The LightOS 1.0 release in Spring of 2019 introduced a new type of storage solution: high performance, low latency, software-defined storage on the most ubiquitous fabric: Ethernet and TCP/IP. Simply utilizing Ethernet was not enough – after all, iSCSI uses Ethernet. NVMe-oF was gaining ground and also worked on Ethernet but utilized RoCE – a protocol requiring special NICs, special OS drivers and special switch settings to emulate a lossless fabric. LightOS 1.0 was special in that it provided NVMe performance over a standard TCP/IP network. It provided a standards based solution that fit into your existing infrastructure with minimum transition cost. In addition to providing NVMe performance, this was no “stripped-down” NVMe-oF solution – LightOS 1.0 provided thin provisioning, Elastic RAID data protection/striping and data compression.
We are excited to let you know that LightOS 2.0 is here. With the LightOS 2.0 release, we have enhanced availability, performance, latency and optimized features for QLC based SSDs. It’s the world’s first and only scale-out, highly available, software-defined NVMe/TCP storage. This is game-changing for existing and new Lightbits customers, and the industry at-large.
Here are the highlights:
1. NVMe/TCP Multipath failover with Domain Awareness
LightOS 2.0 based clusters deployed on standard x86 servers can physically reside anywhere in the data center. Application servers utilize standard multi-path NVMe/TCP for high availability against target or network failures. Our cluster management solution takes into account the failure domains of each server and ensures efficient volume placement for maximal availability and reliability of high performance block storage.
2. Optimization for QLC based SSD’s
LightOS 1.0 implements intelligent flash management that isolates the application workloads from SSD’s, and extends the life of SSD’s by optimizing the internal workload for NAND. LightOS 2.0 extends this capability to QLC SSD’s. Without LightOS, QLC SSDs are limited to read-heavy workloads. With LightOS 2.0 improving the performance and write endurance of QLC, regardless of workload, customers can take advantage of a more competitive media cost without compromising reliability, availability or performance.
3. Per Volume Protection Scheme
LightOS 2.0 introduces the ability to create logical volumes with different protection schemes. Given applications and end users have differing and evolving needs, this flexibility allows end users to provision the storage at the size and protection level they need. A scratch space or ephemeral storage may not require replication, whereas a highly available database may require multiple copies the highest level of server availability. This flexibility allows service providers to offer different services at differing pricing levels to their end users while using the same API and getting maximum utilization of their storage capital investment.
4. Kubernetes CSI Driver
LightOS 2.0 release includes a Kubernetes CSI driver to provide seamless persistent storage for containers with on demand scaling and overcomes the shortcomings of running Kubernetes on local storage by eliminating physical server boundaries. LightOS also provides high availability with automatic failover and no disruption of storage services – eliminating needless network rebuilds and pod movement due to storage failures.
LightOS 2.0 is a revolution in storage; addressing today’s issues with ease of use, scalability, and flexibility. It provides a single storage platform regardless of your deployment model, size of data center and works with heterogeneous server configurations all on standard Ethernet TCP/IP networks. Utilizing standard x86 servers, there is no vendor lock-in; newer technologies, server platforms, chipsets can be deployed as soon as they are available or when their cost is at an optimal level thus eliminating cycles of qualification and characterization.
