Maintaining 100% Availability for more than 20 Years for one of the French National Railway’s most Critical Applications

Overview

Challenge

• Ensure high performance and total availability for the ticketing system serving the entire French mainline rail network, with seamless disaster recovery.

Solution

• Deploy Hitachi Virtual Storage Platform arrays with Hitachi Copy Manager, ShadowImage and Hitachi Universal Replicator, enabling point-in-time snapshots.

Outcome

• Ensured 100% availability of the organization’s commercial distribution system for more than 20 years and reduced recovery time from six hours to less than one.

Challenge

As France’s national railway company, SNCF plays a vital economic and social role, safely delivering more than 5M business and leisure travelers to their destinations every day. To help customers plan, book and pay for mainline rail journeys across France – and neighboring countries through its partnerships with Eurostar, Thalys and SBB (Swiss Federal Railways) – SNCF ran a system called RESARAIL for more than 30 years.

To serve customers effectively, RESARAIL needs to handle approximately ten ticket sales per second, doubling to 20 at peak times. The system must also respond near-instantaneously to customer searches on pricing and travel itineraries, with a load of up to 10,000 transactions per second.

According to Alain Mortier, IT Manager, manager of the RESARAIL system at SNCF, “RESARAIL was one of the most critical applications at SNCF and the most visible to our customers, who expected extremely short response times. Of course, availability was also paramount; even a minute of downtime would have represented an enormous loss of revenue.”

Beyond deploying the most reliable and highly available hardware possible, SNCF wanted the ability to take reliable, non-disruptive point-in-time snapshots of the system. The goal was to enable the rapid and reliable (in terms of data consistency) recovery of services in the event of unplanned downtime.

Bertrand Delaporte, TPF System Team Leader at SNCF, said: “Our real-time operating system cannot simply slow down; you cannot interrupt the I/O operations to the storage system. If those operations stop for even a second or two, the system will crash. We needed to achieve asynchronous and consistent replication of this mission-critical system to the remote site without any impact on the production landscape.”

Solution

SNCF originally selected IBM storage to support the RESARAIL system, and later switched to EMC. Subsequently, the company issued a tender for a more highly available solution.

Alain Mortier said, “Following an RFP, Hitachi was chosen because it was technically superior to the equivalent offerings from IBM and EMC. Hitachi was the only vendor able to guarantee 100% data availability.”

SNCF continued to choose Hitachi Vantara storage through several upgrades and refreshes of the RESARAIL landscape. As energy consumption became a more important issue in an increasingly crowded data center, the Hitachi solution extended its lead over the competitors’ offerings by virtue of its superior efficiency.

“Over the years, for every RFP, we provided specific expectations around response times,” said Bertrand Delaporte. “We ran extremely demanding tests on the Hitachi hardware and the performance was always above what we’d asked for – and even beyond what Hitachi had promised.”

To solve the challenge of enabling rapid recovery without impacting production performance, SNCF used Hitachi ShadowImage to create snapshots of the system and Hitachi Universal Replicator to synchronize them to the other arrays – initially in the same data center, and later at a remote site.

Alain Mortier commented, “Hitachi’s solution was uniquely able to capture point-in-time snapshots while guaranteeing data consistency and, crucially, without interrupting or freezing storage I/O operations. The other solutions generated a significant risk for the system when snapshots were taken, which was not acceptable.”

After 20 years of evolution, the final storage configuration supporting RESARAIL was based on Hitachi Virtual Storage Platform (VSP) 5100 arrays in two data centers 25 km apart. Hitachi Copy Manager for IBM TPF was used to manage the disaster recovery (DR) solution, enabled by ShadowImage and Hitachi Universal Replicator.

Outcome

“We never experienced a single problem in terms of performance or reliability with Hitachi storage,” says Alain Mortier. Bertrand Delaporte confirmed, saying “We never questioned our original decision to choose Hitachi. The solution ran perfectly through multiple generations across more than two decades.”

Even if a failure had occurred, the Hitachi snapshot-based DR solution gave SNCF the ability to recover reliably in less than one hour – compared to the original (IBM) solution, based on reloading from tapes,  which required between five and six hours for complete recovery. The tape-based solution was also very risky. For example, if a single tape failed to load properly, it would have been necessary to reinitialize the entire system, potentially taking more than a week since there was no way to return to the previous state. In addition, the Hitachi Vantara solution offered greater flexibility, enabling SNCF to restore previous versions of the production system to a test environment for diagnosis of issues.

“Beyond the technical superiority of their hardware and software, Hitachi offered us a real partnership,” says Bertrand Delaporte. “If we ever had a concern, we could pick up the phone and get a technical expert on site right away. There was complete transparency around incidents, and we always had the impression of working with them on a person-to-person level. It was much simpler than with other large organizations, where you might have to raise a ticket and then watch as different departments blame each other for the problem.”

Alain Mortier concludes, “Our relationship with Hitachi could not have been better. Whether we were dealing with salespeople or technical experts, it was the same story; a more-than-20-year relationship characterized by trust and partnership, which made it a great success on both sides.”