Nightman1965 - Fotolia
The move toward more cloud deployments is driving demand for distributed database technologies, which is benefiting many vendors, including Cockroach Labs and its CockroachDB database.
CockroachDB is a distributed SQL database that is well-suited for cloud-native deployments. Cloud native is an increasingly common model of workload deployment using containers and the Kubernetes orchestration platform to provision workloads in public, private and hybrid cloud deployments.
CockroachDB 19.2 became generally available Nov. 11, providing increased performance and improved backup capabilities.
In the long range, cloud native will likely be the predominant direction for most databases, said Tony Baer, principal at New York-based consulting firm dbInsight.
"Right now, cloud native might be the exception, but I think, you know, down the pike, it will be the rule," Baer said.
CockroachDB has an advantage for cloud native as it was designed from its inception to be distributed, while many other databases have had to adapt to become distributed, Baer said. Now, relatively few distributed transaction databases are on the market, as it's been difficult to build them.
""That's always been a very hard nut to crack," Baer said.
Cockroach Labs first brought CockroachDB to market two years ago, and the database, as well as the company, has evolved since then. Finding the right market fit to attract paying customers has helped push CockroachDB forward, Cockroach Labs co-founder and CEO Spencer Kimball said.
Tony BaerPrincipal, dbInsight
"Financial services actually is a very important vertical for us," Kimball said. "I guess it wasn't clear, when we started that, you know, whether a big bank, for example, would be ready to consume, CockroachDB, for example, over Oracle for existing use cases or even for net new use cases, but it turns out that they are, and that's sort of where we found our product market fit, is with bigger enterprises."
The distributed nature of CockroachDB helps with both geo-replication, as well as accommodating users in different parts of the world in an optimized way, Kimball said. CockroachDB makes it possible to partition so data is closer to the user, based on geography, he said.
CockroachDB 19.2 features
Among the new features in the CockroachDB 19.2 update is a parallel commits capability that helps to improve overall database performance. Parallel commits are an important step in helping CockroachDB to deal with the challenges of geographic latency, Kimball said.
"Customers can essentially be anywhere and there's an increasing amount of data sovereignty regulations, so we really do have to figure out how to make distributed transactions work well when there is geographic latency," he said.
With a distributed database, there is a need for consensus across the different database nodes to make sure data is consistent. Kimball explained that even though the database might write, for example, 100 inserts in the course of a transaction, with parallel commits, CockroachDB 19.2 is not doing the commits in a serial execution fashion as 100 consecutive commits. Rather, the database executes a single commit, no matter how complex the transaction.
Kimball explained that, for example, if before the CockroachDB 19.2 update it took 10 milliseconds for consensus across data centers, and the database did 100 inserts, then it would take about one second for the entire transaction. In contrast, with parallel commits the same transaction now takes 10 milliseconds, he said.
Backup and restore
CockroachDB 19.2 also boasts improved backup and restore capabilities. Geo-partitioning of data is a core feature of CockroachDB and it's important that data backups follow the same policies, Kimball noted.
"You want to make sure that when you back up the database, you're not sending U.S. users' data to the EU and vice versa," he said.
In the past, with data backups, users would just point to an Amazon S3 storage target, for example. Using another example, the CockroachDB 19.2 update now ensures that data in the EU is backed up only to storage targets at S3 that are in the EU.
"Backup and restore is becoming more intelligent in terms of its understanding of the constraints that the operator wants the database to abide by," Kimball said.