LightBits on SuperMicro

Lightbits’ software-defined, disaggregated and high-performance storage can be deployed on Supermicro SuperServers over standard ethernet networks using TCP/ IP protocol, requiring zero modification to applications.

While software-defined storage is desirable for its ability to lower cost by utilizing standard servers and components, a “do it yourself” approach is not for everyone.

For the ultimate in scalability, performance and convenience, this ready-to-deploy solution delivers composable NVMe® over TCP-based block storage with built- in Intelligent Flash ManagementTM that increases flash endurance by 20X. It’s an ideal platform for applications running on-premises, in private and edge clouds in containerized or virtual environments.


Lightbits offers rich data services such as thin provisioning and compression. A single SuperServer in a Lightbits cluster with redundancy enabled can serve up to 4.7 million random 4K read IOPs, up to 760,000 random 4K write IOPs, up to 21GB/s of read bandwidth and up to 4.3GB/s of write bandwidth.

  • Unmodified Software on Clients
    • Standard NVMeoF 1.1 w/multipathing (ANA)
    • Clients can connect to multiple clusters
  • Clustered/Failover Storage Solution
    • Distributed cluster management with fast failover and no single point of failure
    • Cluster size: 3-16 servers
    • 64K Volumes per cluster
  • Lightbits Cluster Performance
    • 75 Million 4K Random Read IOPs
    • 12 Million 4K Random Write IOPs
    • 336 GB/s Read Bandwidth
    • 68 GB/s Write Bandwidth
  • Lightbits Cluster Capacity (15.36TB drives)
    • 4.68PB Effective Capacity (2x
    • replication, Elastic RAID, 2:1 compression)
  • Lightbits Cluster Latency with
    • 2x Replication
    • 160μs average latency 4K Random
    • Read @1.27M IOPs per server
    • 493μs average latency 4K Random
    • Write @433K IOPs per server
  • Storage Services
    • QoS
    • Thin provisioning
    • Compression
    • Elastic RAID for drive failure protection
    • Volume replication (1x, 2x or 3x) per volume
  • Node Management
    • Replication and failover handling • Failure handling using NVMe/TCP
    • Multipath
    • Non-disruptive cluster upgrades (for 2x and 3x replicated volumes)