Peer Software’s PeerGFS, maintaining choice - you don’t really want another file system, you want a heterogeneous file service

The world is full of innovative storage solutions using great new technology allowing you to take full advantage of modern data developments. That’s good, but what do you do with your legacy storage, that may be “good enough” to meet your needs? Rip and replace is usually considered a bit risky, and it is best avoided if you can. So, you will likely end up with both your legacy storage systems and the innovative new ones – which means added complexity.

This is where Peer Software comes in, with its real-time distributed file synchronisation service, PeerGFS. Note “service”, it is not a new file system, it is more than that.

What it does is to seamlessly integrate existing storage systems. Users have fast (low latency) local access to their data while the PeerGFS ensures that data integrity is protected and provides high availability. Most importantly, installing it won’t disrupt your business – it goes on top of existing on-premises and cloud storage solutions and promises to scale well. New storage systems can be added equally transparently. It can help you to move data between locations and technologies, but often you won’t need to. PeerGFS lets you keep data local to users and their application/computing environment whilst still maintaining low latency, cached, access to their data from anywhere in the organisation.

Its key use cases include Edge Computing, load balancing across data centres, and replication to S3-compatible Object Storage on the Cloud for use in advanced analytics (and, presumably, for training AI models). It targets multiple-vendor enterprises and integrates with existing storage APIs. It runs lightweight Peer Agents on the supported nodes and minimises network traffic by only replicating changes, not whole files. High availability and automatic failover/failback can be provided over edge, data-centre and cloud systems. Integrity is supported via global file locking, which always makes me a bit nervous (it’s great for integrity, but timely release of locks is vital for performance), although it is simple and reliable. I suppose I can think of workload-related ways in which this might go wrong, but I am assured this isn’t a problem in practice. PeerGFS also offers management and observability functions so it should highlight badly designed, badly-behaved, “chatty” systems before they become a serious problem, as well as monitoring malware and other malicious behaviours. There are probably use cases that don’t suit PeerGFS but none that struck me as particularly obvious, as long as systems are designed sensibly.

Peer Software is headquartered in Virginia, USA and has an EMEA HQ in Ulm, Germany. I was talking to Peer Software’s CEO, Jimmy Tam, who obviously knows his stuff. Jimmy is proud of the company’s longevity (it started in 1993) and that it is profitable without being dependent on venture capital. It is privately owned, so perhaps not as high profile as Jimmy would like, but it has an impressive customer list. I also wonder whether Peer Software tends to be called in when a file system vendor can’t fully deliver in a hybrid environment but wants to use as much of its own file system as it reasonably can, and that isn’t a situation that the file system vendor will make a lot of noise about. Peer Software has technology alliances with Autodesk, Citrix, Liquidware, Microsoft, AWS, Wasabi, NetApp, Dell, and Nutanix.

I don’t have in-depth practical experience with PeerGFS but I like what I’ve seen. I think that it helps to maintain customer choice, by providing a single, coherent, view at the logical level of disparate physical storage technologies underneath. This sort of thing matters to the Mutable Organisation, in a state of continual evolution, which can’t afford to be tied to particular technologies.