Micron's HSE Open-Source Storage Engine 1.8 Released
Back in April the folks at Micron announced the "HSE" open-source storage engine optimized for SSDs and persistent memory. Version 1.8 of HSE was released on Friday as the first major update since going public earlier this year.
HSE amounts to a well optimized key-value store database geared for high performance solid-state drives and persistent memory. Micron's original announcement talked of HSE providing as much as doubling the throughput and improving read latencies by around four times. With HSE is also a modified MongoDB implementation as a real-world reference implementation. Besides MongoDB, Micron hopes HSE will see use for various NoSQL, SDS, and big data use-cases along with other verticals.
This HSE storage engine depends upon an mpool object storage media pool that is implemented as a Linux kernel module. Mpool interfaces with SSDs or other persistent memory storage directly, bypassing file-systems and other overhead while also supporting replication across classes of memory and other storage features. So far we haven't seen any work on Micron sending out the HSE kernel bits / Mpool for upstream review in hopes of getting it mainlined in the Linux kernel, but we'll see what more comes out in the months ahead.
With this week's HSE 1.8 release there is support for media class pinning and tiering policies, value compression settings, configurable ingest throttling, and Ubuntu 18.04.4 support. This is the first release seeing official Ubuntu support as previously Micron was focused on RHEL/CentOS support. Micron does note though that HSE on RHEL is seeing around 5~30% better performance than Ubuntu.
Download links and more details on HSE 1.8 at GitHub.
HSE amounts to a well optimized key-value store database geared for high performance solid-state drives and persistent memory. Micron's original announcement talked of HSE providing as much as doubling the throughput and improving read latencies by around four times. With HSE is also a modified MongoDB implementation as a real-world reference implementation. Besides MongoDB, Micron hopes HSE will see use for various NoSQL, SDS, and big data use-cases along with other verticals.
This HSE storage engine depends upon an mpool object storage media pool that is implemented as a Linux kernel module. Mpool interfaces with SSDs or other persistent memory storage directly, bypassing file-systems and other overhead while also supporting replication across classes of memory and other storage features. So far we haven't seen any work on Micron sending out the HSE kernel bits / Mpool for upstream review in hopes of getting it mainlined in the Linux kernel, but we'll see what more comes out in the months ahead.
With this week's HSE 1.8 release there is support for media class pinning and tiering policies, value compression settings, configurable ingest throttling, and Ubuntu 18.04.4 support. This is the first release seeing official Ubuntu support as previously Micron was focused on RHEL/CentOS support. Micron does note though that HSE on RHEL is seeing around 5~30% better performance than Ubuntu.
Download links and more details on HSE 1.8 at GitHub.
1 Comment