You might think we’d cancel the OpenStack Podcast on the rare occasions when both Jeff and I have the flu at the same time. But you’d be wrong. This week, we interviewed EMC Cloud Architect Shamail Tahir (remotely). Or at least we think we did. The transcript of the interview suggests that we were in fact present, but not in the most coherent way.
The good thing is that it didn’t matter. Shamail is a boss when it comes to OpenStack, and he proved that by delivering interesting insights on pretty much every hot topic we could think of, including:
- What EMC is doing with OpenStack
- What he and the other participants talked about at the “hidden influencer” meeting in Paris
- Why you should run certification tests on your OpenStack cloud
- Why we need an OpenStack roadmap
- Why the number of OpenStack projects will continue to grow over the next few years
- What role containerization will play with OpenStack
- Whether or not DIY OpenStack is a good idea
For a full transcript of the interview, click read more below.
Niki: Welcome, welcome, welcome, everyone, to the OpenStack Podcast. It is Tuesday, January 27th. I am Niki Acosta.
Jeff: I’m Jeff Dickey, from Redapt.
Niki: We have an awesome guest with us today, Shamail Tahir. Shamail, introduce yourself for us.
Shamail: Hey, everyone. This is Shamail Tahir. I’m with EMC, and I am a cloud architect there. I’m actually focused on our cloud innovations, in research and development, mainly, along with being a part of the team that is driving OpenStack strategy at EMC.
Niki: Sweet. We were just chatting, and it sounds like we have a lot to talk about today. We typically always start by asking about your personal journey through technology; how you started, and what got you into OpenStack. If you want to dial it back a few years, without potentially dating yourself there, we’d love to hear more about your entry into tech, and how you got to be where you are today, in your awesome position at EMC.
Shamail: Awesome. It’s definitely been a semi-long journey already, but it started off actually at a VAR, working as a consultant. In those days, I was kind of doing a little bit of everything; mainly, going out and doing systems architecture and just some deployments.
From there, I actually turned into a customer for a bit. On the customer side actually, I was focused more on network administration, systems engineering still, as well as getting into some of the enterprise monitoring solutions out there, like HP OpenView, TMG, et cetera.
That was actually a really good experience to have, because that kind of gave me my first purview into challenges with scale, as well as distributed systems. From that perspective, I especially was doing a lot of agent management and large-scale systems management and monitoring, so that was a good background.
From there, I eventually went to a vendor, EMC. Over here, I’ve actually done multiple things as well. I actually started off in our professional services group, doing implementations and deployments. From there, I actually moved into our engineering team for our NAS product, and started focusing on file, as well as cloud gateways. From cloud gateways, I started looking at the cloud platforms that they can use, and OpenStack was one of the things that I started looking at at that point. That was really the start of where I started paying attention to OpenStack and following it.
Over time, that became my main area of focus. Basically, I stopped focusing on the cloud gateway aspect, and started focusing more on the cloud platform itself.
Niki: EMC has definitely been involved with OpenStack to a pretty wide extent at this point. I can imagine being at Cisco, which traditionally was a hardware company, that you guys are going through sort of a similar transition in moving from hardware, now moving up the stack, more into making sure that kind of everything has APIs, or they’re software-defined stuff. Do you see that sort of transition taking place at EMC? Can you tell us about that?
Shamail: Yeah, absolutely, and I think that transition was started very publicly, probably a couple of years ago, even, when we first started launching some of our software-defined storage solutions, and going into market with journey to cloud type messaging. We’ve been there for a while, but from a cultural perspective, absolutely. We’ve been transitioning, in terms of messaging sets, as well as making sure that our products are [inaudible 00:03:42] and they can integrate and operate sufficiently in this new data center architecture.
Niki: Are you guys taking your existing employee base, and kind of leveling up and kind of providing them with new skills? Is there a hiring exercise going on at EMC to bring in that type of talent?
Shamail: A bit of both, and quite honestly, who isn’t hiring OpenStack talent right now? I would say yes to both.
Niki: Nice. Jeff, I’m… As I often do, I’m hogging the mic here. Surely, you have some questions you wanted to ask Shamail.
Jeff: I was curious, and you were talking about, just to step back a little bit, how you got focused on OpenStack. What was that? What was the driver that really got you behind OpenStack? What did you see? What’s that kind of vision?
Shamail: To me, it was about the enablement and the agility that the platform offered. Looking at it as a shift in how IT and infrastructure can be consumed, was I think the thing that really drove me deeper into the platform. It started off, like I said, I was looking at it more as a landing spot for some of our other stuff. When I started looking at the platform, and more specifically at the vast amount of services and endpoints that are available, it became really interesting.
The other big part as the community aspect of it is, OpenStack has a very good community ecosystem. Yeah, there’s a lot of material, and at times, it’s daunting to dig in. People are willing to help you if you jump on IRC or if you talk to people. I think the community is really what got me to the next level with OpenStack, which is being able to interface with people, and see their passion and have their passion be inherited by me as well.
Niki: Did you guys dig into the Eucalyptus and the CloudStacks, and even some of the VMware solutions? Did you just not see what you… Were you not liking what you saw? Was the community factor kind of a win out of the gate?
Shamail: Also, this is me personally. From an EMC perspective, I think we’ve kind of looked at all those platforms. Obviously, with the VMware EAC solution, and VMware stackware, we’re definitely involved with that one.
For me personally, I did look at all of those, and I actually even looked at OpenNebula as well, when I was starting to look at various cloud platforms. For me, it was the modularity as well as the community that kind of attracted me towards OpenStack specifically. The thing was, being in an OpenStack community, if you were a storage domain expert, you could work on storage. If you were a compute domain expert, you could work on compute, et cetera. With some of the other platforms, it was kind of more monolithic and less modular, in a sense.
