5 reasons why choosing Apache Cassandra is planning for a multi-cloud future

Transcription

1 White Paper 5 reasons why choosing Apache Cassandra is planning for a multi-cloud future Abstract We have been hearing for several years now that multi-cloud deployment is something that is highly desirable, with the driver being that companies want to be capable of being cloud agnostic and not beholden to one single vendor. While this is a nice to have, the reality has been that multi-cloud has been viewed as hard and simply moving to the cloud is hard enough. Something has changed over the last 3 months. We are seeing these requirements becoming definitive and real, with larger enterprises now at least planning on being cloud agnostic and/or multi-cloud capable. It is like the first wave of cloud adoption has happened, and the early adopters are now planning for the future. Apache Cassandra goes a long way to underpinning and enabling multi-cloud deployments. So even if you are not doing multi-cloud right now, you should be thinking about future proofing your architecture. This paper provides 5 reasons why Apache Cassandra is an enabler for true multi-cloud deployments.

2 Overview Even if an organization is not ready for a multi-cloud active-active deployment, choosing Cassandra as a primary data store ensures that any future migrations to a different cloud provider, or adoption of true multi-cloud is simple and easy at the data layer, with no data related migrations or downtime. Choosing Cassandra with other widely adopted and standardized technologies, that either have easily portable APIs (like Kubernetes for container orchestration) or support for datacenter awareness or configurable replication like Kafka, will ensure a significant portion of an application stack can be easily spread across multiple cloud providers or easily extended should the need arise. Here are the 5 reasons why we believe choosing Apache Cassandra can be a real technology foundation for planning a multi-cloud future. Reason 1: Cassandra has topology-aware availability Cassandra is, in relation to CAP theorem, an AP database that allows you to tune consistency to your own internal requirements at query time. It also allows you to structure replica placement around common groups of machines that are likely to fail at the same time. All architectures haves a failure domains - groups of machines that are part of a Cassandra cluster could be on the same hypervisor, in the same rack or the same availability zone., Whatever resources your machines share, you can make Cassandra aware of them so that it makes a group outage is less likely to impact availability. Combined with Cassandra s tuneable consistency you can choose how consistent you want a query to be with these failure domains. Cassandra calls these failure domains racks and datacenters. Racks are failure domains within a data center and should have high bandwidth, low latency connections between them. They are named after physical racks in a physical datacenter, where machines in a single rack share common power and networking infrastructure. A typical Cassandra configuration is to use three racks and a replication configuration that ensures each rack has a complete replica of the data If a whole rack fails then two more replicas of the full data set will still be available and strong consistency (QUORUM) queries can still operated on the cluster. Reason 2: Cassandra has remote regional awareness Datacenters within Cassandra are considered to be remote, with potentially lower available bandwidth and higher latency. Cassandra drivers are aware of the local datacenter they connect to, nodes within a Cassandra cluster are aware of which nodes are in their local datacenter. Cassandra s tuneable consistency mechanisms allow you to isolate the strength of consistency relevant 2

3 to the data center a query is sent to. The awareness of remote failure domains, datacenters, baked into Cassandra from day 1 means that developers are made to think about the how their queries interact with database failure and physical topology. Reason 3: Cassandra has flexible global and local consistency Cassandra s fault domain model works incredibly well within the cloud model and actively encourages developers to build applications that work across multiple data centers, availability zones, and regions. Once you have an application that supports multiple regions, the logic is fairly simple to be extended beyond a single cloud provider, as a second cloud providers region is likely to be just as remote in terms of latency as your primary cloud providers remote regions. Cassandra s ability to isolate consistency to within a datacenter also means that adding new regions/ datacenters to the cluster has minimal impact on operations and is a no downtime operation. The other side of this coin is that Cassandra can also let developers obtain guarantees that queries are globally consistent through higher consistency levels where application requirements demand this. Reason 4: Cassandra has easy and powerful replication Region/datacenter replication of data can be configured at the keyspace level, a logical grouping of tables, that allow developers to replicate data to specific subsets of regions in a given Cassandra cluster. This is truly powerful for your data compliance requirements. The configuration of data at the keyspace level enables developers to ensure specific data resides in specific locations to easily meet customer, business and government compliance requirements such as GDPR regulations. Applications/ tables with requirements to be in multiple cloud providers can be separated out from applications/ tables that are only relevant to a given cloud provider. Reason 5: Cassandra is licensed for freedom Combined with the fact that Cassandra is a liberally licensed open source software project under the control of the Apache Software Foundation, it can be readily run in any cloud provider, any colocation environment, or any private datacenter. With Cassandra you have a database that gives you incredible flexibility and freedom of choice when compared to proprietary databases or database services provided by your primary cloud provider. Database services are often structured as a lock-in mechanism within a cloud provider. Apache Cassandra provides an easy, cost-effective antidote to cloud vendor lock-in. 3

4 Final Word: True multi-cloud is hard, Cassandra makes it easy Truly adopting multiple cloud providers and running an active-active deployment across dissimilar environments, such as different cloud providers, is time-consuming and challenging. Tooling such as configuration management, CI/CD, alerting, resource planning and routing/dns all will have specific requirements for each cloud provider creating separate process and code paths to support. Adopting an open source multi-cloud database like Apache Cassandra ensures that the data layer does not have similar challenges. Adopting a managed database provider that supports multiple cloud providers further abstracts away cloud provider incompatibilities. 4

5 About Instaclustr Instaclustr is the open source as-a-service company, delivering reliability at scale with Apache Cassandra, Apache Spark, Elassandra and other related technologies. Our expertise leverages over 15 million node hours and one petabyte under management, allowing us to run the world s most powerful NoSQL distributed database effortlessly. Our open source data management solutions power mission critical, highly available applications for our customers and help them achieve scalability, reliability and performance in their database. Our Services Instaclustr provides range of consulting and support services to help our customers develop and deploy solutions that use Apache Cassandra and related technologies. Managed Solutions Consulting Services Enterprise Support Our managed solutions enable our customers to run seamlessly in the cloud with confidence that our team is monitoring and managing their clusters. Our team of experts have extensive experience with a wide range of use cases and deployed architectures. Our dedicated 24x7 support team of experts have hundreds of thousands of hours of operational engineering and support experience. Build, run and scale your app with confidence Like what you see? If you re looking at building a proof of concept for your application, or looking for production grade nodes, contact a member of our Sales Team to discuss your specific needs. 5