Brian's Brain

Having just filed my March 15^th feature article on imaging, my thoughts now turn to the next few Dipert-slotted writeups on EDN's editorial calendar. Right now, impermanence willing, I've got cover stories in the May 10^th and August 2^nd issues and a feature article in the September 27 edition. For May 10, I was originally planning on doing some hand-on testing of various direct-attached storage interfaces; here's the summary blurb I wrote up for the marketing folks:

Peruse your computer's HDD and you'll likely find that the bulk of the consumed storage space is comprised of multimedia files, thereby pointing out the attractiveness of external storage to supplement and expand on the capacity you've built into your next system design (PC or embedded). This multimedia information must often also be accessed (read and write) at fast rates, driving home the importance of high speed interconnect, with eSATA currently representing (at least on paper) the performance pinnacle. But does it make sense to add an eSATA mass storage connector (and corresponding controller IC and software stack) to the already-burgeoning plug plethora in the back of your next system design? Alternatively, should you just upgrade your existing IEEE 1394a (aka Firewire400) interface to IEEE 1394b (aka Firewire800)? Or, if you currently employ only USB, will a USB2 upgrade suffice to meet your mass storage speed needs?

In EDN's May 10 issue, Senior Technical Editor Brian Dipert puts all of these direct-attached storage interface options (USB2, 1394a and b, and 1.5 and 3 Mbps eSATA) through a hands-on torture test to assess their respective strengths and shortcomings. To this end, he'll employ both synthetic and real-life application benchmarks, on one-to-five drive subsystems, and in various concatenated JBOD (just-a-bunch-of-disk) and RAID (0, 1, 0+1, 1+0, and 5) configurations. He'll also identify the system-level bottleneck(s) to higher performance; the storage subsystems themselves (including the drives contained within them, which he'll vary to assess their impact), the to-system interface technology, the within-a-system interface bridge (PCI, PCI-X, and/or 1-to-multi lane PCI Express), and/or the system software stack.

But now I'm having second thoughts. I have two other in-depth hands-on projects planned for this year:

• File transfer (specifically multimedia stream) benchmarking in my 'technology torture chamber' of various powerline (HomePlug 1.0 and Intellon's proprietary 1.0 Turbo, HomePlug AV, Universal Powerline Association aka DS2, and Consumer Electronics Powerline Communication Alliance aka Panasonic's HD-PLC) and wireless (802.11a, 802.11g and, if I deem it stable enough for my results to be editorially meaningful for a long-enough post-publication duration, draft 802.11n) technologies, versus each other and versus conventional 100BASE-TX (100 Mbps) and 1000BASE-T (1 Gbps) wired Ethernet, and
• A subjective and, if the analysis tools are up to snuff, objective analysis of various high definition video codecs (WMV9 and SMTPE VC-1, MPEG-4 AVC aka MPEG-4 Part 10 aka H.264, DivX, MPEG-2, On2's VP variants, etc).

I'm waffling a bit on which of these three projects I should tackle first, so I thought I'd survey you all to see which one(s) you thought had the highest priority. Let me know your thoughts, either publicly in the comments section of this post or privately via email. Thanks in advance, and have a good weekend!