Niki: It really gave you guys an opportunity at EMC to specialize and hone in, in an area where you guys were comfortable, it sounds like.
Shamail: Yeah, it was a good starting point. There was a lot of value from the structure, as well as we wanted to make sure that we enabled what we were hearing from our customers, and we were hearing that from our customers as well.
Niki: One thing I find myself doing, even now quite a bit, especially in the enterprise space, is explaining what OpenStack is. I’m starting to realize and recognize that everyone has their own sort of take or perspective, when they talk about OpenStack and the value that it provides.
How are you explaining OpenStack to your customers and your prospects?
Shamail: OpenStack is a full-blown platform. I would say, the way I look at it is, the base layer is the IaaS layer that it offers, but then it offers so much more beyond that. The best thing about OpenStack is, it offers multiple services and multiple widgets, if you will, that you can combine to build your own platform, that’s relevant for your organization.
I kind of explained it as multiple stacks or layers, if you will, with the IaaS layer being one layer, then the next step being services that actually help with the cloud management deployment itself. These are things like Ceilometer, TripleO, et cetera. The last piece is like advanced services, or services that actually consume the IaaS layer, to provide additional value-add; things like Sahara, Trove, et cetera. It’s a multi-layer stack, in a sense.
Also, if we look at what is it from a directional perspective; is it an API product, distribution, et cetera, you know the standard taxonomy or categories that are being used, I would say straight out of [repo 00:08:09], it’s probably more than an API has to be. Otherwise, I think it just becomes a standard. At the same time, it’s probably less than a distribution, because it doesn’t really have the packaging and the coupling of projects together. I think it’s something more than an API, but less than distribution.
Niki: Are you finding that people are looking… What are your customers, by the time you get to them, what are they looking for? Have they already deployed, and gone down that DIY path with OpenStack? Are they looking for a VMware alternative? What is your sweet spot when you talk to customers? What are they coming from?
Shamail: I think the two major types of customers that I’ve seen, are the ones that are researching it, and have kind of built an open source first strategy, if you will, where going forward, they’re kind of looking at open source as being something they’re evaluating, and if it meets their needs, that’s where they’re going to go. At the same time, they’re at the research phase of that strategy right now.
The one ones that I see typically are customers that have started with the do-it-yourself path, and going from a POC to pilot to production, have hit an inflection point where they’re looking at someone to come in from either a skill set perspective, or from a product perspective, to help them scale to that next level, if you will.
Niki: Nice. I think we’re seeing similar stuff, too. Jeff, you look like you’re pondering a question.
Jeff: Yeah, it’s funny to talk about that, because I just want to go on kind of a little rant here about… It seems like folks have a different view. I’m working with a lot of folks that are doing OpenStack, that are trying to build RACK scale architecture, and kind of forgetting about the whole DEV/TEST, QA type environments. It seems like we’re skipping that in a lot of these deployments, and it’s kind of… Are you seeing people in these deployments? What’s the lifecycle look like?
Shamail: When you say DEV/TEST and skipping that, are you referring to kind of from a data center use case perspective? They’re saying, “How do I deploy production resources and work loads on OpenStack, versus using it for DEV and QA?
Jeff: Yeah, that’s kind of… What we’re not seeing is, we’re not seeing the infrastructure similarities that we used to see back in the day of, this is… If I was going to be deploying VNX in production, then I would have the similar VNX in TEST, DEV and QA, and kind of have this roll out. We’ve gotten this stuff… Infrastructure is code, but we still have this underlying hardware, and I’m curious of EMC’s take on that, and being such an instrumental part of the infrastructure stack.
Shamail: I think we are seeing actually both. From the enterprise customer perspective, I would say that I see more adoption initially, at least from a pilot program perspective, in the DEV QA space, because it seems, to be honest, more like low-hanging fruit, from their perspective, where they can try it out, understand it, learn the ins and outs, the pain points, the benefits, et cetera, before they start adoption from a production perspective.
I’m actually seeing DEV and QA as being a very prominent starting point for a lot of our customers in enterprise, actually.
Jeff: That’s great to hear. Another question I had was around Ceph. What’s EMC’s view on Ceph? Are you able to talk about what you guys are doing, or are you talking about Ceph? What does that look like to you guys?
Shamail: Ceph is obviously a really strong community player, and they are very involved in a lot of aspects, and they have been involved in a lot of aspects, from a storage perspective, both on the object and sender side as well. They’re actually a really good enabler, because they are an open source project for some of the reference of limitations, and open source deployments that are happening, of OpenStack.
From our perspective, of course, we are looking at them, and we are also figuring out, where does our definition and their definition of software-defined storage intersect and deviate at the same time?
Jeff: Are you seeing Ceph change at scale, where people are starting with Ceph, and you kind of come in with a solution?
Shamail: Me personally, no, probably because I’m more into Office of CTO, R&D space, so I’m not getting that feel to feedback directly, so I probably couldn’t comment on that directly myself.
Jeff: What about… Sorry, Niki, I feel like I’m… Now I’m dominating.
Jeff: What’s going on with… Are you able to talk about anything with Cloudscaling, or what our friends are doing? Are they in the…?
Niki: How’s Randy?
Jeff: Yeah, I haven’t heard from him in a while.
Shamail: He’s doing good. He was actually supposed to be in Palo Alto, but he had some conflict come up, so unfortunately he’s not here. I was looking forward to seeing him actually, in person again. No, they’re doing great, and they’re actually all over the place still in the community. I’m sure you run into them in meet-ups and various aspects of community management engagements.
