Cloud Computing, Big Data and mobility arguably share similar intrinsic features such as distribution, and geographical dispersion. How can you utilize these common intrinsic features to invite Big Data and mobility into your organization, leveraging an existing cloud infrastructure, at a low risk, low cost and with high resilience to meet changing business demand?
- Wayne Shurts, Executive VP & Chief Technology Officer, Sysco
Given the similarities between cloud, Big Data and mobility, a new question is being raised as to how an organization’s existing cloud infrastructure can be leveraged in order to invite Big Data and mobility into the organization.
Although organizations have been able to scale up and adapt to introduce new applications and capabilities thanks to the new possibilities provided in cloud vendor options, there are naturally network design, data transport and infrastructure challenges to take into consideration beforehand.
Likewise, Big Data specifically has technical and non-technical pre-requisites that are important for a CTO to keep in mind. Does your organization have the right skill set, the right data scientists and the right readiness to leverage Big Data, and how will it be used to inform business decisions? A cloud provider will be able to help organizations with the technical requirements, but the additional questions are for the leaders themselves to determine.
“The purpose of the cloud it to provide enhanced capabilities and elasticity that you can’t do easily in house. And big data and mobility are simply two use cases that fit that mold.” However, it’s crucial to ensure you’re maintaining critical data and differentiating between what are core, transactional functions and others that can be operated elsewhere.
The best advice is to work with a partner, don’t just learn as you go, experience some of the pitfalls and horror stories to truly understand it, and avoid analysis paralysis in pursuit of perfection on the cloud by trying some of the applications on the cloud in order to obtain the necessary experience.
Sanjog Aul: Welcome listeners this is Sanjog Aul, your host, and the topic for today’s conversation is “Leveraging Cloud to Invite Big Data and Mobility,” and I have with me Wayne Shurts. Wayne is the Executive Vice President and Chief Technology Officer for Sysco. Hello Wayne, thank you for joining us.
Now, Big Data and Mobility have proven to be of great value to many and more and more it is making sense for companies to embrace them, but we wanted to ask a question that hasn’t come up as often, which is, can we do so while leveraging our existing Cloud infrastructure? That’s easier said than done, and that’s why we have you here today to discuss what the state of our Cloud infrastructure should be in order to accommodate it. So with that said, what do you think is the current Cloud infrastructure, where does it stand, and to what degree is it ready to take on Big Data and Mobility with minimal changes?
Wayne Shurts: First of all, I’ll definitely agree that it’s easier than done Sanjog, and you need the architect solutions differently to take advantage of Cloud capabilities like elasticity. It’s a new skill that we’ve had to learn and had to invest in to take advantage of that. Often times you’ll opt out of Cloud capabilities due to security, control, other reasons that sub optimize the solution, so you might keep some of those things in house. For example, if you simply want to lift and shift your environment that was designed five to 10 years ago with an end-tier architecture that is data bound, most traditional relational database systems can scale up but not out by adding more commodity hardware needs. There are definitely risks and challenges; Cloud providers will absolutely tell you about the wonderful possibilities, which are true in a vacuum, however they don’t know your infrastructure, they don’t know you network design, and they don’t know about your data transport capabilities, and those things in particular are some of the challenges that we ran into early on.
“Cloud providers will absolutely tell you about the wonderful possibilities, which are true in a vacuum, however they don’t know your infrastructure, they don’t know you network design, and they don’t know about your data transport capabilities, and those things in particular are some of the challenges that we ran into early on.”
For example, we made a decision to migrate to Salesforce.com. When we did that, we had to rethink the way we transport application data. Historically we didn’t rely much on middleware or web services and most things were point to point integrations. Or when we migrated to Office 365 for email, we ran into network issues because the network had a single point of failure to the Internet. We just didn’t think of the network as being that critical for things that we were doing in the past, and now with the Cloud they are incredibly important. So the Cloud infrastructure has come a long way in the last several years, but the key thing to consider is your applications, your network and how you set them up for success in the Cloud when you go there. It just doesn’t happen by getting the vendor and turning the key.
Sanjog: If an organization believes they have a reasonably mature and stable Cloud infrastructure and they can slap Big Data and Mobility onto it easily, can they do so with reasonable assurance that it will work out of the box? Or will they have to make basic tweaks first to invite these new paradigms?
Wayne: I think you can get Big Data and Mobility and you can start taking advantage of them on the Cloud; there is no doubt about it. There are a couple of things though that are prerequisites that you will need to look at.
