Wendell went to Penguicon 2016 and had a lot of fun. This is a quick recap of some of the fun stuff at Penguicon.
More about the avenger, EV http://theavenger-ev.com/
More about Lucida/Sirirus AI Assistant:
http://sirius.clarity-lab.org/sirius-suite/
Lansing Makers Group:
https://www.lansingmakersnetwork.org/
All Hands Active Charity:
http://www.allhandsactive.org/
Thanks for a look at Penguicon 2016 looks interesting. Do you have an video of the panels you went to they would be nice to see.
Have a day
That open source AI project you mentioned at the end looks interesting. Future video maybe? Might be cool to get some raspberry pi projects or something out of it. (using an external home server for data)
Edit: I just looked at their Github page and this looks amazing. A tutorial video on TekLinux would be amazing. Doesn't need to be anything too in depth or fancy, but a starting point would be cool. Unfortunately I'm on an extremely limited internet connection and can't download anything. I'll see if I can leach from a friend.
Yes please have the Artemis people come to LAN Syndicate 2017 if possible, that would be awesome!
Somebody fix the keystoning on that monitor ahhhh.. But yes to the touch screen Artemis.
With enough interest we could have multiple crews playing together, or against each other.
You know, it would be really interesting if you could marry the Lucida/Sirirus AI Assistant and the Artemis ship simulator to be able to ask the ships computer questions.
Great recap, I wish you could expand on the ZFS part a little.
What do they actually use? raid 5? raid 10?
Except for IOPs, I think the big players are looking beyond ZFS - for redundancy at the rack or row level.
Their servers run as installed for the useful lifespan.
The problem of how to grow a VDEV SMBs run into,
stuff mid-range dumb RAID like PERC can do, does not exist.
Pick the right tool for the job.
That being said, the resilver time, and performance hit,
for these new drives is cause for concern.
A lot of people had really impressive workloads. So the new hot thing is PCIe interconnects and some of the big datacenter folks have enclosures that attached to the storage controllers directly via pci-exrpress. 48 drive cabinets with a zillion channel LSI HBA + local PCIe SSD storage (usually intel NVMe storage for l2arc/zil). This was just one setup that got my interest. So their cache + drives were local. That 48 drive setup might have several vdevs.. most people there did not like more than 12 drives on a vdev.
one of the 500tb+ setups (which was a mirrored vdev iirc so like 1200tb raw space) all in one rack with 4 heads) had a scrub time of something like.. 8 hours? which is pretty awesome.
I have been to that gun range, it's quite nice.
Certainly better than the usual "yooper" shenanigans of going out to a hunting camp in the woods and just letting loose; although you were a good two-hundred miles away from the Upper Peninsula.
Great recap from Wendell. I am now also looking at the Sirius/Lucida AI project. I have have not had a look at this kind of system before. Already with the speech ai and the wiki, maybe it is possible to roll your own Echo like system as this is not sold/shipped to Sweden. (Not that I expect it to be easier to roll my own - more like beating the sales strategy, not be connected to their cloud and it seems more fun to do.)
Definitely great recap! Certainly the workload when I have write my paper one after another increases drastically so that the storage control directly via pci-exrpress is only but beneficial to have full time access to the data.