From an internal perspective, they’ve been really good for us, because I think some of it is infusing that DNA that Cloudscaling brings with them. Another is infusing the young talent that they bring with them as well. They are busy all over the place at EMC, and the challenge is making sure that there’s a shield to let them do what they need to do, effectively.
Yeah, it’s a great team, great products, and they still have their existing customer base as well.
Jeff: That’s good to hear. I think it was a good acquisition. That’s just a bunch of great folks over there.
Niki: I have to jump in real quick and let folks know this is… Jeff and I are both probably likely coming down with the flu. Mine is from a co-worker, who probably caught it in the airport when someone hurled on him. Jeff’s, I think, is from a children’s birthday party. If we seem a bit off today, it’s because both of us are a little under the weather. Thank God we have a great guest who is pulling it together for us.
You mentioned being… I just had to preface that, because we’re both like, “Uh.”
Jeff: Yes, I have a fever, so I don’t think anything coming out of my mouth is coming out right.
Niki: I’m walking around with sanitizer spray, sanitizing everything. You are in Palo Alto now, and for those of you who don’t know, there is a product working group … a product management working group within the OpenStack community. Both today and I believe tomorrow, or is it yesterday and today, you guys are in Palo Alto and meeting, and having some really interesting conversations.
We were talking about some of the things that are going on there. For those that couldn’t make it, and for those who don’t know what the group is, can you give us a definition of what that group is, what it does, and tell us about what’s going on there in Palo Alto, both yesterday and what’s planned for today?
Shamail: Yeah, absolutely. Just some background and context is, this group was originally started when a few people got together. I believe it was Alison, Rob, Shawn, Randy, and a few others that decided to figure out, how does OpenStack work as a product, and how does it move forward as a product, beyond being a set of projects, and being an integrated release, if you will?
From that perspective, there was a kick-off meeting at the Paris Summit, called hidden influencers. In that meeting, we kind of got together, and we kind of discussed, what’s our charter, and what are we trying to achieve?
A few things that we discussed were things along the lines of … Again, as I was saying … How does OpenStack, from a person who is new to OpenStack, consume it as more of a product versus downloading an integrated release and figuring out, “Which projects do I use? Which ones don’t I use?” It can be pretty daunting, to someone who is not in the community, and not immersed in OpenStack day-to-day, the way it’s structured today.
The other big thing was, how do we work on strategically identifying a multi-release road map, if you will, or a strategy around multi-release? Right now, everything’s kind of operating on a 6-month cadence, if you will.
Those were the two big things that we wanted to do, and the last thing was getting a new persona involved in the community, which is a product manager which, the product manager is a loosely-used term here, because to be honest, in the meeting, we have product managers, we have operators, we have engineering and R&D managers, and the whole purpose is, how do we get this group of people that’s probably been around OpenStack for 2, 3 plus years in some cases, but they’ve been kind of behind the scenes. How do we get them to kind of express their viewpoints as well, going forward?
Niki: Yeah, being a product manager I think is really interesting, because you sit somewhere in between… You’re accountable to sales, I guess. It’s your responsibility to deliver what sales is asking you for. You’re also… You have this other responsibility, to really be involved in the OpenStack community, understand what’s going there.
A lot of times, you’re the one prioritizing your own product road map. You’re going to take OpenStack from trunk, you are going to pick which components you think customers are going to want, based on what sales is asking you for, or based on which projects are on the wish list for a lot of your customers, and then it’s your responsibility to prioritize what you are going to release in your own product.
It’s a tough place to be. I think you talk to people who’ve gone to the Summit, there’s two groups, two main camps. You have the business sort of vendor side of the summit, and then you have your developer and engineering component at the summit. I think there’s that kind of desire to figure out a way to bring those two groups together. I think the product manager is a very good place to start, if you are trying to do that.
You also talked about, before the podcast, some of the stuff that Rob Hirschfeld is working on. We had a fantastic podcast with Rob a couple of weeks ago, but fill us in on what Rob presented to this group as it pertains to RefStack and DefCore, and if you could define those terms for people who are kind of new to the OpenStack space. Fill us it on what that is, and what the intent is, and tell us what Rob’s working on.
Shamail: Yep, absolutely. When Rob was on your podcast last month … a while ago … he actually covered a lot of the same content yesterday at the Summit, or the meet-up, for us. For people that are new to DefCore RefStack as initiatives, DefCore is an initiative to kind of define the core of OpenStack, from a marking perspective.
Going forward, DefCore hopefully will identify some criteria that lets people understand whether something is or isn’t OpenStack, as well as what type of mark they can legally use, from an OpenStack perspective. RefStack is the component that actually runs the testing to validate that functionality, that’s needed. DefCore might define certain functionality, such as can you load a certain type of… Can you load images? RefStack would be the testing mechanism to make sure that you are compliant with that functional requirement.
Secondly, in DefCore, there’s also a designated code section, which not only means that you have to have functional parity with the definition of DefCore, but you may also have to have some portions of the open source code as well, within your product.
Niki: Yeah, I think this is important for a lot of reasons. The big reason is, from a user perspective, if you are evaluating OpenStack, you have a multitude of different options, and you have a lot of things that may look and feel like OpenStack, that are very, very different than what OpenStack is about.
I think the effort around RefStack and DefCore is figuring out what it takes to be called OpenStack. There’s a list of criteria, like do you have to have some of these ancillary projects included? Which projects should be included in that definition of OpenStack? Just to give users around the globe some consistency in terms of what they expect from OpenStack.
