Having information stored on multiple servers that are not connected will often leave your files scattered and misplaced. By clustering your servers, it enables you to have all your data in a single namespace.
Clustering has often been perceived as something complex and expensive, but technology advances have made it easier to build a high availability cluster that delivers petabytes of raw data in an open storage platform.
Our cluster operates by using high availability software running on redundant servers, which provides continued service if any system components fail. The cluster also offers easy scalability, meaning if at any time you require more storage or performance – just add another server.
Using multiple Storinators, a system called GlusterFS and the ZFS file system, we built a high performance, high availability cluster solution that has tremendous capacity and is easily scalable.
Here’s an overview of the cluster built be 45 Drives, showing what was used and why:
- Minimum of two Storinators (open storage platform)
- ZFS, for device-level redundancy.
- GlusterFS, for server-level redundancy.
The ZFS layer provides HDD redundancy through the support of various parity RAID levels (Z1,Z2,Z3) and allows users to benefit from the many features of ZFS such as compression, snapshots and SSD caching with L2ARC and ZIL. The GlusterFS layer allows for scalability beyond one server and protects against server level failure by providing a highly-available, high-performance storage volume. Combining the two on an open storage platform like our Storinator allows for incredibly safe, reliable, large, and really powerful storage volume.
Storinator Cluster Solution Features:
- Fully scalable storage (we built 1 PB but you can configure what you need)
- Simple and fast configuration, supported by 45 drives.
- Read/write in excess of 3 gigabytes/second
- High availability
- ZFS file system based, delivering fast, secure, highly scalable storage
- Hard drive locking mechanism to prevent excess vibration
- High efficient “85%” server grade redundant power supply with PMbus remote monitoring.