We started Solid State Storage in January of 2000, with the goal to develop a smart solid state storage card that would plug into a computer's PCI slot and provide ultra high IOS and throughput storage for high demand systems. Our initial goals were to cram 4 GB of battery-backed DRAM on the device and use a 64-bit MIPS processor to run specialized embedded software that would allow the card to interface with disk-based storage, essentially making the card a fancy fully-associative persistent cache. The card would monitor usage patterns and lazily flush under-utilized data back through the PCI bus to a third party disk array, as well as perform read-aheads to load data blocks before they are needed. The other advantage of the card would be that it could be made to appear to be much larger in storage capacity than the amount of installed DRAM, by virtue of its integration with disk.

As we were developing the software for this device, it became clear that the software could potentially stand on its own as a generic memory manager and would provide a nice layer of functionality on top of the host OS's virtual memory manager (VMM). We also realized that most of what we wanted this card to accomplish could be achieved through this memory manager working in conjunction with the typical unmanaged solid state storage device that simply mimics a disk drive, which already existed at the time. By foregoing a hardware approach, however, we would lose one principal advantage thereof, which is that it would work with all existing software. The memory manager, on the other hand, would require the recompilation and re-linking of applications that would utilize it.

Fundamentally, information technology is about algorithms, and a common debate is what to implement in software versus hardware. Based on a combination of factors including feasibility and economics we chose the software route. Approximately four years later, we released our first product, the Taiyen Memory Manager, the culmination of thousands of hours of research, development, and testing. This system is loaded with innovation, and we believe will produce at least two patents, which we a currently securing.

Going forward, a key technical goal is to complete development of Objedata, which is a powerful transactional object DBMS that rides on top of the Taiyen system. While object-based at its core, it will also allow for relational access via SQL as well as SOAP, XML, and its core API. We believe the competitive advantage of this system will be three-fold: One, operational and transactional speed will be exceptional by virtue of its Taiyen kernel, which provides enterprise performance on budget platforms. Two, it will be vastly more flexible in terms of data models than competing solutions, won't have to be endlessly tuned to achieve optimal performance, and won't require full-time administrators. And three, like the Taiyen system, static libraries will be distributed allowing users to embed Objedata code directly into their applications.

We are also moving ahead with the development of our first hardware product, a persistent solid state storage device, which will offer the features of our initial concept but via a much more clever design. On the horizon, we intend to develop other innovative storage technologies that we believe will meet and exceed the needs of ever more demanding information systems.

Morgan Weinberg,
President