The next BriefingsDirect Voice of the Customer IT modernization journey interview explores how a global candy maker depends on increased insight for deploying and optimizing servers and storage.
We’ll now learn how Ferrara Candy Company boosts its agility as a manufacturer by expanding the use of analysis and proactive refinement in its data center operations by bringing more intelligence to IT infrastructure.
Stay with us to hear about unlocking the potential for end-to-end process and economic efficiency with Stefan Floyhar, Senior Manager of IT Infrastructure at Ferrara Candy Co. in Oakbrook Terrace, Illinois. The discussion is moderated by Dana Gardner, Principal Analyst at Interarbor Solutions.
Here are some excerpts:
Gardner: What are the major reasons Ferrara Candy took a new approach in bringing added intelligence to your servers and storage operations?
Floyhar: The driving force behind utilizing intelligence at the infrastructure level specifically was to alleviate the firefighting operations that we were constantly undergoing with the old infrastructure.
Gardner: And what sort of issues did that entail? What was the nature of the firefighting?
Floyhar: We were constantly addressing infrastructure-related hardware failures, firmware issues, and not having visibility into true growth factors. That included not knowing what’s happening on the backend during an outage or from a problem with performance. We had a lack of visibility into true real-time performance data and fully scalable performance data.
Gardner: There’s nothing worse than being caught up in reactive firefighting mode when you’re also trying to be innovative, re-architect, and adjust to things like mergers and growth. What were some of the business pressures that you were facing even as you were trying to keep up with that old-fashioned mode of operations?
IT meets expanded candy demands
Floyhar: We have undergone a significant amount of growth in the last seven years -- going from 125 virtual machines to 452, as of this morning. Those 452 virtual machines are all application-driven and application-specific. As we continued to grow, as we continued to merge and acquire other candy companies, that growth exploded exponentially.
The merger with Ferrara Pan Candy, and Farley’s and Sathers in 2012, for example, saw an initial growth explosion. More recently, in 2017 and 2018, we were acquired by Ferrero. We also acquired Nestlé Confections USA, which has essentially doubled the business overnight. The growth is continuing at an exponential rate.
Gardner: The old mode of IT operations just couldn’t keep up with that dynamic environment?
Floyhar: That is correct, yes.
Gardner: Ferrara Candy might not roll off the tongue for many people, but I bet they have heard a lot of your major candy brands. Could you help people understand how big and global you are as a confectionery manufacturer by letting us know some of your major brands?
Floyhar: We are the producers of Now and Later, Lemonheads, Boston Baked Beans, Atomic Fireballs, Bob’s Candy Canes, and Trolli Gummies, which is one of our major brands. We also recently acquired Crunch Bar, Butterfinger, 100 Grand, Laffy Taffy, and Willy Wonka brands, among others.
We produce a little over 1 million pounds of gummies per week, and we are currently utilizing 2.5 million square feet of warehousing.
Learn More About Intelligent,
Self-Managing Flash Storage
In the Data Center and Cloud
Gardner: Wow! Some of those brands bring me way back. I mean, I was eating those when I was a kid, so those are some age-old and favorite brands.
Let’s get back to the IT that supports that volume and diversity of favorite confections. What were some of the major drivers that brought you to a higher level of automation, intelligence, and therefore being able to get on top of operations rather than trying to play catch up?
Floyhar: We have a very lean staff of engineers. That forced us to seek the next generation of product, specifically around artificial intelligence (AI) and machine learning (ML). We absolutely needed that because we’re growing at this exponential rate. We needed to take the focus off of infrastructure-related tasks and leverage technology to manage and operate the application stack and get it up to snuff. And so that was the major driving force for seeking AI [in our operations and management].
Gardner: And when you refer to AI you are not talking about helping your marketers better factor which candy to bring into a region. You are talking about intelligence inside of your IT operations, so AIOps, right?
Floyhar: Yes, absolutely. So things like Hewlett Packard Enterprise (HPE) InfoSight and some of the other providers with cloud-type operations for failure metrics and growth perspectives. We needed somebody with proven metrics. Proven technology was a huge factor in product determination.
Gardner: How about storage specifically? Was that something you targeted? It seems a lot of people need to reinvent and modernize their storage and server infrastructure in tandem and coordination.
Floyhar: Storage was actually the driving factor for us. It’s what started the whole renovation of IT within Ferrara. With our older storage, we were constantly suffering bottlenecks with administrative tasks and in not having visibility into what was going on.
During that discovery process and research, HPE InfoSight really jumped off the page at us. That level of AI, the proven track record, and being able to produce data around my work loads.
