More than ever, Kubernetes applications need enterprise-grade data protection and resiliency. As organizations expand their workloads on Kubernetes to include a growing number of business-critical applications, they need tier 1 data protection and data management capabilities at scale. They need services like high availability, data protection, resiliency, and efficiency.
With NetApp® Astra™ Data Store, you can meet these requirements head on. It’s therefore no surprise that Astra Data Store is architected based on the tenets of resiliency and scale. Further, the highly flexible service-based architecture makes each service fault domain-aware so that you can independently scale services while preserving resiliency.
This blog post explains the key pillars of the Astra Data Store architecture:
Service-based architecture decouples application data management from storage management
NetApp Astra Data Store is built on a service-based composable architecture that’s integrated natively with the Kubernetes control plane. This flexible architecture is very well suited for your applications that are built for cloud scale. Unlike traditional storage systems that are built on tightly coupled subsystems, with the service-based architecture of Astra Data Store, the system automatically initializes loosely coupled services as needed.
Each service serves a specific system function and can be scaled independently of other services. This flexible architecture decouples data management services (application data management and data lifecycle functions) from storage management services (data and metadata block management, data reduction techniques, redundancy, fault tolerance and device management). The services run in a disaggregated fashion on different node pools as long as they’re attached to the same network fabric—making Astra Data Store a true composable architecture.
With this capability, the data management services can scale with your application as part of the application fault domain, while the storage management functions remain part of the storage fault domain and scale accordingly. Thus, decoupling of the data management functions from the storage management functions enables segregation of these functions to separate fault domains, which is critical for your system to achieve resiliency at scale.
Resiliency at scale is why Astra Data Store stands out
Resiliency is important for any business-critical, stateful application, especially when those applications run on commodity software-defined infrastructure. NetApp Astra Data Store delivers industry-leading resiliency for products in this class. You get a combination of cloud resiliency models and enterprise-grade data protection schemes to protect your file system data and metadata.
You also get a topology-aware data protection scheme. Every block of your data is replicated within the cluster based on fully customizable protection domains (availability zones). Your system is protected against node, rack, or entire availability zone failure. To enable automatic recovery, your system proactively monitors and auto heals at node, rack, and availability zone granularity. To complement the replication, each replicated block is further redundancy-encoded within the node. This approach enables protection and recovery from local drive failures or from bad blocks reported by the drive firmware. Astra Data Store can locally repair such data or data blocks within the node, thereby avoiding large internode data transfers to rebuild your data.
Astra Data Store also has built-in block-granular data consistency checkers, so your system maintains data consistency not only during a rebuild or other background operations but also during normal client read operations. Astra Data Store performs data consistency checking at multiple levels and can repair inconsistencies of data either inline or in the background.
This resiliency is what really differentiates Astra Data Store from any other product in its class. To achieve this level of resiliency, Astra Data Store has incorporated decades’ worth of NetApp innovation in reliability and data protection and combines the best of several industry leading data protection algorithms like Lost Write Protection, Automatic Incremental Repair, and bad block detection to protect your data from software, hardware and network failures.
Highly scalable shared file services promote consistent performance
Astra Data Store gives today’s cloud-native applications a highly scalable shared file system with ReadWriteMany or ReadOnlyMany access. Your cloud-native applications can scale horizontally by accessing the same persistent volume from multiple nodes. If you have legacy applications that use shared file system architectures, you can easily migrate them to Kubernetes while maintaining the same levels of data availability.
The service-based architecture combined with the distributed data placement algorithms of Astra Data Store gives you linear scaling of the file system with compute and storage resources. Any myths about the scalability of shared file systems have now been dispelled. By using all available system resources uniformly and by preventing any hotspots, the distributed algorithms manage your system load consistently across the cluster. The data layout and the corresponding data structures enable you to perform lookup and read operations as constant time operations. The application granular quality of service (QoS) of the filesystem prevents noisy-neighbor problems and ensures that each tenant obtains the performance levels that you specify.
See Astra Data Store in action
Are you ready for the next step in achieving Kubernetes application data protection and resiliency? Learn more about Astra Data Store and how it provides enterprise-grade data management, resiliency and protection. Then watch Astra Data Store in action. Get details about the Astra Data Store preview release and then sign up for your own preview.