Pliops unveils accelerated Key-Value store boosting RocksDB performance

1 min read

Pliops, a leading provider of data processors for cloud and enterprise data centres, has introduced XDP-Rocks, its new accelerated Key-Value store.

Designed to overcome architecture limitations of traditional RocksDB deployment, XDP-Rocks both streamlines architecture and eliminates complexities to deliver improved levels of workload performance, scalability and efficiency.

According to Pliops, XDP-Rocks increases throughput by 20x while reducing tail latency by 100x and normalised CPU by 10x in RocksDB-based databases.

XDP-Rocks is API-compatible with RocksDB, an open-source storage engine technology that has become the de-facto standard for modern data management software services. It provides orders-of-magnitude performance improvements versus RocksDB on ubiquitous NVMe storage by leveraging Pliops’ Extreme Data Processor (XDP) accelerator hardware. Its drop-in compatibility with RocksDB also eliminates the complexities of migrating the applications off the legacy storage engine.

XDP-Rocks optimises CPU and storage resources resulting in:

  • Massive application performance acceleration
  • Lower infrastructure costs
  • Greater scalability with use of larger data sets
  • Faster analytics and time-to-insight
  • Reduced blast radius
  • Increased endurance

According to Pliops XDP-Rocks was designed to overcome the unstructured data challenges facing modern enterprises, including cloud computing and data services providers.

“With XDP-Rocks, we are strengthening our industry leadership and offering a leap forward in data infrastructure performance improvement while reducing associated costs,” said Uri Beitler, Pliops founder and CEO. “We are constantly working to enhance our data services engines in terms of value and features – and the introduction of XDP-Rocks is just the tip of the iceberg. XDP-Rocks is enabling our customers to simplify and accelerate key-value based applications including MySQL, MyRocks, Redis, KV-Rocks, TiDB, Kafka, Spark and many more.”