Ultra High Performance
Open vStorage is the fastest distributed block store on the market. Some facts to prove this bold claim: 500,000 IOPS per host, 50 million IOPS in a single rack and sub 0.5 millisecond latency. To achieve this best-in-class performance lays in the Open vStorage architecture. The log structured write buffer in RAM or on NVMe flash offers the stunning write performance even for random writes. The distributed local flash pool takes care of the read performance.
Questions? We got answers!
How fast is Open vStorage?
Open vStorage is wicked fast! On standard hardware you will reach 500,000 IOPS for a 4K random IO workload (70% read, 30% write). Open vStorage also makes use of the latest technologies in networking such as RDMA to build a distributed read cache across all SSDs in a rack or even datacenter.
How does Open vStorage ensure performance across datacenters?
Open vStorage uses a tiered approach where hot data is kept in the SSD cache and cold data remains on the capacity tier which is distributed across datacenters. By using this approach it can offer stunning performance while distributing the data across multiple datacenters for safety.
Does the performance scale?
The performance scales linearly with the addition of each host. Each additional host can add upto 500,000 IOPS to the storage pool. For a full rack the performance scales up to 50,000,000 IOPS.
Is Open vStorage also fast for random IO workloads?
Yes, Open vStorage uses the same log structured approach as today’s leading databases such as RocksDB and MongoDB to serialize random IO into sequential IO. All incoming writes are processed in an append-only fashion in the write buffer. Once enough writes have been accumulated the log, a Storage Container, is closed and stored safely on the backend.