I mentioned that the Kurobox would get its own post, so here goes. It’s shaping up to be kinda long-winded, and I apologize for that.

Currently, I have a fileserver (named “pangloss” in a fit of pure optimism) based on Gentoo built from old whitebox components. It’s about a 1GHz i686 processor (don’t remember exactly, it’s been a long long time since i’ve dealt with CPU internals), a 40GB parallel ATA HD housing the root partition and swap space, and a 400GB PATA HD housing /var and /usr but mostly /home.

Pangloss hosts backups for both my laptop and Fred’s desktop, as well as some common files. It listens for SSH from the outside world as well, so that we can access it from work or the road, and runs a couple services for the local network like DHCP, DNS caching, and an NFS-available portage tree.

Pangloss is also a full-sized tower that has a gajillion fans and consumes a bunch of electricity. I have yet to approach full CPU load or memory capacity without the help of the BOINC client. The last time I rebuilt it, I had it at my then-office, and whenever it was on my co-workers jokingly called it the “jet engine”. The last time the house electricity went out, pangloss drained the UPS to its last little LED within 20 minutes, but once I shut it off the UPS lasted another 20 minutes or so. These are non-optimal.

So, in the interests of size, noise, power consumption, and that geeky itch to adjust things incessantly, I’m itching to re-do the local file server in a much smaller, quieter, more efficient, and more challenging way.

Enter the Kurobox. The Kurobox is, essentially, a Linux-based Buffalo NAS appliance without the built-in hard drive. There are two models currently on the market, the HG/WR with a PowerPC chip and a PATA interface, and the Pro with an ARM9 chip and SATA. Each one can hold precisely one 3.5″ hard drive, but there’s no real limit on the capacity of that one drive.

(Before I continue: I know there is a purist out there grumbling about how I could just build my own network file server that’s more powerful and can do RAID and can make julienne fries or whatever. Great, but I’ve already taken that approach [cf. earlier in this very post] and would now like to adjust priorities away from sheer computing power.)

So here’s the list of advantages for the two Kurobox models as they apply to my situation. They’re equivalent on RAM and network speeds, and both hold only one drive but that’s OK as I’ll be finding a duplicity buddy for backups.

HG/WR
• Cheaper for hardware if purchased from the US distributor
• Already have the drive!
• I’m already familiar with the PPC architecture
• More software packages are ready to compile on/for PPC processors than ARM processors
• The PPC architecture is generally more powerful
• All sleek and pretty

Pro
• SATA has smaller cables, which means better airflow, which means less heat
• SATA is faster than PATA
• SATA doesn’t use the PCI bus, so the drive won’t compete for resources with network activity or USB
• SATA is more future-proof than PATA
• The ARM architecture is generally more power-efficient than PPC
• This particular ARM chip is faster than the PPC in the HG/WR (400MHz vs 266 MHz)
• More ports for future expansion
• I can leave the existing data where it is while setting up partitions et al. on the Kurobox Pro. This is a very big thing.
• For now, the US distributor has the Kurobox Pro in stock in Texas for $169; shipping starts at $9. At the moment, the HG/WR is only available online from a shop overseas, with both product and shipping costs based on the Euro and converted to USD and therefore lots more than what they’d be if it were also in stock in Texas.

So the balance favors the Pro, for now. Anything else I should consider?