Did you know that object storage can be high-perf and easy to scale?
OpenIO can be used for Big Data and AI workloads, and even as a warm tier for HPC use cases. We explain how our object storage technology is designed to deliver the highest performance and easiest scalability.
When we designed our object storage solution back in 2006, we made the decision to build an architecture as a grid of nodes, instead of using the more common consistent hashing algorithm. With a grid architecture, you can easily add nodes and make them available in real time. And this can be done without rebalancing data, which saves a ton of hassle and time.
Dynamic data placement
We then developed our ConsciousGrid™ technology on top of the Grid of Nodes. The ConsciousGrid dynamically places incoming data on the most appropriate node, every single time. It even works when you add new nodes, so you maintain consistent performance while scaling.
OpenIO scales really well, and this is one of the main benefits that differentiates it from other storage solutions. When you want to scale with OpenIO, you can add new nodes instantly. A lot of vendors require you to plan your storage needs a few months or even years in advance, which is not always practical and can become a huge expense. With OpenIO, you can add capacity to your platform as you need it, in small or large batches.
OpenIO is also highly resilient. Even if an incident were to damage a node, performance would not be affected, and quality of service would remain consistent.
OpenIO has proven that a new generation of object stores is now ideal for workloads which require high bandwidth performance, such as Big Data and AI. We recently benchmarked our solution against Hadoop’s HDFS, and the results demonstrate that OpenIO Object Storage can outperform HDFS on many commands, especially those with large workloads. So if you want to separate your storage and compute clusters, OpenIO is great alternative to HDFS.
Alternative to the public cloud
These benefits are something that several of our customers have taken into account, as they migrate data from the public cloud to a private cloud or even on-prem, with OpenIO. Both of these solutions have given our customers a drastic reduction in their monthly costs. And, as a bonus they now have dedicated resources to optimize performance.
OpenIO's architecture is also unique in the way that it really scales when you add more machines. We are able to just keep adding spinning disks if it looks like we're running out of bandwidth.
We have improved our performance, and reduced costs by 40% from public cloud. And we’re still agile: the web host adds nodes when necessary to increase the capabilities of the platform.