7 min to complete
So why do we exist? Why have we spent so much effort building this technology? Four key benefits make us stand out, and they are:
- A lower node count
- Consistent performance
- Reduced complexity
- Apache Cassandra compatibility
So what do we mean by a lowered node count? Let’s go over a specific example. There is a company who was running 120
nodes of Apache Cassandra. They were running on AWS and spending over $600,000 a year in data center costs. They were doing about a few million operations per second on about a 10 terabyte data set. Clearly, they wanted to save money and see the same or better performance. They came to ScyllaDB and asked us if we could do better. ScyllaDB not only rose to the challenge but delivered on all points. The first step was to just cut down from 120 to 24 boxes. This delivered significant cost savings and allowed them to reduce their data center cost by 80% and their node count to a fifth of what it was before. Naturally, this node reduction also yielded a significant reduction in administration efforts in terms of maintenance, upgrades, and so on. But we
weren’t finished just yet. There was one more step that had to be done. ScyllaDB doesn’t only scale-out and provide a ton of flexibility, but a key difference from Apache Cassandra is our ability to scale up. Why manage 24 small machines when you can do the same throughput at relatively the same cost with three machines? So we move them to 3 i316XLs. What this meant for our customer was significant savings and management overhead with better performance. This was a definite win-win.
We’ve talked about lower node count. What do we mean by consistent performance? This graph which comes from the ScyllaDB Monitoring stack, tells the entire story. This graph depicts a customer-serving customer-facing traffic. Each time the traffic spikes means your customers suffer poor performance. And if you’re an Apache Cassandra user, you’re familiar with the upper yellow line, right? Right there, that’s garbage collection, compaction, compaction, compaction, garbage collection, and so on. Without going into too many details, these are maintenance operations done by the database in the background every once in a while. The timing for doing them is critical. What you’ll notice about the lower line, the green one, which is ScyllaDB, is how flat it is, how tightly bound it is. And why is that?
Because we are much better about providing consistent performance, so not only are we more performant, but we’re consistently performant.
So what do we mean by reduced complexity? Here we’re talking about the complexity of running Apache Cassandra that we solve. When you install ScyllaDB, it benchmarks against the hardware that it’s run on. For example, ScyllaDB says, how many cores do I have? What’s my IO throughput? What’s my network throughput? And then it tunes itself to those parameters. 90% of the time, you’re never going to have to touch those parameters: your workload changes, no problem. You don’t have to change anything. If you change hardware, no problem. ScyllaDB will simply re-benchmark. The parameters change themselves. That means less headache, less worry, and it’s all managed by ScyllaDB. So that’s what I mean when I say reduced complexity.
ScyllaDB is Apache Cassandra from an architectural standpoint, from a use case standpoint, and from an API standpoint. It’s just Apache Cassandra with a lower node count, consistent performance, and the reduced complexity. ScyllaDB is not only a drop-in replacement for Apache Cassandra. It can also be used with integrated solutions, such as Presto, Spark, Kubernetes, or Kairos. ScyllaDB provides you with drivers to get the job done. Whatever you’re using to speak CQL, we got you covered. If you’re using the Java driver, or you’re using the C driver, or the C++, driver or the C-sharp driver, or the PHP driver, or the GO driver. Whatever you’re using, you can simply take the current toolset and the current engineering and just change the IP address to a ScyllaDB cluster, and it will simply work.
ScyllaDB is offered in three variants. The open-source version licensed under the AGPL is a fully functional drop-in replacement for Apache Cassandra. It offers great performance, a lower node count, and reduced complexity. ScyllaDB Enterprise is based on the ScyllaDB open-source project. It includes tested and certified production-ready binaries, software updates, and hotfixes. Additionally, it includes technical support that guarantees that you have access to the engineers who developed ScyllaDB. This version also includes the ScyllaDB Manager, which provides centralized cluster administration and recurrent task automation. This includes automation of periodic repair, as well as other features which are exclusive to ScyllaDB Enterprise customers. In addition to all those features, the ScyllaDB Cloud offers a fully managed solution. It allows you to run ScyllaDB Enterprise as a highly affordable cloud service. Sparing your team from database administrative tasks, ScyllaDB Cloud gives you access to fully managed ScyllaDB clusters with automatic backup, repairs, performance optimization, security hardening, and 24/7 maintenance and support. This is offered for a fraction of the cost of other database as a service offerings.