Storage drove that need for change. We looked at a lot of different storage area networks (SANs) and providers, everything from HPE Nimble to Pure, VNX, Unity, Hitachi, … insert major SAN provider here. We probably did six or so months’ worth of research working with those vendors, doing proof of concepts (POCs) and looking at different products to truly determine what was the best storage solution for Ferrara.
During that discovery process, during that research, HPE InfoSight really jumped off the page at us. That level of AI, the proven track record, being able to produce data around my actual work loads. I needed real-life examples, not a sales and marketing pitch.
By having a demo and seeing that data being given that on the fly and on request was absolutely paramount in making our decision.
Gardner: And, of course, InfoSight, was a part of Nimble Storage and Nimble became acquired by HPE. Now we are even seeing InfoSight technology being distributed and integrated across HPE’s broad infrastructure offerings. Is InfoSight something that you are happy to see extended to other areas of IT infrastructure?
Floyhar: Yes, ever since we adopted the Nimble Storage solution I have been waiting for InfoSight to be adopted elsewhere. Finally it’s been added across the ProLiant series of servers. We are an HPE ProLiant DL560 shop.
I am ultra-excited to see what that level of AI brings for predictive failures monitoring, which is essentially going to alleviate any downtime. Any time we can predict a failure, it’s obviously better than being reactive, with a retroactive approach where something fails and then we have to replace it.
Gardner: Stefan, how do you consume that proactive insight? What does InfoSight bring in terms of an operations interface? Or have you crafted a new process in your operations? How have you changed your culture to accommodate such a proactive stance? As you point out, being proactive is a fairly new way of avoiding failures and degraded performance.
Proactivity improves productivity
Floyhar: A lot of things have changed with that proactivity. First, the support model, with the automatic opening and closure of tickets with HPE support. The Nimble support is absolutely fantastic. I don’t have to wait for something reactive at 2 am, and then call HPE support. The SAN does it for me; InfoSight does it for me. It automatically opens the ticket and an engineer calls me at the beginning of my workday.
No longer are we getting interrupted with those 2, 3, 4 am emergency calls because our monitoring platform has notified us that, “Hey, a disk failed or looks like it’s going to fail.” That, in turn, has led to a complete culture change within my team. It takes us away from that firefighting, the constant, reactive methodologies of maintaining traditional three-tier infrastructure and truly into leveraging AI and the support behind it.
Learn More About Intelligent,
Self-Managing Flash Storage
In the Data Center and Cloud
We are now able to turn the corner from reactiveto proactive, including on applications redesign or re-work, or on tweaking performance improvements. We are taking that proactive approach with the applications themselves, which has rolled even further downhill to our end users and improved their productivity.
In the last six months, we have received significant praise for the applications performance, based on where it was three years ago compared with today. And, yes, part of that is because of the back-end upgrades in the infrastructure platform, but also because as we’ve been able to focus more on the applications administration tasks and truly making it a more pleasant experience for our end users -- less pain, less latency, just less issues.
Gardner: You are a big SAP shop, so that improvement extends across all of your operations, to your logistics and supply chain, for example. How does having a stronger sense of confidence in your IT operations give you benefits on business-level innovation?
Floyhar: As you mentioned, we are a large SAP shop. We run any number of SAP-insert-acronym-here systems. Being proactive on addressing some of the application issues has honestly caused less downtime for the applications. We have seen into the four- and five-9s (99.99-9 percent) uptime from an application availability perspective.
We have been able to proactively catch a number of issues, whether using HPE InfoSight or standard notifications. We have been able to proactively catch a number of issues that would have caused downtime, even as minimal as 30 minutes. But when you start talking about an operation that runs 24x7, 360 days a year, and truly depends on SAP to be the backbone, it’s the lifeblood of what we do on a business operations basis.
So 30 minutes makes all the difference on the production floor. Being able to turn that support corner has absolutely been critical in our success.
Gardner: Let’s go back to data. When it comes to having storage confidence, you can extend that confidence across your data lifecycle. It's not just storage and accommodating key mission-critical apps. You can start to modernize and gain efficiencies through backup and recovery, and to making the right cache and de-dupe decisions.
What’s it been like to extend your InfoSight-based intelligence culture into the full data lifecycle?
Sweet, simplified data backup and recovery
Floyhar: Our backup and recovery has gotten significantly less complex -- and significantly faster -- using Veeam with the storage API and Nimble snapshots. Our backup window went from about 22.5 hours a day, which was less than ideal, obviously, down to less than 30 minutes for a lot of our mission-critical systems.
We are talking about 8-10 terabytes of Microsoft Exchange data, 8-10 terabytes of SAP data -- all being backed up, full backups, in less than 60 minutes, using Veeam with the storage API. Again, it’s transformed how much time and how much effort we put into managing our backups.
Again, we have turned the corner on managing our backups on an exception-basis. So now it’s only upon failure. We have gained that much trust in the product and the back-end infrastructure.