It’s tricky, because you have so many vendors who have so many different perspectives that are involved on so many different layers, at what point is what you’re offering not really OpenStack? How do you add value in the context of OpenStack, and represent that value, if your flavor of OpenStack, or your distro of OpenStack, or your product around OpenStack, is so vastly different? Can you expect the same APIs from one OpenStack cloud to another OpenStack cloud, and get to that ideal world of not being locked into a particular vendor, or being able to move workloads from one provider to another?
I think it’s a hugely important effort that is going on in the community, and based on what we discussed with Rob, it’s a hard thing to do. There’s going to be a lot of people who are going to want to make sure that they can still call their thing OpenStack, even if they don’t have everything in that DefCore definition included in their product.
Shamail: Yeah, and it’s even harder, because you’re establishing this definition after four years of the project being up and running. It’s a lot harder to actually go and figure out, how do we get people that maybe invested a certain way two or three years ago, and then do we go back to them and tell them that you’re no longer OpenStack now, because of how we’re defining it?
It will be interesting, so that definition and those tests will be really critical. I think one of the things that people should do is, right now, DefCore has Havana certification standards defined, which are advisory. The encouragement that Rob gave everyone in the room, and he wants to give to the broader community as well, is go out there and run the certification tests, just to see, does your cloud pass? Does it not pass? That will help the committee get additional data that’s needed, to be able to really fine-tune some of the testing, as well as understand, where do most clouds sit today?
I think that data will be really useful, and it’s already out there. The encouragement is, if you have an OpenStack cloud running, please run RefStack against it to see what the outcome is.
Niki: Could we talk about the roadmap component a little bit as well? I think we’ve had themes within OpenStack. One theme, for example, was stability. Let’s just focus on the stability of OpenStack. Lately it’s been like, let’s focus on OpenStack in the context of enterprise. I think where a lot of people are getting tripped up, is the road map. There’s so many people now contributing to OpenStack, there’s a lot of sort of wish lists that a lot of people have, from a user and a vendor and an operator perspective.
Tell us why the road map discussion matters, and tell us about the conversations that are going on around it, trying to create a sort of broader road map for OpenStack.
Shamail: Absolutely. There’s multiple factors on why it’s important. One is, so obviously, we have some clear cadence of what we’re trying to do, and we can hopefully use those. Those themes still play into this, but then you take the themes, and you kind of break them into hopefully smaller, tangible requirements that you can kind of iterate through in a multi-release manner, to get to the eventual goal of whatever the theme was.
Targeting stability as a theme for a 6-month release is good, but there’s so much to do there, that it’s going to be really hard to accomplish that theme, within one release.
Niki: It’s so broad, right?
Shamail: Exactly. The other piece of that is also, you have to get cross-project buy-in as well. If we say that we want to improve, let’s say, logging, then just because one project says, “Yeah, this is something that we agree with. It’s all well, and we’re going to do it.” That’s good, but at the same time, if there’s no clear alignment and consensus with all the projects and the various people that are involved, that yes, logging is something that addresses operational burden or ease of use, and that’s the theme that we’re targeting, so therefore we’re going to implement that, if two out of 12 projects change your logging, that, to the user, does not deliver the benefit yet. We have to get more alignment on a cross-project level as well, which I’m hoping the roadmap helps with as well.
Last but not least is also the fact that once we have a release roadmap that’s probably broader, and probably more long-term, hopefully having it publicly available and be transparent in that nature, will help customers look at, what are the capabilities of OpenStack? What is OpenStack focusing on? Being able to make that decision internally, within their IT strategy of when it intersects with their needs, I think we’ll benefit from that.
Basically, people will see that based on where it is, and based on where it’s going, and based on where I am and where I want to go, this is the intersection point of where we should look at OpenStack, or consume OpenStack, or deploy OpenStack.
Niki: If I was an enterprise decision-maker watching this podcast, I might say, “Wow, I’m seeing how the sausage is made, and frankly, OpenStack just seems F’d up.” What kind of reassurances can you give people about these efforts that are going on, and why OpenStack actually is a good decision? Why are these discussions important?
Shamail: These discussions are important because again, overall, the platform is really relevant. It can serve multiple needs for multiple customers, and customer types, even. The reason these discussions are relevant, is because we are experiencing a phenomenal amount of growth, and we have to figure out, how do we take that growth, and also build new mechanisms around it, that let us scale for the volume that we’re seeing now, from an adoption perspective, from an enquiry perspective, et cetera?
I think all of these things are good, because they kind of, in the long run, not only work on making the product effectively more stable, let’s say, but it also gives more transparency and more long-term vision to end users that are looking at OpenStack. Again, if you’re new to a project, it’s really hard figuring out, where’s it going?
As you mentioned, stability being a theme in the community, we know that. I think from a user perspective, it’s just a lot harder to figure that out, when you’re trying to jump into OpenStack initially. I think these things give clarity, and they give some sort of, what’s the word I’m looking for here… Commonality, as well, across projects.
Jeff: What’s … If you’re wearing your octo glasses, what does it look like two years from now? Are we going to have less projects, like tightly integrated projects, or are we going to have more projects, like decoupled, more … I don’t want to say strategic, but like Neutron may kind of break apart. Which way, when we look back two years from now, do you think we’re going to be?
Shamail: I definitely think there’s going to be more projects two years from now, and some of the writing is on the wall already, right? If you look at Neutron, there’s discussion around a Neutron service split going on right now. At the same time, if you look at the changes that just are still a work in progress but are happening, moving from an integrated release cycle, to moving to a de-tagging approach, if you will, and being able to say that going forward, things that were once in the StackForge namespace, and had to go through the incubation process to get into the OpenStack namespace, now can apply for application, and move under the OpenStack namespace. That will open a lot of doors.