“The Cloud infrastructure has come a long way in the last several years, but the key thing to consider is your applications, your network and how you set them up for success in the Cloud when you go there. It just doesn’t happen by getting the vendor and turning the key.”
Your network capacity suddenly the network is incredibly important as you do these Cloud solutions. Can it handle the volume and the load and the resiliency that you need to make sure that these applications run, perform well and are stable and scalable? The big thing to look at is your transactional performance. Your mobile apps are typically going to be transactional in nature. But some will just be content oriented, and as a result you’ll likely to need web services and APIs for those transactions.
So it’s definitely doable, and it’s important that you pay attention to those parts of the equation and what’s going to be different in the way that a Cloud application performs and Big Data and Mobility performs over the internet versus through an internal data center.
Sanjog: What other than network bandwidth are some specific pre-requisites for a successful deployment of Big Data on cloud would like to point out? Are these things you will typically find in a cloud infrastructure, or will some tweaking be required?
Wayne: There will be some tweaking, I think you’ll find a fairly open environment out there to do Big Data in the Cloud. We’ve had success using some companies like 1010data and Pure Predictive. But there are things like network capacity, and you’ve got to make decisions on whether you are going to transport raw data, which we would recommend, versus transformed data or exactly how you’ll transport that data interacting with Cloud APIs or web services or batch or whether you are going to use static data model versus dynamic data model with version control. You need to decide whether the Cloud’s going to be the final destination for the data you’ve planned or do you plan to bring that data back internally again. It gets more complicated if you are going to bring it back internally.
“Most of the good Cloud providers that you talk to can help you with the technical prerequisite. I think you’ll find that they are fairly mature, but you on your own are going to have to figure out those non technical prerequisites and where your organization is.”
These are just some of the more technology related prerequisites, but to me, there are some non-technical pre requisites that are really important. For example you need to have everything well thought out before you do it. Who will have access to the data to perform analysis in your organization? Do we have the right skill sets to analyze Big Data? Statistical data plus data mining, plus some light programming skills plus problem solving plus story telling all equal data scientists. And do we have those skills to take advantage of it? How will the data and analytics from the data be used to inform business decisions? Are you doing the Big Data for operational reporting or ad hoc analysis? You need to think through how that would work and ask, “Is our organization ready to leverage the data? So I think most of the good Cloud providers that you talk to can help you with the technical prerequisite. I think you’ll find that they are fairly mature, but you on your own are going to have to figure out those non technical prerequisites and where your organization is.
Sanjog: You’re making a bold statement in saying that most of the providers are fairly mature. But it’s one thing to say you’re mature enough to provide a Cloud infrastructure and another thing to say they have a complete understanding of when we want to adopt Cloud. When it comes to Cloud as the foundation of Big Data and Mobility, are we mature or are we just learning to crawl rather than run?
Wayne: I think we are beyond crawling but I think we are not running yet for sure. A key that we found is to really define what you are trying to do business wise, don’t so much listen to the sales brochure that you get from the Cloud provider but think through exactly what you are trying to do Mobility and Big Data wise and see if what they are providing is a fit for you. In most cases I think it’s going to be a fit, and you would be more in the run versus crawl area, but depending on what you are going to be doing with it and the more complex that it gets, you are going to be more in the crawl area. It’s looking at each situation; I don’t think there is a hard rule. Sanjog. I think you have to look at exactly what you are trying to do with the data and scrub that against what your provider is offering.
“You have to look at exactly what you are trying to do with the data and scrub that against what your provider is offering.”
Sanjog: What are the differences between what may be required from a cloud infrastructure standpoint to deploy Big Data versus deploying Mobility? What are the commonalities that can be leveraged?
Wayne: The big difference is transactional performance. Your mobile apps will typically be transactional in nature, like ordering, sales, inventory management. Some will just be content oriented, product information and training for example. And as result, you’ll likely need web services or API for those transactions. For us, the Big Data components are rarely transactional. The design of the data feeds are distinctly different for us than transactional APIs, and the data miles are common.
Sanjog: Could our effort to make this cloud infrastructure bear the burden of Big Data and Mobility actually jeopardize the core function the cloud was originally deployed for? To what degree is it overkill to make the environment more complex before you need a separate cloud for another line of business or need another infrastructure?
