A Honeycomb Belongs in a Garden: Data Storage meets the world

The Honeycomb software group had a "work week" two weeks ago out in the San Francisco office. I and several of my East Coast colleagues traveled there for a week of brainstorming sessions and design meetings. It was cool!

One theme that showed up during the meetings is a transition in the "sweet spot" the Honeycomb group is aiming for. Transitions often (maybe always) call for appreciating and then letting go of the past. So the shift in the sweet spot gives me a chance to acknowledge and thank all the people who got Honeycomb this far. I am a relative newcomer here. This is my report on what I have seen after having arrived here relatively late in the game.

Honeycomb (now soon to reach general availability as the Sun StorageTek 5800 System) grew up focused around a particular sweet spot: take cheap, commodity hardware (disks, cpu, switch) and then pull it together with cutting edge software. Start by making it reliable, scalable, and affordable. Focus on the archive storage area, because that is where people need super-cheap super-reliable smart storage. With the "excess" computing power inherent in the design, bring software functions (such as metadata search and programmability) right into the storage server where it can be done more efficiently. And then embody those good ideas in a real, shipping product. (Even from its earliest days, the Honeycomb architecture was designed not just to research good ideas but to ship them.)

It is all the people who came before who got us here, right from the earliest inception of Honeycomb as the gleam of an idea in Sun Labs, all the way through the Sun Chairman's Award, and then through all of what it took to shift it from a research vehicle to a shipping Sun product. This took lots of people: marketing people, software designers, hardware designers, manufacturing experts, test designers, customer engineers, and even customers! This has taken not just the vision and clarity of the original design, and not just the work of talented engineers of many kinds. It also takes a LOT of testing to identify potential weak spots in all the technologies we draw on and then work around those weak points so the customer rarely has to see them.

Besides acknowledging and letting go of the past, the other side of a transition is pointing toward the future. For Honeycomb and the 5800 System that embodies it, the game is changing. The new game is all about making that "sweet spot" available. I don't just mean to ship it out; that is happening now and will happen more once we reach general availability. I mean to do the hard work of listening to our customers and potential customers -- and listening to the rest of Sun -- to hear what people really need to put this great new technology to work in their environments. And that simple statement requires a new focus on a new "sweet spot", which I'll have more to say about in future posts.

========================

What is Honeycomb? And what has been accomplished so far? Here are two quick links to get you started. First, here is a recent Sun EduConnection newsletter on Intelligent Storage Comes to Campus: Sun StorageTek 5800 delivers a petabyte of storage with brains and brawn. Second, many pieces of the Honeycomb software are planned to be released as Open Source Software via theHoneycomb Page at the OpenSolaris software storage portal.

(For you non-geeks out there, open source means that we make the software publicly available so that every one can read it and build on it -- everyone from prospective customers to third-party application developers to even, yes, our active competitors. In today's Internet, releasing your crown jewels to open source is just good business sense: it helps your product build the momentum it needs to stand out above the crowd. For an archive product, where people want to trust their data to it for decades at a time, it is essential.)