Archives

Date
  • 01
  • 02
  • 03
  • 04
  • 05
  • 06
  • 07
  • 08
  • 09
  • 10
  • 11
  • 12
  • 13
  • 14
  • 15
  • 16
  • 17
  • 18
  • 19
  • 20
  • 21
  • 22
  • 23
  • 24
  • 25
  • 26
  • 27
  • 28
  • 29
  • 30
  • 31

Vislab does successful test on real streets

A nice result for Vislab of Parma, Italy. They have completed a trial run on public roads using their mostly vision-based driving system. You can see a report on the Vislab site for full details. The run included urban, rural and highway streets. While the press release tries to make a big point that they did this with a vacant driver’s seat, the video shows a safety driver in that seat at all times, so it’s not clear how the test was done. They indicate that the passenger had an emergency brake, and a chase car had a remote shutoff as well.

The Vislab car uses a LIDAR for forward obstacle detection, but their main thrust is the use of cameras. An FPGA-based stereo system is able to build point clouds from the two cameras. Driving appears to have been done in noonday sunlight. (This is easy in terms of seeing things but hard in terms of the harsh shadows.)

The article puts a focus on how the cameras are cheaper and less obtrusive. I continue to believe that is not particularly interesting — lasers will get cheaper and smaller, and what people want here is the best technology in the early adopter stages, not the cheapest. In addition, they will want it to look unusual. Cheaper and hidden are good goals once the cars have been deployed for 5-10 years.

This does not diminish the milestone of their success, making the drive with this sensor set and in these conditions.

RAID, backyard backup and the future of backup

Had my second RAID failure last week. In the end, things were OK but the reality is that many RAID implementations are much more fragile than they should be. Write failures on a drive caused the system to hang. Hard reset caused the RAID to be marked dirty, which mean it would not boot until falsely marked clean (and a few other hoops,) leaving it with some minor filesystem damage that was reparable. Still, I believe that a proper RAID-like system should have as its maxim that the user is never worse off because they built a RAID than if they had not done so. This is not true today, both due to fragility of systems, and the issues I have outlined before with deliberately replacing a disk in a RAID, where it does not make use of the still-good but aging old disk when rebuilding the replacement.

A few years ago I outlined a plan for disks to come as two-packs for easy, automatic RAID because disks are so cheap that everybody should be doing it. The two-pack would have two SATA ports on it, but if you only plugged in one, it would look like a single disk, and be a RAID-1 inside. If you gave it a special command, it could look like other things, including a RAID-0, or two drives, or a JBOD concatenation. If you plugged into the second port it would look like two disks, with the RAID done elsewhere.

I still want this, but RAID is not enough. It doesn’t save you from file deletion, or destruction of the entire system. The obvious future trend is network backup, which is both backup and offsite. The continuing issue with network backup is that some people (most notably photographers and videographers) generate huge amounts of data. I can come back from a weekend with 16gb of new photos, and that’s a long slog over DSL with limited upstream for network backup. To work well, network backup also needs to understand all databases, as a common database file might be gigabytes and change every time there is a minor update to a database record. (Some block-level incrementalism can work here if the database is not directly understood.)

Network backup is also something that should be automatic. There are already peer-to-peer network backups, that make use of the disks of friends or strangers (encrypted of course) but it would be nice if this could “just happen” when any freshly installed computer unless you turn it off. The user must keep the key stored somewhere safe, which is not zero-UI, though if all they want is to handle file-deletion and rollback they can get away without it.

Another option that might be interesting would be the outdoor NAS. Many people now like to use NAS boxes over gigabit networks. This is not as fast as SATA with a flash drive, or RAID, or even modern spinning disk, but it’s fast enough for many applications.

An interesting approach would be a NAS designed to be placed outdoors, away from the house, such as in the back corner of a yard, so that it would survive a fire or earthquake. The box would be waterproof and modestly fireproof, but ideally it is located somewhere a fire is unlikely to reach. It could either be powered by power-over-ethernet or could have its own power and even use WIFI (in which case it is only suitable for backup, not as a live NAS.)

This semi-offsite backup would be fast and cheap (network storage tends to be much more expensive than local drives.) It would be encrypted, of course, so that nobody can steal your data. Encryption would be done in the clients, not the NAS, so even somebody who taps the outside wire would get nothing.

This semi-offsite backup could be used in combination with network backup. Large files and new files would be immediately sent to the backyard backup. The most important files could then go to network backup, or all of them, just much more slowly.

A backyard backup could also be shared by neighbours, especially on wifi, which might make it quite cost effective. Due to encryption, nobody could access their neighbour’s data.

If neighbours are going to cooperate, this can also be built by just sharing servers or NAS boxes in 2 or more houses. This provides decent protection and avoids having to be outside, but there is the risk that some fires burn down multiple houses depending on the configuration.

A backyard backup would be so fast that many would reverse what I said above, and have no need for RAID. Files would be mirrored to the backyard backup within seconds or minutes. RAID would only be needed for those who need to have systems that won’t even burp in a disk failure (which is a rare need in the home) or which must not lose even a few minutes of data.