The company, which is not named in the case study, is a global leader in hospitality and entertainment with 65,000 employees and $5 billion in annual revenue. It places a strong focus on customer loyalty and operational efficiency, and set out on a digital transformation program to streamline its marketing processes, increase agility and speed to market, and create curated online experiences based on customers' hospitality and entertainment preferences. Its data analytics ran on a two-petabyte on-premises Apache Hadoop platform spread across 246 servers in 14 racks.

The company wanted better insight into customer preferences through an improved data analytics system in the cloud, so its marketing team could gain new capabilities in a shorter time frame and the business could gain a competitive advantage within the hospitality and entertainment industry. To get there, it chose to migrate its existing 2PB on-premises Apache Hadoop platform to Microsoft Azure.

Cost was a second driver. The Hadoop environment occupied 246 servers across 14 racks, and decommissioning it would return costly data center space and avoid spending on additional infrastructure and an expanded footprint.

Connectivity was the constraint. VPN throughput into Microsoft Azure is limited to 1.25Gbps, which meant moving the full 2PB would take more than 370 days. That would extend project estimates, hurt financial forecasts and delay other parts of the digital transformation initiative. A second challenge surfaced mid-project: a separate entity within the organization was copying incremental data from Hadoop to Google Cloud Platform for marketing analytics, and once the platform lived in Azure, the internet egress costs for that transfer exceeded the department's budget.

The company used Megaport's high-performance Software Defined Network (SDN) to complete the Hadoop migration. Its networking team established two 10Gb Megaport Ports with Virtual Cross Connects (VXCs) to Azure, providing fast, flexible and direct access to the cloud. Over that dedicated connectivity the entire 2PB of data was transferred to Azure within 26 days, a 93% decrease in the time frame a VPN migration would typically require.

With the data in Azure, the company decommissioned the 14 racks of on-premises equipment as planned and began building new features for its data analytics platform in the cloud.

In the second phase, the company addressed the Google Cloud egress problem. Provisioning two additional 10Gb Ports through its self-serve Megaport account, the networking team set up private, high-speed, dedicated connectivity to Google Cloud Platform. The new VXC provided multicloud connectivity between Azure and Google, reducing costs and increasing efficiency as data moved between cloud environments.

The two-phase project was completed on time and under budget. The company's next phase is to migrate its 1.4PB enterprise data warehouse from its on-premises data center into Azure, again using Megaport-enabled cloud connectivity so the move does not disrupt existing operations, and to allow more elastic expansion of warehouse services without investing in additional infrastructure.

• 2PB Apache Hadoop platform migrated to Microsoft Azure in 26 days
• 93% reduction in migration time compared with a typical full migration over VPN
• 246 servers across 14 racks decommissioned, returning costly data center space
• Two-phase digital transformation project completed on time and under budget
• Private multicloud connectivity between Azure and Google Cloud Platform, cutting egress costs
• Path cleared to migrate a 1.4PB enterprise data warehouse to Azure without disrupting operations