As NoSQL databases catch on, companies that deploy them may realize they need new types of backup to protect them.
MDL uses NoSQL-based document database MongoDB to store 1.4 TB of customer data across a 15-node sharded cluster deployed in three data centers. The company must restore and refresh daily to a single, non-sharded cluster — a quality assurance (QA) test database.
MDL manager of infrastructure Eric Gutmann said he struggled to find backup software that could handle restores fast enough. After trying MongoDB native backup and a traditional backup vendor, Gutmann switched to startup Imanis Data to improve backup and restore times.
Imanis Data specializes in backup of NoSQL distributed databases.
“We had done a lot of research prior to using our other software, which, at that point, we were informed was the only company that could do what we wanted to do,” Gutmann said.
MDL sells software to connect with RFID, Bluetooth and license plate recognition devices to help car dealerships streamline customer service orders, manage lots and track loaner cars. The MongoDB database contained information on the roughly 250 dealerships MDL works with, as well as data for their customers, such as vehicle identification numbers. Gutmann said it is crucial to the business to protect that information.
Native MongoDB backup couldn’t meet his needs, so Gutmann turned to an outside backup software vendor. That software fulfilled his most basic requirements, but presented a new problem — the restore took more than 48 hours. By then, the data restored to the QA database was outdated and useless to the QA team.
“After 48 hours, some of that data is dead,” Gutmann said.
Quick setup, quick restores
Gutmann purchased Imanis Data for MongoDB in late 2018, drawn in by its incremental backup and granular recovery capabilities. With the new software, he said he’s now able to perform hourly backups and nightly restores. Backing up only took 10 minutes, while restores only took two hours. All in all, he said it was a vast improvement over what he had before.
When asked for other vendors he looked at before purchasing Imanis Data, Gutmann said he did not find many choices for a MongoDB backup vendor.
But Gutmann said after attending MongoDB World last year and doing other research, he found few options that fit his data protection needs.
Gutmann said he was impressed by how quickly he was able to get Imanis Data up and running. His old backup system took six months to deploy, with his IT team spending a minimum of four hours a day devoted to setting it up. Gutmann said he got Imanis deployed on four servers in just two hours.
Gutmann is looking for a few improvements to Imanis Data software, however. He wants its interface to display database numbers with fewer clicks. Also, he hopes the recovery sandbox feature introduced in Imanis Data 3.3 will someday work with MongoDB. Gutmann said he would like to create virtual test environments for restores and find ways to incorporate that into his current QA plan.