As much as I hope Sun’s Project Honeycomb will be a success, there doesn’t appear to be a single element of it that is a first. Like most high-tech products it picks from threads of innovation that have sometimes been hanging around for decades. So when Sun claims:

Honeycomb is a scalable, extremely dense, highly reliable storage system, consisting of intelligent software components integrated with purpose-built commodity server hardware. The integration of storage with metadata and search systems into one easy to manage, high reliability, scalable appliance is unique among storage systems available today.

They go a bit over the top. For example, Seattle’s Isilon Systems claim they’ve shipped over 2,000 clustered storage system in the last two years. Isilon is very focused on the media business, but AFAIK there is nothing in their system that confines them to that niche. The founders knew media serving from working at Real Networks and stuck to what they knew. In fact they’ve recently been expanding into geo-physical markets.

Metadata extensions are also interesting as well. Apple added some to their OS X file system recently. They are also at issue in the ILM marketecture. As I’ve mentioned in my ILM critiques (“Gee, “Users Cite ILM Shortfalls” – Maybe ILM IS Bunk” and “ILM is Bunk”) IT doesn’t own the data so their leverage over end-users to get them to classify their data is nil. Without classification it is difficult to make the case for ILM — and even with classification it still isn’t easy. Yet there are other potential reasons to add metadata to files.

That is what a little Silicon Valley startup named Abrevity claims to do. Except they do for all the NAS boxes on a network and they use a “free entity” database purpose built to handle metadata. Once installed a “. . . lightweight utility scans Windows and UNIX network or desktop storage systems and extracts target information” and sticks it in the database. They also say they can add additional metadata to files in the database — effectively extending metadata just as Sun says Honeycomb can do. Further, “. . . unlike traditional enterprise search engines, which require up to 300% storage overhead for indexes and days or weeks to index data, FileData Manager offers less than 3% overhead and order of magnitude faster indexing.”

I have no idea if any of this is true, but you can download their software and try it. Let me know how it goes.

What I do know is that Abrevity picked the least liked pricing scheme in storage software: capacity-based pricing. Sure, EMC does it, but little storage startups should know better.

Ideally, Sun’s marketing team is looking at these companies and others to figure out what these guys have figured out already and what applies to Honeycomb. You aren’t the first and that is a very good thing.