The other barrier that was removed in that transition from integrated release to the tagging model, or potentially will be removed, I guess, from what I’m hearing, is that you can have multiple projects, even, that might provide similar services. That doesn’t exist today. Now, if I have multiple projects providing similar services, at the end of the day, consumption is going to be the dictator of which projects win and which ones lose. I don’t think there’s going to be any restriction, so that’s why I think there’s going to be more projects with some getting more success, some getting less success. At the same time, the number will grow.
Niki: I can’t imagine this process. It makes it really difficult for folks to be able to A, find and build a team to be able to deploy OpenStack, but B, how to make sure that your needs will continue to be met, and your wish list items will be developed in the context of running OpenStack within your organization.
Do you need a vendor to do this for you? Do you think people need a vendor? What is the level of effort required, to be able to do this on a sort of DIY basis?
Shamail: I think that partly goes back to what organizational skill sets you have internally as well. There’s possibly a shift as well, from a skill set perspective. I think for some customers, that might have engineering as one of their core values. I think they might be able to do it off-hand. In other instances, if there is … If engineering is something that you have to build as a skill set, then a vendor might be able to give you that assistance you need, until you can make that transformation happen internally.
I think it’s a bit of both, but I think a lot of it depends on the organization culture and structure and skill sets within customers, on whether they can do do-it-yourself or not.
Niki: Do you worry that people who go the DIY route are just so overwhelmed by everything, and all the incubated and core projects and decisions, and how you’re accomplishing HA and all that other stuff, that they’re getting a bad taste in their mouth as it pertains to OpenStack?
Shamail: Yes and no, and I think coming back to the tagging stuff we were just discussing a second ago, I think that is some of the basis of why that change initiated as well, was from the fact that if you look at Terry’s blog posts on the topic, as well as Randy Bias’, one of the things they discussed was the fact that right now, we have an integrated release. That integrated release includes all incubated and integrated… sorry, integrated projects.
When you do that, all those projects are available. There really isn’t anything telling a person what services are needed, at a minimal, viable level, to build an OpenStack cloud. You get all packages, or you get none. I think going forward to a tagging system, hopefully we’ll get to a point where we can have releases that basically make deploying OpenStack initially a lot easier. I think right now, there are some challenges, because it is daunting. You get the integrated release, and it has everything.
At the same time, I think in the future, with some of the work that Terry and the TC are doing, hopefully that’ll become easier for new adopters.
Niki: It definitely sets the stage for more ecosystem options and participation…
Niki: …Which is killer. As a consumer of OpenStack, I don’t necessarily want to have to tap into 8 different vendors to try to get a full cloud deployed. Hopefully, a vendor is doing a good job in bringing that together. Just the possibilities in terms of, let’s say I’m a customer, and I want to deploy OpenStack for purposes of a massive Hadoop environment. Being able to go with the intent to say, “I want to deploy Hadoop. Show me what I need to do that,” or “Show me some best practices on what I can use to make that possible,” is going to be huge.
That probably would apply to the whole part of the stack, I would imagine, all the way down from the concrete all the way up through the API.
Shamail: Absolutely. You combine that with things like DefCore, which give us some level of predictability of what any given vendor or product will have, in terms of functionality and capabilities; I think it definitely sets the stage for a broader ecosystem, as well as a more standardized expectation from OpenStack as well, when you consume it from a customer perspective.
Niki: We have a … I don’t know if this is a question. Larry Smith asked, “Looking at ELK stack for logging across projects … is the community looking at ELK stack for logging across projects?”
Shamail: I know that… from the community perspective, I know there’s multiple StackForce projects they’re using to ELK stack right now, to look at logging. One in particular is Monasca, which the folks at Helion kind of open sourced initially, but Monasca is one of those that’s kind of tracking both logging as well as events correlation, and they are definitely looking at ELK right now, and they use them in memory solutions right now as well. Yeah, there’s multiple projects that I can think of… two or three at least… That are looking at ELK right now.
Niki: Awesome. Jeff? How you doing, Jeff? You look so out of it, Jeff. You’re on mute, Jeff.
Jeff: I am out of it. I need some Gatorade and a cold towel. It’s been a pleasure talking with you, because again, we’re so fortunate to get guests like you that are so knowledgeable and awesome. How do you stay on top of everything? How do you… What does this look like for you, consuming and gathering and collecting this massive amount of knowledge around OpenStack that you have?
Shamail: I think, as I mentioned initially when I was saying how I got into OpenStack as well, there’s a lot of information out there. At the same time, there isn’t a single source, either. The answer’s going to be multiple sources here.
One is obviously the Wiki pages on OpenStack. I do actually try to participate in IRC, in certain projects that I’m more closely aligned with, and then with others, I follow the meeting minutes after the meetings have ended in IRC as well. The other piece is going to local meet-ups and A, learning … Each meet-up usually has some subject that they’re focused on, and so that helps expand its new areas, as well as network, so networking helps as well, from that perspective.
The final piece is to kind of get… When I look into OpenStack, I like IRC and I like real-time discussions, because that’s giving me a present and future picture of what’s happening. I like to go back and look at old summit videos as well, to get some context on what’s the history here.
I think getting a snapshot of a position, of a project in a point in time, is good, because it gives you a scope of what’s the project capable of, and how can I use this today, but at the same time getting some historical context helps as well. You can see, what were some of the original thoughts, and how did the morph, and what were the use cases that drove those changes as well?
Jeff: That’s good.
Niki: It’s good to get some context. I’m assuming, based on your Twitter profile, that you probably get a lot of information there as well, right?
Shamail: Yeah, absolutely.
Niki: Do you have search columns on Twitter? What are you searching for every day?
Shamail: Not too much. I’m doing mainly OpenStack. I follow that tag every once in a while. For the most part, most of my information is coming more from Wiki and the community itself.