We specifically watch for failure, and any time something comes up that's what we address as opposed to watching everything 100 percent of the time to make sure it's working.
We specifically watch for failure, and any time something comes up that’s what we address as opposed to watching everything 100 percent of the time to make sure that it’s all working. Outside of the backups, just every application has seen significant performance increases.
Gardner: Thinking about the future, a lot of organizations are experimenting more with hybrid cloud models and hybrid IT models. One of the things that holds them up from adoption is not feeling confident about having insight, clarity, and transparency across these different types of systems and architectures.
Does what HPE InfoSight and similar technologies bring to the table give you more confidence to start moving toward a hybrid model, or at least experimenting in that direction for better performance in price and economic payback?
Headed to hybrid, invested in IoT
Floyhar: Yes, absolutely, it does. We started to dabble into the cloud, and a mixed-hybrid infrastructure a few years before Nimble came into play. We now have a significantly larger cloud presence. And we were able to scale that cloud presence easily specifically because of the data. With our growth trending, all of the pieces involved with InfoSight, we were able to use that data to scale out and know what it looks like from a storage perspective on Amazon Web Services (AWS).
We started with SAP HANA out in the cloud, and now we’re utilizing some of that data on the back end. We are able to size and scale significantly better than we ever could have in the past, so it has actually opened up the door to adopting a bit more cloud architecture for our infrastructure.
Gardner: And looking to the other end from cloud, core, and data center, increasingly manufacturers like yourselves -- and in large warehouse environments like you have described -- the Internet of Things (IoT) is becoming much more in demand. You can place sensors and measure things in ways we didn’t dream of before.
Even though IoT generates massive amounts of data -- and it’s even processing at the edge – have you gained confidence to take these platform technologies in that direction, out to the edge, and hope that you can gain end-to-end insights, from edge to core?
Floyhar: The executives at our company have deemed that data is a necessity. We are a very data-driven company. Manufacturers of our size are truly benefiting from IoT and that data. For us, people say “big data” or insert-common-acronym-here. People process big data, but nobody truly understands what that term means.
Learn More About Intelligent,
Self-Managing Flash Storage
In the Data Center and Cloud
With our executives, we have gone through the entire process and said, “Hey, you know what? We have actually defined what big data means to Ferrara. We are going to utilize this data to help drive leaner manufacturing processes, to help drive higher-quality products out the door every single time to achieve an industry standard of quality that quite frankly has never been met before.”
We have very lofty goals for utilizing this data to drive the manufacturing process. We are working with a very large industrial automation company to assist us in utilizing IoT, not quite edge computing yet, but we might get there in the next couple of years. Right now we are truly adopting the IoT mentality around manufacturing.
And that is, as you mentioned, a huge amount of data. But it is also a very exciting opportunity for Ferrara. We make candy, right? We are not making cars, or tanks, or very expansive computer systems. We are not doing that level of intricacy. We are just making candy.
But to be able to leverage the machine data at almost every inch of the factory floor? If we could get that and utilize it to drive end-to-end process, efficiency, and manufacturing efficiencies? It not only helps us produce a better-quality product faster, it’s also environmentally conscious, because there will be less waste, if any waste at all.
The list of wonderful things that comes out of this goes on and on. It really is an exciting opportunity. We are trying to leverage that. The intelligent back-end storage and computer systems are ultra-imperative to us for meeting those objectives.
Gardner: Any words of advice for other organizations that are not as far ahead as you are when it comes to going to all-flash and highly intelligent storage -- and then extending that intelligence into an AIOps culture? With 20/20 hindsight, for those organizations that would like to use more AIOps, who would like to get more intelligence through something like HPE InfoSight, what advice can you give them?
Floyhar: First things first -- use it. For even small organizations, all the way up to the largest of organizations, it may almost seem like, “Well, what is that data really going to be used for?” I promise, if you use it, it is greatly beneficial to your IT operations.
Historically we would constantly be fighting infrastructure-related issues -- outages, performance bottlenecks, and so on. With the AI behind HPE InfoSight, the AI makes all the difference. You don't have to fight that fight when it becomes a problem because you nip it in the bud.
If you don't have it -- get it. It’s very important. This is the future of technology. Using AI to predictively analyze all of the data -- not just from your environment -- but being able to take a conglomerate view of customer data and keep it together and use predictive analytics – that truly does allow IT organizations to turn the corner from reactive to proactive.
Historically we would constantly be fighting infrastructure-related issues -- outages, performance bottlenecks, and so on. With the AI behind HPE InfoSight, and other providers, including cloud platforms, the AI makes all the difference. You don’t have to fight that fight when it becomes a problem because you get to nip it in the bud.