Includes Materialized Views, Secondary Indexes, Filtering, and more.
AWS i3.metal Benchmark:
Comparing Scylla on 4 i3.metal nodes versus Cassandra on 40 nodes of i3.4xlarge
Samsung Benchmark:
Throughput of ScyllaDB vs Apache Cassandra Using 2TB Data for Different YCSB Workloads
YCSB Benchmark:
Scylla vs Apache Cassandra Using only 3 Scylla Nodes
Lockless implementation and an independent memory management stack remove the inefficient reliance on JVM or Linux page cache to deliver consistently low latency.
Automatic failover and replication across multiple nodes and data-centers makes for reliable fault tolerance.
Written in C++ to squeeze every bit of performance from hardware and allow up to 1,000,000 read/write operations per node.
Auto-sharding, homogeneous servers, and native multi-datacenter implementation allow seamless linear scaling without compromising on application downtime or performance.
Apache Cassandra’s wire protocol, a rich polyglot of drivers, and integration with Spark, Presto, and Graph tools make for resource-efficient and performance-effective coding.
ScyllaDB has been open sourced since day one and is backed by a growing community of contributors.
Workload conditioning provides an umbrella of dynamic scheduling algorithms to minimize database operation latency jitter and reduce compaction streaming and repair time.
C++ framework boots effectiveness to capture up to 10x throughput over existing infrastructure.
Fully automated and dynamic tuning effectively manages internal resource, requires zero configuration, and enhances performance straight out of the box.
| What makes “a good” vs. a great database deployment? | Apache Cassandra | Scylla |
|---|---|---|
| High Availability | 
Cassandra: Multi-region and fault tolerant |
Scylla: Multi-region and fault tolerant |
| High Scalability |
Cassandra: Homogeneous nodes scale to hundreds of nodes per cluster |
Scylla: Homogeneous nodes scale to hundreds of nodes per cluster. Additionally, Scylla scales up with the number of cores in our server |
| Low Latency | 
Cassandra: Unpredictable and unbounded latency, mainly the result of the JVM’s GC |
Scylla: C++, shared-nothing and lockless novel design allows < 1 millisecond tail latency |
| High Throughput |
Cassandra: Limited per-node performance. Cannot fully exploit the disk/memory/cpus |
Scylla: Fully asynchronous core engine with shard-per-core architecture allows perfect scale-up |
| Less Management |
Cassandra: Intricate tuning, endless compactions and GC storms makes it hard to maintain |
Scylla: Out-of-the-box tuning, isolation and SLA between background and foreground jobs free DevOps time during day and lets them sleep at night |
February 19, 2019
Meetup: How to Write a Database / Avi Kivity, CTO & Co-Founder, ScyllaDB
February 21, 2019
February 21, 2019
Meetup: Materialized Views in distributed systems like Scylla/Cassandra
SE-Radio Episode 354: Avi Kivity on ScyllaDB
February 11, 2019 | Software Engineering Radio - A Podcast for Professional Software Developers
ScyllaDB Makes Enterprise Cluster Administration and Task Automation Tool Available to Scylla Open Source Users
February 5, 2019
SD Times Open-Source Project of the Week: Scylla
January 22, 2019 | SD Times
Sail towards lower latencies and higher performance for your NoSQL databases with Scylla
January 22, 2019 | JAXenter.com
ScyllaDB Announces Major Release of its NoSQL Database and Support for Concurrent OLTP and OLAP
December 17, 2018 | InfoQ
ScyllaDB achieves Cassandra feature parity, adds HTAP, cloud, and Kubernetes support
November 7, 2018 | ZDNet
ScyllaDB: Product Overview and Insight
August 21, 2018 | eWeek
ScyllaDB is better than Cassandra, and here’s why
April 1, 2018 | EPeak
Getting started takes only a few minutes. Scylla has an installer for every major platform and is well documented. If you get stuck, we’re here to help.
Terms of Use Privacy Policy ©ScyllaDB 2019
Apache®, Apache Cassandra®, are either registered trademarks or trademarks of the Apache Software Foundation in the United States and/or other countries. No endorsement by The Apache Software Foundation is implied by the use of these marks.