Jeff: What about software companies that we should be watching, including on the EMC side? What’s exciting you right now? What should we have in our radar?
Shamail: I think we kind of set the stage, from our perspective anyway, with the Cloudscaling acquisition that happened, as well as some of the transitions you’ve seen to software-defined storage. I think 2015, 2016 will be really exciting years, from our perspective, as we have some interesting announcements coming up in this space.
Also, I think just following naturally Docker, and anything in that ecosystem, is also interesting right now. There’s companies coming out with backup solutions, SCM solutions for containers, et cetera, and that’s also interesting and intriguing.
Niki: What does EMC… If you could sum up everything at EMC’s doing around OpenStack, and do a couple sentences, what are you guys helping customers do in terms of work cases? What are things that maybe people don’t know that you guys are doing, especially as it pertains to the Cloudscaling acquisition?
Shamail: Everything that we’ve been doing is… This is a pretty common understanding for most people that are at EMC, that we are huge believers and enablers of customer choice. From that perspective, around the OpenStack ecosystem, we’re doing a lot there from A, moving up from just pure driver contributions, to actually helping from a community perspective, as well as moving more into core contribution processes as well.
We’re actually looking at, how do we take some of our enterprise customer feedback, and some of our storage domain knowledge, and kind of work with the community to discuss some topics that we might be able to help with, from that front.
We’re also obviously participating in multiple working groups, as we were talking about the product working group a while ago. We’re also involved with a few other working groups there as well. That’s kind of what we’re doing with OpenStack.
Again, going to the choice perspective, and from a product perspective, EMC is A, we announced a referenced architecture program around OpenStack at the last Summit, and that referenced architecture program basically will allow people to consume EMC assets with distributions such as Mirantis, Canonical, and RedHat, in a certified manner.
We’re also building some additional technologies and products, based on some of the Cloudscaling talent, and where they wanted to go as well.
Niki: We have a question that popped up in the chat, and I’m not sure you know the answer to this, but we’ll ask it anyway. How is Open Contrail going to help with Neutron scaling? Do you have any thoughts on that?
Shamail: Unfortunately, I don’t. You’re on the mark. I could talk about it, but I don’t have enough background to give an insightful answer here.
Niki: Yeah, Neutron is definitely one of those where I think a lot of people are getting super-tripped up. I was at an internal OpenStack summit at Cisco a couple weeks ago, and I was just completely floored at all of the different perspectives on Neutron, and the different things that are integrating into the Neutron layer. Obviously, Cisco is a networking company, but I feel super-behind in terms of Neutron. It just seems like one of those things that you almost definitely need to have some really good sort of networking background and expertise to be able to sort through.
Shamail: Yeah, absolutely, and then just from my opinion, and the conversation we’ve been having for the last 45 minutes or so, it’s going to be interesting. What happens there is, when the ecosystem expands, like the kind of conversation we were having is, when we can have multiple projects in multiple domain spaces, it’s going to be interesting to see what that means for some of these projects.
Niki: Sure. I think one of the big takeaways from the internal OpenStack Summit is that if you’re looking at Amazon or something, AWS has like 3 different networking products. I think a lot of the focus is going to be on trying to simplify all of the Neutron functionality in a way that makes it easy for cloud application developers to navigate through all that stuff.
Larry’s commenting, “NSX integration with Neutron suite …” We need to probably have like 5 different shows on Neutron, it’s such a beast of a project.
Shamail: Yeah, I agree.
Niki: Let me ask you another question, because this is one that kind of came up, and that I’ve been thinking about a lot lately. Do you think that workload portability really matters? Do people care about that? Of course, in the context of having a disaster recovery plan and whatnot, but do you see people sort of asking about workload portability, and asking you to help solve that for them?
Shamail: Yeah, I think that is a real need in the market. I would say there’s probably two ways that I would look at it as well. There’s the HADR concept that you mentioned. There’s also the notion of actual data mobility, of being able to move workloads that are already there, from one place to another.
The other piece would be having common API or being able to even burst workloads, I think, is the other piece that would come into that as well. I think there’s a definite need there, and I think some of the definitions, and some of the things that we’re talking about … standardized APIs will definitely play a part in that.
I definitely see the conversation happening, and I see the conversation happening more, though, from being able to say that I can provision a workload here or there, versus I can move a workload from here to there, even though that comes up every once in a while as well.
Niki: What do you think of a VMWare strategy? It looks like they’re tipping the scales, and going full out on OpenStack. I don’t know how close you are to what they’ve been up to, but if you’re at liberty to talk about that, I’d be curious to find out more.
Shamail: Yeah, I’m not too close to it myself right now, but they had the announcement about VMware integrated OpenStack at their last trade show. They’re moving forward in that direction. Of course, they’re heavily entrenched in the Neutron area that we were talking about as well, with their NSX strategy as well.
I think we’re going to be seeing more and more from them, and I think hopefully either in the next summit, or the one after that, we’re going to see a huge amount of new information coming out from them, hopefully.
Niki: Do you think that they have kind of an advantage, based on the fact that it seems like every enterprise is pretty well-invested with VMware at this point?
Shamail: I think possibly, because if you have a… I’ll just use a [inaudible 00:43:19]. If you have a product already on the floor at a customer, I think you have an open channel to go in there and have further conversations. I think just the access to customers will be really interesting, from their perspective.
I think just from that alone, I think they did have a good shot at getting at least their message heard.
Niki: Yeah, I think so, too.
Jeff: Yeah, they’ve got such a stronghold. They’ve got such a great position in the enterprise. Frequently, I’m talking to folks that want to replace VMware with OpenStack, and I thought Boris had some great thoughts around that, where it’s kind of the wrong use case.
