Operating a business with outdated or end-of-life, unsupported operating system (OS) and software technologies can pose myriad vulnerabilities. Fortunately, this nightmare can be remedied quickly and cost-effectively by moving IT infrastructure into the cloud.
Swyft provides a software and data analytics platform that gives brands and retailers everything they need to build and deploy automated retail stores. With Swyft, customers such as Apple, Best Buy, and CVS are able to provide consumer goods 24/7 through next-generation vending machines, even in satellite locations where they don’t have stores.
“Swyft creates sensor-based technologies that are cost-effective, simple to operate, and deliver a great consumer experience,” says Gower Smith, CEO of Swyft. “We always source fantastic, high-traffic, high-profile locations for our brand and retail partners. These location verticals include airports, hotels, mass transit, and tourist locations.”
Unlike traditional mechanical vending machines, Swyft’s turnkey solutions operate on cloud-based software and use cutting-edge robotic and sensor-based hardware. Self-service retail installations include automated kiosks, lockers, and “grab & go” cabinets, all supporting QR-code-based contactless payments. Swyft also offers services ranging from location procurement and customizable store buildout development to replenishment and reporting.
When Swyft was founded 15 years ago, they used Microsoft-based technology for their infrastructure and ran three racks of servers in a data center. “Everything was on-prem, including our Active Directory server, databases, and application servers,” says Vikranth Katpally, Chief Technology Officer of Swyft.
Three years ago, when Katpally joined the company, he found that much of the hardware was outdated, the server software near end-of-life, and extended support frequently unavailable. Two full-time engineers were dedicated just to keeping Swyft’s IT operations functioning. “Each Monday, we would run various jobs, and they would tell me about that weekend’s failures. Every time I saw that email from the engineers, my heart skipped a beat because there were a couple times when we lost our main database. Luckily, we always had a backup we could plug in,” says Katpally. “Still, we knew our on-prem days were numbered; we needed to move to the cloud.”
Swyft’s IT team tried a lift and shift to migrate their virtual machines (VMs) over to the cloud. However, this process was expensive and didn’t provide any new technical capabilities or leveraged innovations. Additionally, some VMs couldn’t be migrated due to end-of-life OS support. This project was sidelined when the team couldn’t clearly cut over to the cloud and abandon the data center without threatening its Payment Card Industry (PCI) compliance–one of a retailer’s most important requirements for ongoing operations.
“We wanted to do things at our own pace, but at the same time have experts available to guide us. That’s why we chose SADA.”– Vikranth Katpally | Chief Technology Officer at Swyft
SADA, a Google Cloud Premier Partner and three-time Google Cloud Partner of the Year, worked with Google Cloud to design a hybrid migration path. Together, the team configured a VPN tunnel between the data center and Google Cloud Platform (GCP) so Swyft could migrate its entire VMware environment to Google Cloud VMware Engine. This allowed an easy lift and shift of VMware-based applications to GCP without any major changes to apps, tools, or processes. “That gave me a lot of confidence when we were able to migrate all the VMs that we didn’t want to lose from the on-premises environment in less than a weekend,” says Katpally.
Next, SADA assisted with a complete infrastructure migration to Google Cloud’s PCI-compliant cloud native environment. The phased deployment started with the Active Directory migration, then Quality Assurance (QA) and Development servers, and ended with the production environment migration.
“We had about 1,200 stores operating in the world at the time we conducted the migration,” says Katpally. “The stores were always online and conducting sales; we were able to sync all the transaction data from the stores into our backend server and release pending software updates for these stores without any issues.”
To begin their move into more predictive inventory management, Swyft replaced a custom tool with a machine learning model on Google Cloud Platform. “We had a homegrown supply-chain algorithm that would allocate stock to our stores based on past sales and shipping criteria, among other factors,” says Katpally. “We put those criteria into Google AutoML and ran some experiments over a few weeks to determine how well it forecast demand. Now, we use that in some stores to allocate products.”
As a result of Swyft’s engagement with SADA and Google Cloud, they were able to move from 12 VMs to just three instances in one Active Directory (AD) group. “That’s one of the biggest changes; we could move multiple VMs for primary and secondary AD backups to a single cloud AD.”
Also, Swyft can now restore its production database in minutes running an automatic script whereas it used to take two engineers a full day to restore it manually in a data center. “Because our production and backup servers are both on GCP, we can just run the script in a few minutes, as opposed to 24 hours of on-prem work,” says Katpally. “We’re down a couple IT people, but we’re able to make do with one IT intern because we don’t have as much to do to maintain our infrastructure.”
In terms of computing upgrade cycles, moving from a Microsoft-based on-prem environment to GCP has fast-forwarded Swyft’s tech stack by over a decade. “We were able to leapfrog generations of technology,” says Katpally. “We moved away from running native Windows with VMs on bare metal to a Docker-based container running inside a GCP instance. That equals huge gains for our application development and deployment.”
Overall, SADA helped Swyft to:
- Migrate 15 TB of database and server data from on-prem to GCP
- Reduce 12 VMs to three instances in one cloud Active Directory
- Gain real-time visibility into store operations
- Reduce infrastructure staff from two engineers to one intern
- Implement machine learning to predict and allocate stock in stores
SADA continues to provide Swyft with ongoing technical support for design reviews, security vulnerability audits, and feasibility analysis.
“It is cost-prohibitive for me to have a large staff that is purpose built for specific areas. Having SADA gives us that ability on demand and has proven to be a huge advantage.”– Vikranth Katpally | Chief Technology Officer at Swyft