It's easy to understand why so many retail organizations have moved their infrastructure to the cloud. Cloud-based architecture allows provisioning nearly infinite compute instances on-demand in minutes. There is no need for upfront licensing or infrastructure costs, meaning a faster return on investment can usually be realized.
The most lauded benefits of all, however, are the unmatched agility and scalability of the cloud, given deployments can be fully automated using application programming interfaces, or APIs. This enables retailers to deploy new services or capabilities at equal pace with market demand and quickly expand or shrink their operations.
Overall, cloud technology can provide an invaluable competitive edge and allow retailers to react quickly to ever-changing consumer needs.
Hasty Cloud Implementation Leads to Chronic Challenges
The numerous benefits made possible by cloud technology are real. However, many retail organizations follow a herd mentality because they are afraid of being left behind. They reason that many of their peers and competitors are leveraging cloud-based systems and tapping into their seemingly infinite benefits, so they should as well.
However, the problem with the cloud attaining a cult-like status is that many retailers implement the technology too eagerly. Without thinking through their unique needs and goals, retail organizations quickly experience chronic challenges that compound over time.
For instance, retailers that adopt a single-source public cloud may initially experience the cost, agility and ecosystem benefits such technology can bring. Over time, though, complexities will inevitably arise that not only negate the initial benefits of the public cloud but can also impede business growth, anger customers and damage brand reputation.
Following are four common challenges that many retail organizations face when embracing cloud technology too hastily:
1. Rising Costs: In some cases, cloud services are initially free, and the affordability of the technology can serve as a true lifeline. This is particularly true for startups. However, as a business grows, it's going to need to perpetually spin up more server instances.If long-term growth and scalability aren't considered and accounted for, cloud bills can quickly multiple to five or 10 times their initial cost. Complicating matters further, many retailers mistakenly assume the solution to rising cloud costs is simply moving to another or an additional cloud provider, but in doing so, service fees can become even higher and more difficult to effectively manage.
2. Service Incompatibility: The open source landscape has opened up opportunities for retailers to improve customer experience through various frameworks and applications. Some cloud providers may provide their own managed versions of these. However, they are upgraded less frequently compared to the open source releases, leading to lack of availability of the latest features.
The equivalent proprietary services offered by the cloud vendor are often less agile when it comes to adding features to the service. For example, a retailer that wants to take advantage of microservices technology to improve its customer experience is going to require a high-performance, scalable database capable of integrating with a variety of operating system applications. Open source databases often have a leg up compared to the cloud-vendor databases.
3. Vendor Lock-In: Being beholden to one cloud vendor can lead to both inability to leverage groundbreaking technology from other cloud vendors and uncompetitive pricing mechanisms that don't scale along with business growth.
Even more concerning, vendor lock-in can lead to serious data portability problems, and retailers must always be prepared and able to move their data to another cloud vendor.
Consider the recent example of Walmart demanding its technology vendors stop running its applications (which hold sensitive Walmart data) on AWS, given the competition between Amazon and Walmart.
A lack of data portability has the potential to be a major sales roadblock for any organization in the retail industry, especially if any of their customers could be considered Amazon competitors.
4. Operational Complexity: Managing cloud environments isn't for the faint of heart, especially as a business and its infrastructure scales. For instance, any zone or regional failures in public cloud environments can bring down data access, leaving IT teams scrambling in an attempt to recover their instance and assets as quickly as possible.
Compliance headaches can also occur when data isn't in multi-region deployment, and traffic spikes can quickly drive up costs and lead to over-provisioning or the need for operations teams to manually monitor the cluster.
Best Practices for Cloud-Based Retail Systems
For retailers, ensuring that the cloud architecture has low latency and high availability is paramount. Customers don't have a lot of patience and expect responsive online storefronts.
Amazon famously did a calculation showing how a page slowdown of just one second could cost it $1.6 billion. This means that retail organizations must be able to support large traffic spikes while retaining low read latencies during certain times of the week, month or year.
If an e-commerce platform isn't able to manage incoming traffic due to a poorly designed, implemented or managed cloud infrastructure, the site will slow down, inventory and product review lookups will come to a crawl, and the checkout process will bottleneck. As a result, the customer experience suffers, dollars are lost, and the brand's reputation gets dinged.
To prevent common cloud challenges from impacting a business's bottom line, retailers should adhere to the following four best practices:
1. Ensure high availability. The unfortunate reality is that in the cloud, failures are inevitable. When they happen, retailers need to be able to recover quickly and mitigate the severity of the failure.For business-critical applications in particular, be sure to implement sound geo-distributed deployments architectures. Tolerate node and zone failures automatically with multi-zone and fault-tolerant deployments.
Replicate data across regions to ensure against regional outages. This could involve using additional, different regions or even cloud providers like Google Cloud or Microsoft Azure to mitigate any business impact resulting from AWS failures, for instance.
Be sure to perform region outage fire-drills to ensure well-exercised contingency plans for diverting traffic to other regions when an outage occurs.
2. Ensure low latency. User experience is affected by latency, and it is important to serve requests with a low latency. Retailers must ensure low query latency by keeping data close to the users, which requires geo-replicated deployments with data present in multiple regions.
It is also important to understand the impact of outages and the necessary recovery time. Outages typically cause a temporary increase in latency because user requests have to be served by a datacenter that is farther away.
3. Ensure horizontal scalability. For many retailers, there is much, much more online traffic in the last few months of the year (typically October through December) than the rest of the year.
Spikes in traffic also can occur as a result of various actions, such as a well-placed advertisement. The impact of a traffic spike can be a dramatic increase in the number of simultaneous connections, number of reads, and the number of writes. It is important to ensure that all of them are scalable in the application.
4. Comply with data regulations. With data governance and privacy laws like GDPR and the California Consumer Privacy Act becoming more prevalent and stringent, ensuring cloud-based data remains within compliance is essential.
Retailers should restrict their data distribution to specific geographies to best comply with data protection laws, as regulations like GDPR mandate that data be stored in particular regions (not leaving the region's boundaries) and deleted whenever the data subject exercises their right to be forgotten.
Multicloud, Globally Distributed Data Architectures
Retailers' unbridled enthusiasm for the cloud is admirable, and it's also understandable how overwhelming effective cloud management can be. However, the key to successfully reaping the benefits of cloud technology is simple: Don't put all your eggs in one cloud basket.
To support ever-changing infrastructure demands as a business scales and customer demands evolve, prioritize globally distributed data architectures rather than being beholden to one cloud vendor and all the limitations it imposes.
Once used only by technology giants like Facebook and Google, globally distributed data architectures can be applied to any business-critical enterprise applications.
It is also essential to pick the right geo-distribution strategy, depending on the exact microservice -- such as retail product catalogs, checkout processes or logistics systems.
By adopting global data distribution, while also establishing contingency plans that ensure low latency, horizontal scalability and compliance with data governance regulations, retailers can confidently defend their most critical business assets against infrastructure failures and realistically take advantage of all the benefits cloud technology has to offer. 
Karthik Ranganathan is CTO of
YugaByte.
OhNoCrypto
via https://www.aiupnow.com
Karthik Ranganathan, Khareem Sudlow
