When you work in a StorNext collaborative environment every day you can take for granted what goes on behind the scenes to make it cross-platform, simultaneous workflows on a shared storage possible. It turns out that managing users all wanting to access to files on a shared resource is one of the hardest computer challenges – how can you manage user access and file transfers to the shared storage seamlessly?
A network shared storage approach such as NFS was an early solution – as is Network Accessible Storage (NAS) over ethernet. While its true that Ethernet accessibility makes shared files easily accessible, moving these files via Ethernet means that these files are delivered in Ethernet packets, are not guaranteed to be delivered in order, and subject to network interference, other Ethernet traffic bottlenecks and more.
For more direct, high-performance, more controlled group access to shared folders requires a Storage Area Network or SAN. StorNext has emerged as the preferred choice for high-speed, collaborative shared storage – meaning that files are delivered over a high-speed, low latency network, that groups of users all see and share files on storage, and that there is no ‘file copy’ penalty waiting for a user to finish work on a file and copy back to shared storage before others can continue working on it.
The way StorNext delivers this ‘magic’ is by ensuring that a highly redundant ‘controller’ for your StorNext environment manages the users (which client workstations can connect to shared storage volumes), permissions (which user can see which shared storage volumes), and file traffic (allowing clients to move files to/from shared storage). This StorNext Metadata Controller tracks information about each file and folder on the shared storage that it manages, which can approach millions of instances in large, busy environments.
With the recent release of StorNext 5, this critical area of storing and managing this ‘metadata’ or, ‘information about the files in the shared storage environment’ received a dramatic performance improvement thanks to three main innovations:
Redesigned Metadata Layout and New Algorithms
StorNext 5 lays metadata information on disk far more efficiently thanks to a new b+tree structure on disk. This operation is fully multithreaded and takes advantage of modern, multi-threading CPUs, and allocates free space intelligently.
Redesigned Metadata Journal
StorNext 5’s Metadata information’s Journaling feature keeps track of file system changes and allows for rapid resumption of file operations, this too takes full advantage of multi-threaded access and is optimized for SSD usage.
Compressed Metadata Cache
The caching of this Metadata is so efficient that it can be compressed and recompressed on the fly and in most cases keep almost all of the needed metadata cache in memory for the fastest possible access.
To test these improvements, the StorNext 5 engineering team tested up to millions and even billions of files (and let me tell you it takes a long time to write out and move billions of files). All of this means that StorNext 5 file operations feel more ‘liquid’ and natural, even for extraordinarily large environments. So while every environment benefits immediately from upgrading to StorNext 5, you will appreciate the performance even more as your environment grows and evolves.
StorNext 5 is built into the latest StorNext 5 Metadata Appliances such as the StorNext M440, StorNext M445SSD with SSD drives, the StorNext M660 and each of these are the heart of the very successful StorNext Pro Solutions. Upgrading to StorNext 5 is highly recommended for all customers – be sure to contact your StorNext Reseller or Integrator or Quantum representative to go over your upgrade options!
Want to make the move to StorNext 5? It’s easy to upgrade and we’ll even give you a $5,000 trade-up credit too.