CockroachDB: The Resilient Geo-Distributed SQL Database

Abstract

We live in an increasingly interconnected world, with many organizations operating across countries or even continents. To serve their global user base, organizations are replacing their legacy DBMSs with cloud-based systems capable of scaling OLTP workloads to millions of users.CockroachDB is a scalable SQL DBMS that was built from the ground up to support these global OLTP workloads while maintaining high availability and strong consistency. Just like its namesake, CockroachDB is resilient to disasters through replication and automatic recovery mechanisms.This paper presents the design of CockroachDB and its novel transaction model that supports consistent geo-distributed transactions on commodity hardware. We describe how CockroachDB replicates and distributes data to achieve fault tolerance and high performance, as well as how its distributed SQL layer automatically scales with the size of the database cluster while providing the standard SQL interface that users expect. Finally, we present a comprehensive performance evaluation and share a couple of case studies of CockroachDB users. We conclude by describing lessons learned while building CockroachDB over the last five years.

Studi Literatur: Perbandingan Basis Data NewSQL

“…CockroachDB menjadi jembatan untuk pusat data dalam menyediakan layanan otomatis secara terus menerus dan transparan dalam menghadapi kegagalan sistem (fault) skala besar. Sehingga aplikasi tidak perlu memiliki pengetahuan atau penanganan khusus untuk kondisi tersebut [39][40] [41]. Basis data ini mendukung data historis, dengan menawarkan bounding data untuk menyediakan transparansi dan kemudahan servis yang berkelanjutan [42].…”

Spitz

“…Second, the abort rate can be high in a write-intensive workload. To address the first limitation, we can adopt the hybrid logic timestamp scheme that allocates timestamps by each individual node and still has serializability guarantee [28,50]. For the second limitation, it is possible to adopt the combination of OCC and MVCC by dynamically adjusting the transaction order to reduce abort rates [19,34], and verifying the transactions in batch to reduce the verification cost [20].…”

Scalable but wasteful

“…Strongly consistent replication is the backbone of modern cloud services, ranging from configuration management [13,29] to cloud datastores [8,22,26,28,33,39]. To ensure strong consistency, these services typically implement state machine replication using consensus protocols [4], henceforth referred to as replication protocols, which need to scale and provide high-throughput, low-latency services to the users.…”

“…As such EPaxos Modern cloud services, however, rarely operate on dedicated resources; on the contrary, physical resource sharing with some resource isolation is the norm. Moreover, most systems scale horizontally by running multiple protocol instances side-by-side to support different data partitions [8,26,28,36]. In such shared environments, the resources are allocated and scheduled using some sort of task packing [11], leaving only limited free resources on each node.…”