Wayne: I don’t really see us bringing the cloud to its knees. You asked earlier though if we have some core transactional things happening on the cloud, do you want to separate them on a different cloud for things like Big Data, etc., and I think that’s a very good question. I don’t think there’s a pat answer for that. I think each organization needs to look at its transactional volume, its network capability the cloud that it’s using. Certainly in our case here at Sysco, we’d probably have some dedicated pipes for core transactional things that we were doing. So if we were doing Big Data and mobility in other places, they’re not interfering with it, but I think every organization has to look at that.
“The purpose of the cloud it to provide enhanced capabilities and elasticity that you can’t do easily in house. And big data and mobility are simply two use cases that fit that mold.”
The purpose of the cloud it to provide enhanced capabilities and elasticity that you can’t do easily in house. And big data and mobility are simply two use cases that fit that mold. The biggest risks are certainly the network capability, connectivity and capacity that you have is obviously going to be very important and a key failure point for your cloud applications. Implementing another is implementing solid design patterns and principles for the cloud solutions that you have and adopting those principles and making sure you use them across all your apps, as well as maintaining critical data and mobile application knowledge in house.
Lots of times you’re using third parties to do your mobility work or to do your big data work or to do some of your cloud app work, which is very common because it’s kind of a new space, but I think it’s also very important to make sure that you’re maintaining some of that critical data and that mobile application knowledge in house as you use those third parties to do it.
Sanjog: Is everyone cognizant of the pre-requisites you’ve mentioned, or are there some pitfalls they should look out for upon trying this for the first time?
Wayne: I think everybody understands it conceptually how important the network is in the cloud application. But until you get into it and you get burnt once or twice, you realize there’s a whole lot more bandwidth that we’re pressing through this, there’s a whole lot more data that needs to pass through our network that wasn’t before or there’s so much critical data that we need a very redundant network so that if it goes down, I don’t shut my business down or shut this portion of it down. Until you live some of those horror stories, you don’t fully get it.
“I don’t think we live in a world at all where the cloud is the right answer for every application that you have. I do think that there are specific characteristics in an application in a solution that you’re trying to put up that the cloud can handle well.”
So doing some good network analysis and network studies with some network partners and the application partners prior to going into cloud or putting a big app up I think is really important. Conceptually it’s self-evident. But I don’t think you really understand where your flaws are until you actually try to do it and fail in a couple of places. That has been our experience at Sysco, including something as simple as email on the cloud and Office 365.
Sanjog: Many of the choke points you mention refer primarily to scalability. Is that the main area for pitfalls, or do complexity and extensibility also produce some horror stories?
Wayne: All of those things Sanjog. There are certainly the scalability issues around the network, but then also there are some applications or some use cases that the cloud is a very good solution for. There are other use cases that you would look at and say, the cloud is not good for that, and you should keep that in house. And we found there’s a lot of partners out there that can help you do those assessments, but I don’t think we live in a world at all where the cloud is the right answer for every application that you have. I do think that there are specific characteristics in an application in a solution that you’re trying to put up that the cloud can handle well, and therefore you should do it and avoid the others. I think that’s a good analysis, but the answer to everything is not the cloud.
“Don’t learn as you go and get a partner, and avoid analysis paralysis. Make a good bet, take a good risk, dip your toe in and start the work around getting something up on the cloud.”
Sanjog: Now of course the end game we all have is to make big data and mobility get adopted on a common cloud foundation, and if possible, one cloud foundation. And we want to do that with the lowest amount risk, lowest cost and a very high resilience. So no matter what happens, the business continuity is not jeopardized. What is the learning people have discovered that our listeners can take away from this conversation?
Wayne: What we’ve learned here at Sysco and from colleagues that I’ve talked about is I think it’s important in this space that you don’t just learn as you go. I think it’s real important that you work with a partner who has experience in the cloud who has been there and done that and can help shorten your learning curve and help you avoid some common pitfalls that others make. We’ve certainly learned that learning as you go is important, but not a good strategy when it comes to cloud; get connected with some experts who can help guide you and help save you some of
And at the same time, I think it’s important that you avoid analysis paralysis in the pursuit of perfection when you go to the cloud. Just start getting going on some of those applications where you can learn and can gain some experience. Take your organization up the cloud learning curve.
Those are the two things: don’t learn as you go and get a partner, but at the same time, avoid analysis paralysis. Make a good bet, take a good risk, dip your toe in and start the work around getting something up on the cloud. The learning will be there and you will grow.