What are your thoughts around folks that are … Is OpenStack a replacement for VMsare, or is it an addition? Is it … Where is that place?
Shamail: I’ve seen both arguments, and I would say it’s probably, from my perspective, it’s probably in the end world right now, where I think there is a use case for both, in environments. To the point that I have seen customers that are pretty much coming in with that one use case, though; “It’s going to replace my VMware environment,” and there’s other customers coming in with the use case of, “No, my VMware environment is going to host everything I’m doing. This is for my native applications I’m developing as my next platform, in a sense.”
I think some of this actually comes down to, I don’t know if we’re going to get an answer from the community here, but at some point I think we also have to understand, is there an identity that we want to portray as a community as well? I think we hint at an identity, from a workload use case perspective, but I don’t think we’ve ever clearly come out and said that we’re either going to be this VMware replacement, or we’re not, or what our position is from that community perspective.
Niki: Yes, should ESX just be a standard hypervisor option within OpenStack, right? That’s a question we get a lot.
Niki: Larry, man, you are full of good questions today. Are there any other solutions that make deployments more user-friendly, like Mirantis Heat? Anything EMC is working on from a solutions-type perspective?
Shamail: Not at the moment. I’m sure there’s some conversations going on internally, with some of the stuff that’s happening around, maybe the reference architecture program, or even some of the stuff that the Cloudscaling team might be examining, but I don’t know offhand.
Niki: I know Rackspace has sort of a packaged installer. I know that Tristack is still very much a thing, that people are just trying to get past the whole installation process, that’s a pretty good mechanism for doing that. We had Kord Campbell on the show the other day, and he worked on some stuff and some code to be able to deploy OpenStack locally, like on your laptop, so there’s definitely options out there.
I’m not sure that the OpenStack community actually breaks down installer things.
Jeff: Yeah, Mirantis really has their … They’ve got something good going on with Fuel. They just came out with … This goes into another question, about they’re running Fuel on Docker, and what is your vision and where does Docker and containers and OpenStack all play? Where do they play together?
Shamail: I think there’s probably intersection at multiple points. There’s containerizing the cloud itself, i.e. running your controllers within containers. There’s also the notion of being able to consume containers, or provision containers, within the cloud as well, from a use case perspective.
I think long term, it probably has both as scenarios that are viable. Just with some of the history there, between Nova and then transition to Heat, and now Magnum coming and starting as well, I think the consumption path is well underway.
There’s tons of people that have done presentations on how to actually deploy a cloud using containers. The last one, I think, was from IBM actually, that I saw.
Niki: Yeah, speaking of IBM, what are your thoughts on what IBM’s doing? I have to say that both HP and IBM, from a marketing perspective, have definitely helped change my opinion about them. They don’t feel kind of like these staunchy old IT companies anymore. HP with their… I saw a billboard on the 101 that said, “Python, Penguin, and Wales; HP Helion,” and that’s it. I was like, “Brilliant.”
IBM, with their Bluemix stuff, looks like they’re doing some more to sort of attract that developer type of customer, and make their platform more applicable to what I believe are absolutely the influencers in the enterprise at the moment. What are your thoughts on those companies?
Shamail: I don’t know, I think they’re doing good. I think they’ve defined a strategy around open source and OpenStack in particular, and they’re going forward with it. As you said, they’re making huge pushes to change their identity in this space as well, or in general as well, and I think they’re both doing really good there.
Niki: It’s pretty exciting to watch. It’s definitely exciting to see, talking about EMC as going from writing drivers initially, just to make sure their stuff worked … Make sure all stuff worked with OpenStack … To now kind of getting involved in a deeper level, and starting to help shape some of the other projects based on the domain expertise.
I think when EMC’s doing that, everyone else is doing that, too, it seems like. It seems like the future for OpenStack looks pretty bright. I’m surprised it’s taken this long, but then I’m also surprised that it’s happened so fast.
Shamail: Yes, I get it. I get that statement.
Jeff: What about, is there anything you can talk about that’s kind of exciting around VNX and OpenStack? We’ve got some stuff going on with folks.
Shamail: The VNX is doing pretty good in OpenStack, from an integration perspective and a consumption perspective. That team is… I like the way they’re approaching this, because they’re actually approaching it from an OpenStack compatibility perspective, but then also using the drivers and the way OpenStack integrates, to expose some of VNX’s differentiation as well.
The example that I usually give is, for example, in a standard configuration found normally, you have clear text credentials stored in the configuration file, for your back end. For example, on the VNX side, you can actually use a certificate file, instead of storing that stuff.
Little things like that, but they’re doing good. They’re thinking about it very creatively, which is awesome.
Niki: Vancouver’s coming up. I’m assuming you’re going to be in Vancouver. Do you already have a… If you’re like me, you’re waiting until the last freaking minute, but do you have any talks planned that you want to submit?
Shamail: Yeah, I do, actually; one hopefully with you, the product manager panel. I think that will be a really good discussion just to see, A, what are the different best practices, if you will, from an OpenStack product management perspective, that multiple people are using, and how that can benefit customers as they consume and deploy OpenStack within their organizations. That will be a good one.
I’m also working on a couple other ones that are storage-centric, mainly one around Cinder, and how… Based on how they do quality control, based on their CI test model, and based on just their developer ecosystem, how Cinder ensures that regardless of which back end you’re using, whether it’s LVM, Cinder, Salt, et cetera, you’re getting at least some parity in operations and capabilities.
It’s kind of talking about Cinder and how it … I think the way we titled it is “Cinder and the Sea that Rises All the Boats,” if you will, is the way we positioned that one.
