Cloud Migrations in Retail: Managing the Risk to Ensure Customer Satisfaction
Given the extremely crowded e-commerce industry, where the barrier to entry is as low as simply having access to an internet connection and a set of web pages, retailers’ technology investments must drive competitive advantage. At the same time, whether it’s inventory systems running on SAP or web-based shopping carts, retailers need to do everything possible to avoid an unplanned outage or significant slowdown for their critical applications. This means that the underlying IT infrastructure supporting these business-critical apps plays arguably the most important role in the entire customer experience.
The problem is, while performance issues and outages can be caused by an unanticipated increase in traffic, many are caused by retailers’ underlying legacy infrastructure and unexpected congestion due to rapidly changing workload patterns on a shared, virtualized infrastructure. Look no further than last year’s Amazon Prime Day debacle, caused by issues with Amazon’s back-office servers. And if Amazon can fall victim to outages, then so can every other retailer on the planet.
Meanwhile, the cloud is quickly becoming a key differentiator in all aspects of retail from operations, merchandising and marketing to the supply chain, sales, service and support. The potential benefits of the cloud – including reduced costs, greater scalability, and increased flexibility – has encouraged many retailers to migrate their workloads from on-premises to the public cloud. The adoption numbers support this: according to Gartner, the public cloud market is growing at a healthy compound annual growth rate of 16.5%.
However, retailers are increasingly realizing that cloud migrations can be risky endeavors. Many are finding that, after employing a cloud first strategy, and moving an application and its accompanying workloads from on-premises to the cloud, they then need to move the app back to on-premises – otherwise known as “reverse migration” – due to a variety of factors, including security, cost, performance, reliability or manageability. And a trial-and-error method for cloud migrations is also too risky for the following reasons:
- Workloads could be migrated to the cloud without a detailed understanding of performance characterization, leading to unacceptable performance or over-provisioned cloud infrastructure – which could reduce or eliminate the expected cost savings
- No empirical evidence (citing success or failure) can become available until a workload has been migrated and tested in production in the cloud
When it comes to cloud migrations, retailers need guidance – and it starts with asking questions and understanding your workload profiles.
In order to avoid the significant costs – namely, money, time and resources – of a failed cloud migration, retailers must understand their application workload profiles and infrastructure performance requirements beforehand. Without that initial understanding, there is no way to accurately determine if a workload or application is a good candidate for migration to the cloud. And by conducting this pre-migration analysis, retailers can:
- Identify and incorporate workload dependencies into their cloud migration plan
- Create and analyze workload profiles that represent their production retail applications
- Understand seasonality of workloads so that cloud configuration can be architected to scale up/down as required
- Turn the workload profiles into highly realistic simulations of real-world retail application workloads
- Run those simulated workloads in any of the public cloud providers, or even on other on-premises infrastructure, and use the resulting information to identify the suitability of any application or workload for the cloud
- Identify the ideal workload cloud configuration and optimize costs while ensuring acceptable performance
Along each step of the migration process, retailers should constantly remind themselves that delivering a stellar customer experience is the number one goal, and cloud migration projects are a means to that end. Indeed, the customer retention stakes are extremely high: for its 2018 Future of CX Report, PwC surveyed 15,000 consumers and found that 1 in 3 customers will leave a brand they love after just one negative experience.
In order to avoid this doomsday scenario, retailers must take action now on their cloud migrations. In addition to taking the above steps to determine which workloads and applications are suitable for migration, retailers should understand how available technology can help de-risk the process – through automation of application infrastructure discovery, workload profiling and cloud performance load testing, supported by sophisticated analytics. The cloud could very well be the next step in a retailer’s journey to continued growth – but as with any worthwhile journey, preparation is required.
-Rick Haggart, SVP of Professional Services, Virtual Instruments