Dbvisit Replicate Case Study - Galileo Processing
Galileo Processing used Dbvisit Replicate to successfully migrate over 1.5 TB of data to new database infrastructure whilst maintaining a service level of 99.999%
Galileo Processing is a rapidly growing provider of international payment solutions within the United States and internationally. Continued growth meant they needed to migrate their extensive customer and transaction databases to a new platform.
Galileo's systems have an up-time service level requirement of 99.999%, so they needed a migration approach which would minimize downtime. Their solutions also process upwards of 200,000 transactions per hour, so the migration approach needed to be reliable, robust and repeatable.
Galileo Processing used Dbvisit Replicate to successfully migrate over 1.5TB of data to new database infrastructure whilst maintaining service level of 99.999%.
Galileo Processing had a range of software applications which had been implemented over 10+ years. These applications sat on databases spanning Oracle versions, including 10g and 11g. They also operated on aging hardware that was beginning to restrict performance, prevent scalability and cause concerns about reliability.
To improve performance, reliability and supportability, Galileo decided to upgrade to new storage infrastructure that used Oracle 11g exclusively. The upgrade process had to fit within their available outage windows and needed to be completed without impacting service levels.
The traditional data migration approach involves stopping the database and associated application to lock the data, making it unavailable to users. The full content of the database is then copied to the new environment before the application is connected to the migrated database and restarted. The application is unavailable to users during this time.
With Galileo’s systems comprising 1.5TB of data across more than 50 schemas, some of which were up to 300GB in size, this approach was simply not feasible. It would not be possible to complete the full migration in the available outage window. An alternative approach was required; one that mixed alternative technology with innovative thinking.
Following research, Galileo Processing identified that a database replication approach would enable the incremental copying of databases while leaving the source database available for updating. This would allow the application to remain in use for the majority of the data migration, helping to maintain service levels.
Galileo assessed a number of data replication tools, including both physical and logical options. On assessing physical replication tools, Galileo found that this approach required the replication of the entire database in one step, necessitating a “big-bang” migration. This wasn't feasible for Galileo since it would have required them to migrate the full 1.5TB of data in a single pass.
The logical replication tools assessed by Galileo included Dbvisit Replicate which allows migrations to be performed at the database schema level, enabling data migration to be spread over a period of time. Dbvisit Replicate thus allowed Galileo to perform the migration in a series of steps, schema by schema, and gave them the flexibility to migrate applications as and when they chose. Dbvisit Replicate also allowed the data to be migrated while the applications remained available for production use.
Galileo selected Dbvisit based on the results of exhaustive testing, including a pilot migration, the cost effectiveness of the product’s licensing and the level of pre-sales support.
The migration process followed by Galileo started with exporting the data from the legacy database and simultaneously initiating data replication. At this point in the process the application continued to run in normal production operation due to the ongoing replication of changes by Dbvisit Replicate. Without this, the application would have needed to have been taken offline for an extended period of time during data exporting and importing. Galileo then imported the data to the new database before applying the replicated transactions to the base imported data.
Timings for the migrations exceeded expectations. These initial migrations were followed by migrations of data from non-Galileo systems. As an added benefit, and a by-product of the logical replication approach, the migration significantly improved the structure of the Oracle data files themselves.
Dbvisit Replicate's use of logical data writes, as opposed to the block-level writes used by a physical replication approach, created the target data files from scratch and, in doing so, removed the data fragmentation that had crept into the databases over many years of operation. It also reduced the size of these data files on disk.
This improvement in the physical structure of the data delivered its own performance benefits that are visible within the database itself.
Dbvisit develops disaster recovery and database replication products that ensure an organization’s critical data is available and up-to-date when and where it is needed; whether on-premise, in the cloud or a combination of both.
Dbvisit was formed in 2006 to develop Dbvisit Standby, an off-the-shelf, low-cost, high performance disaster recovery solution for Oracle Standard Edition (SE) database users. Today, Dbvisit Standby is the world's leading Data Guard alternative solution for Oracle Standard Edition. Dbvisit Replicate was developed in response to a market need for a smart alternative to Oracle GoldenGate and Oracle Streams products. It delivers the full power and potential of database replication at an affordable price.
Over 1000 business, academic, and public sector organizations in 110 countries trust Dbvisit products to deliver replication for their Oracle databases. Since 2011, Dbvisit has been ranked in the Deloitte Technology Fast 500 for Asia Pacific for the fastest growing technology companies in the region. Dbvisit is an Oracle Gold Partner.
Businesses in over 110 countries worldwide with Oracle® Database Standard Edition trust Dbvisit to protect and share their data.