The other one is more broad from a storage perspective, which is just kind of an OpenStack storage use case discussion, so stepping out of the block, stepping on object, and just looking at storage as being an infrastructure component within your cloud, and what are the different points in OpenStack, through multiple projects, that storage has implications on, if you will?
Niki: It sounds like some really interesting talks. I’m a big fan of the use cases talks. It definitely makes it real, I think, for a lot of people. That’s what’s going on in Vancouver. What are you most excited about in OpenStack, and what’s on your OpenStack wish list?
Shamail: I think you know, from my perspective, I’m really excited about the products working group, and the product management working group. It’s our first meeting, but I’m really curious to see, by summit, what this group can agree on, and how we do our first pilot, if you will, of the program, in a sense. I’m really excited about that.
I’m also participating, as well as looking forward to, there’s some HA guide updates going on right now, so the concept there is to kind of refresh the guides to include more information about project level HA, as well as update it based on the various projects that are available right now, and technologies that are available.
Niki: Do you think OpenStack is doing a good job of solving for HA, and in many regards, kind of leaving that up to the operators to figure out? Do you think OpenStack should be prescriptive?
Shamail: I think it should be a little bit more prescriptive, but at the same time, it has to have a single voice there. How do we get that single voice? I think it’s going to come down to, how do we define that as a strategy, as a theme, and then how do we implement it cross-project?
I think it … You have to have that vehicle or process, before we can have that conversation on being more prescriptive.
Niki: Yeah, it’s crazy how differently people are handling HA right now. It’s just all over the map. I guess when you leave it up to customers, they’ll figure out a way to make it work, or vendors.
Jeff: We’re just about out of time here. We’re hitting the hour. I think one of the questions we like asking are, who are two people you’d like to see on the show, or listen to on the show?
Niki: That’s OK, if you mention someone that we’ve already talked to, we won’t call you out.
Shamail: Fair enough, thank you. I think, since we were just talking about wins to enterprise, I think [Carol Buret 00:53:43] from Intel would be a good person to have on. Some of our time here was spent talking about enterprise and enterprise adoption use cases, challenges, et cetera. That team is kind of trying to approach OpenStack from an enterprise lens and seeing, what changes are needed? What works and what doesn’t work? I think she’d be a good guest to go into that further with.
The other person I would say is probably Stefano Muffilli. He’s the OpenStack community manager and developer advocate. He’s been involved in a lot of interesting things that are going on right now, and as we were discussing, OpenStack from the government’s perspective is kind of transitioning a little bit as well, from integrated to tagging, et cetera. I think he would be a good person to offer just a lot of perspective on what this growth means, and where OpenStack is heading in ’15 and ’16.
Niki: Man, those are really good suggestions.
Jeff: Yeah, thank you. Those are good.
Niki: Speaking of women on the show, too, we’ve got Jessica Murillo, from IBM. I say Murillo, she might say Murillo, I’m not sure, but Jessica is going to be on the show. Talk about an amazing woman; I think she holds hundreds of patents, and really, really sharp, really amazing woman who’s doing a lot for the women of OpenStack movement.
Speaking of the women of OpenStack movement, a couple weeks ago, if you are noodling on what to submit for the OpenStack summit in terms of an abstract and a title and a topic, and thinking about panels or teaming up with someone, we did a Webcast a couple weeks ago that kind of covered best practices. It was Diane Mueller from RedHat, myself and Ann Gentle from RackSpace. If you guys are needing some help or some ideas, it might be something good to watch to either help you come up with an idea, or to make sure that your abstract will be as polished as possible, in preparation for the voting period.
Jeff: Where do they find that?
Niki: I believe it is on … I think it’s on the OpenStack YouTube channel, and I think they’re also posting it on the OpenStack “call for speaker” submission page as well, if it’s not there already. It might be there already.
Jeff: OK, great. What about … Where do people kind of reach you and stalk you at? Where’s the … At Twitter, is it IRC?
Shamail: Twitter’s probably the best place. It’s @ShamailXD, which I think is in the Podcast Tweet that you guys sent out as well.
Jeff: Yeah, great. Next … We’ve got Jessica coming up on the show next. The week after that, Yuri Brodsky; he’s a user from Symantec, and then Joe Arnold is going to be on, from SwiftStack. We’ve got a whole lineup of folks coming up on the show.
I think that’s it. We have the housekeeping staff, and anything else, Niki?
Niki: Really appreciate you coming on, Shamail. Really interested to see what comes out of that product working group in Pasadena, at the Metacloud, now part of Cisco headquarters. Unfortunately, couldn’t make it up there, but really I’m hoping you can talk to some of the folks there, and hopefully you can maybe draft a blog post or something, to kind of talk about what comes out of that.
Niki: Super important for the community, and a good investment of time, to make sure that OpenStack continues to meet and exceed the expectations of users all over the planet.
Shamail: Appreciate it. It was a great conversation, and again, thanks for having me on. On the products side, there probably will be a summit talk at some point, to go deeper into the conversation. Sean Roberts has a blog post for day one, and he has a live transmission going on for day two right now. If you can’t join locally, you can always join remotely. His domain is Sarob.com, and so you can follow the working group meet-up Midcycle from there as well.
Niki: I had no idea. That’s awesome, sweet.
Jeff: That is awesome.
Niki: Cool. Thanks, Shamail. Thanks, Jeff.
Jeff: Yeah, thank you, Shamail.
Niki: Hopefully next week, Jeff and I will be in better working condition.
Jeff: Better shape, yeah.
Shamail: Awesome; get well.
Jeff: We’ll see you in Vancouver.
Niki: All right.
Jeff: Thanks, everyone. Thanks for listening, and we’ll see you next week.
Niki: Bye bye.