New Workload Prioritization Feature Enables Businesses to Combine Operational and Analytics Workloads on the Same Database Cluster
ScyllaDB announced the availability of Workload Prioritization, an industry-first feature that enables developers to run different kinds of workloads at the same time, on the same NoSQL database cluster. ScyllaDB’s latest innovation promises to cut database hardware spend by up to 50%.
Workload Prioritization does this by eliminating the need to segregate database workloads into different clusters. Operations and analytics are a prime example. Operational workloads mainly consist of rapid-fire writes, reads, updates and other small, latency-sensitive transactions. Analytics workloads are less latency-sensitive but more dependent on high throughput for scanning large datasets. To avoid conflicts between the two, organizations typically maintain separate database clusters for each use case, even though it means duplicating resources and effort.
Workload Prioritization makes these sorts of silos disappear by enabling teams to allocate “shares” of total resources to each workload on a cluster. As a result, a single Scylla cluster can efficiently serve heterogeneous workloads—some operational, some analytical—while meeting performance expectations even in the face of traffic spikes and background maintenance operations.
“Database users are tired of dealing with sprawling topologies, tweaking clusters and customizing them to support different workloads,” said Dor Laor, CEO and co-founder of ScyllaDB. “With our Workload Prioritization feature, they don’t have to duplicate resources, isolate analytics workloads or dedicate an entire cluster to every microservice. They can consolidate their data in a single cluster, get rid of idle hardware resources and dramatically reduce maintenance costs. Workload Prioritization makes everything simpler, easier and less expensive.”
Workload Prioritization is the latest achievement in ScyllaDB’s’s mission to shrink database infrastructure to the smallest, most powerful and most highly utilized footprint possible.
A Growing List of NoSQL Advancements
ScyllaDB’s other industry firsts and unique differentiators include:
- Shard-per-core Architecture: Scylla has a two-level automatic sharding design, with cluster-level shards and core-based sharding, which creates a perfect shared-nothing architecture. This architecture enables linear scale-up and maximum utilization of all available CPU cores and I/O devices.
- Unified Cache: Scylla’s unified cache simplifies internal operations by eliminating multiple competing caches, while also dynamically tuning itself at runtime to accommodate varying workloads. For Scylla users, this design means higher ratios of disk to RAM, enabling each node to serve more data. This, in turn, enables smaller clusters with larger disks.
- Self-optimizing Operations: Scylla relies on a scheduler that tags foreground and background requests according to the origin of the I/O operation. Requests can then be metered, enabling the scheduler to prioritize operations. Operators spend less time tuning, confident that background operations will complete as quickly as possible without affecting performance.
- Materialized Views and Secondary Indexes: These features allow users to organize, search and filter data in different ways. Scylla’s Materialized Views are production ready (whereas they remain experimental in Apache Cassandra). Scylla’s Secondary Index feature is built on top of Materialized Views and improves the granularity of both global and localized queries.