period. For technology and especially cloud as there are many opportunities and many challenges to figure out. Thank you guys for talking through them with us today.
The post OpenStack vs. Hybrid vs. AWS : What Does This Mean for Databases in the Cloud? appeared first on Tesora.