12 September 2011 Posted by Paul Burns
This is the fifth in a series of podcasts that were recorded at VMworld 2011 in Las Vegas. In this twelve minute podcast, Chris Pinkham, CEO of Nimubla, talks about his transition to Nimbula, how Nimbula Director is different than other offerings that help organizations build and operate clouds, hybrid cloud computing, the future of cloud computing and much more.
Chris was formerly Vice President, IT Infrastructure at Amazon.com, responsible for the company’s global infrastructure engineering and operations. While in this role, he conceived, proposed and, together with Willem Van Biljon, built Amazon’s Elastic Compute Cloud (EC2), the highly successful public cloud service.
Here is the transcript of this podcast:
| Speaker | Question or Answer |
|---|---|
| Paul: | I’m with Chris Pinkham cofounder and CEO of Nimbula here at VMWorld. And Chris you’ve got a great history with not only ISPs in the background but being a key member of the team that developed Amazon EC2. Maybe we could start with your thoughts on your experience with Amazon EC2 and how that lead you up to Nimbula? |
| Chris: | So Amazon EC2 was really about how do we enable developer productivity by giving them a tool set at pretty high scale that allows developers to more rapidly deploy innovative applications and focus more on building their applications rather than worrying about configuring infrastructure. But doing that in a safe, secure, shared environment which offered a very low cost, very high scalable infrastructure. After a break from Amazon I’m back on the same mission, except by developing technology that allows other organizations, technology companies, technology service providers, infrastructure service providers, to build large sets of data centers, and offer all kinds of technology services to their customers. |
| Paul: | That’s Nimbula director I guess you’re talking about. Maybe you could tell us what the state of that is right now. |
| Chris: | Yes. So we’ve had version 1.0 out for a few months and lots of interesting adoption and evaluation over the last several months. We’re seeing our first production customers come to market with their own services towards the end of Q3. We’ve recently announced we’re bringing out a new major version of Nimbula Director. We’re calling it 1.5 and it’ll be out before the end of September. It fills in a large number of services gaps around storage, persistent storage, using Nimbula disk service, around network address translation support, persistent migratable addresses for transient applications, and a bunch of other service provider features and support for things like better monitoring and metering of application usage, so service providers can charge that or account for it on a client by client basis. |
| Paul: | There’s, as you know, a number of products out there that are taking somewhat similar approaches. Maybe some of them are going after service providers only, or maybe enterprise private clouds only. But you’re going after both. Any thoughts on what it takes to make your product work in both of those environments? |
| Chris: | Yeah, our product has a lot of service provider concepts which we think are applicable even in private organizations. Things like a strong identity model where individual end users could be grouped together, policies can be applied to these groups, infrastructure is sharable, but in a much more controlled, much more secure, and much more auditable kind of way than with traditional enterprise infrastructure products. So those are service provider concepts, but we think that they improve enterprise compute a great deal. So in areas where customers get value out of the dynamic self service that a product like Nimbula Director offers, then those notions of security, and resource isolation, and so on, they are very valuable.
In terms of other products that folks are using, there is a lot of overlap in terms of vision and mission, but that is another infrastructure as a service software providers and projects. I would say that a large percentage of the difference has to do with our absolute focus on systems automation, and having all of these concepts work in a large shared pool of compute. Not to have these small little islands into which users are constrained and have those replicated everywhere. So we’re big on automation, we’re big on scale, and we think the ideas of identity, authorization, and security apply equally well whether you’re selling services for a living or just want a more secured, sharable, self-service infrastructure internally. |
| Paul: | I spoke with Willem a few months back and he sort of indicated what you guys view as the importance of the hybrid cloud going forward. At least the co-existence of public and private clouds. Do you have more thoughts on that in terms of the need for hybrid clouds, or maybe even dynamic cloud bursting? Are you seeing those use cases? Do you think they’ll be coming up in the future? |
| Chris: | Yeah. I think that is being approached from a number of different ways. Nimbula was founded on the premise that there will be large infrastructure service providers, some of which hopefully will be running Nimbula software for a long time. There’ll be smaller service providers, perhaps with more customer business models in niche market segments. But then particularly large customers, large enterprises will end up owning their own computers for a long time. What is important is that we figure out a way to stitch all of those together so that users have access to this global pool of compute so that they can just get what they need done as quickly as possible in the right place.
So that’s sort of the hybrid world we visualize. It’ll take time to get there. It’s a multiyear project for sure. It won’t happen overnight. Through this infrastructure there’ll just be an absolute flood of applications in ways that we have yet to see which allows ultimate business users to figure out new ways of consuming applications. We think smaller organizations will probably stop owning servers, stop running their own email systems, stop running their own ERP, CRM, HR oriented systems in favor of using application hosted in this environment. Larger organizations will continue to host some of their own. A lot of the measure of whether somebody has their own data centers or not, are to what extent they are a technology company. Some of the larger EC2 customers for example are now diversifying by having a mixed environment. So going the other direction to the usual VMware virtualization into public cloud direction, they’re trying to figure out how to replicate the Amazon environment with which they have grown up with inside their own facilities. And that’s just about diversity, it’s around having options in the event of business issues or technical disaster issues. There are lots of use cases for those models. |
| Paul: | So it sounds like a lot still to develop and things that are sort of unpredictable at this point as we’ve got some people out there innovating and coming up with new ways of using hybrid clouds. Are you seeing any desire yet for dynamic cloud bursting? I guess when I say dynamic I just mean sort of on the fly, you know you’ve got an application running, maybe in a private cloud, and then that application wants some resources in a public cloud soon. So not necessarily planed, but more dynamic. |
| Chris: | I guess we’re not seeing use cases where applications automatically spread themselves around in different data centers. I mean there are data locality issues and other kinds of management challenges around that.
Use cases that are more “I need to spin up a service, I need to spin up instances of an application, and what is the best place to do that, and I want to do it really quickly.” Those are real use cases. Large EC2 customers, some Nimbula customers make those decisions on a regular basis. What’s the right location for this application to run? But then they want it immediately, they don’t want to have to wait multiple hours or days for virtual data centers to be configured and so on. They need the system. They need to have trust in the system such that when they make that request it’ll be fulfilled instantly. |
| Paul: | Yeah that is really helpful. I come across that question quite a bit and it seems to be fairly rare so far. Maybe some people experimenting with kind of that dynamic application bursting. But then still hoping or believing that might develop in the future. So I guess time will tell.
What sort of things are you doing around the eco-system? Do you have some active partner programs going or anything there that you’d like to talk about? |
| Chris: | Sure. We’ve announced a range of partnerships with large technology companies as well as small. Most recently we’ve been working with Gluster and Cloud Cruiser and Elastic Cloud. These are smaller technology companies who provide niche services around metering or developer test environments in cloud infrastructures. We’re integrating with their applications or their storage services in the case of Gluster so that customers can have a seamless experience no matter which infrastructure they want to operate in.
In addition we’re working on a new services platform which allows developers to build composite applications consisting of services developed by others. Where these can be just declared and launched without having to download and install and manage software. So there is a range of partnerships with the larger companies such as RedHat for example whom we announced recently as well as the smaller organizations. |
| Paul: | What do you think the role is going to be for Nimbula Director in terms of getting applications provisioned or maybe changed on the fly? Change and configuration management and that sort of thing… |
| Chris: | So we’ve definitely started with getting the infrastructure concepts right, and getting those scalable and building them out to all forms of infrastructure. Over the next twelve months we’ll be very much focused on extending the platform concept so that it’s not just about configuring machines. I think we need to get away from thinking of these containers as machines that we get in and configure and we build networks and we address things and we worry about those things. We need to change what we worry about. What we need to worry about is exactly what does the application need. So what is the application definition? In what kind of format is this? It could be as simple as a Java file. It could be a virtual machine container. It could be some overlay on top of a virtual machine container. So that is just “How do we support in a basic kind of way different kinds of applications in the environment?”
The second thing that we’re working on is enabling new services. So, one of the reasons that Amazon web services is so successful is that they have a range of different kinds of services that allow their customers not to worry about configuring other kinds of services. We’re going to take some of those concepts a bit further where the services themselves become sharable in very controllable kinds of ways. So developers can come up with, for example, a no-SQL database service configuration, and then offer that to named customers anywhere in any Nimbula datacenter. And those customers can simply invoke that service with some parameters. For example: how big do you want it to be? Where do you want it to be? And then within minutes you’ll have database clusters dedicated to you, running in the place that you want them to be. So we’ll be working on those kinds of services and talking more about those towards the end of the year. But there you can see the shift from raw infrastructure components, networks and machines, and so on, into how do I very rapidly stand up a useful productive application environment. |
| Paul: | That’s really helpful. I’m sure you’ve probably observed this to some degree, but with your heritage from Amazon EC2 and now into Nimbula I think there are a lot of people in the industry really rooting for you and kind of watching out for where things are headed.
So with that in mind, is there anything else you’d like to add or anything else you’d like to share? |
| Chris: | I appreciate that. The weight of expectation falls heavily on the shoulders. I certainly hope that we meet everybody’s expectations and needs.
The main thing that I’d like to share is that this is a long term journey. I was at the start of the internet evolution in the late 1980’s and we all had very bold predictions about how the internet was going to change everything. And of course it has changed everything, but it happened gradually over twenty years. And I think the same will be true about cloud computing. There is a lot of noise in the market place at the moment. Noise being made by describing old infrastructure as cloud computing, by making small changes, calling that cloud computing. The long term vision of Nimbula is that of this infinite, seamless pool of application support resources that developers and customers around the world can consume with ease and at low cost. It’ll take a couple of decades for that to be universal in the same way that the Internet is universal. But I have no doubt that it will happen. |
| Paul: | That’s a really helpful perspective. I really want to thank you for your time today and we’re all looking forward to hearing more. |
| Chris: | Thank you Paul |
| Paul: | Thank you |
Recent blog posts
- Compliance Drives Buying Decisions in the Cloud
- Changing Competitive Dynamics for IaaS Providers
- Good News for Cloud Job Seekers in 2015
- Colocation: The New Center of Hybrid IT
- The Problem with Forced Cloud Reboots
- OnApp Federation Grows Stronger with SolusVM Acquisition
- HP to Acquire Eucalyptus: Questions, Comments and Concerns
- Is It Really ‘Game Over’ for Most IaaS Providers?
- The Big Cloud Winners Are…Not Just the Hyperscale Players
- Cloud Federation: Dream or Reality?