@whuber we're a third party logistics providing these estimates to varioius costumers based off our historical data. Because these impact 100s of costomers across hundreds of industries - the impacts are broad. It could mean that a costumer doesn't plan enough empty cars to pick up a shipment of wood because they though they were getting there in 5 days but it took 7. It could mean that somebody doesn't ship enough oil because they thought in transit time would be 10 days but its 15.
When I say lane, this is rail data - so trips on a lane typically go from the same origin to the same destination along the same route - so we can precalculate averages. That origin to destination is a lane.
Its easy to say abs(ActualTime - ETA) is accuracy for a single ETA (how far it is from what actually happened). But how to roll that up to a meaningful measure for a trip, and then for all trips on that lane is where I'm having difficulty making a jump.
@gung So. I have a bunch of ETA (estimated time to arrival) data for logistics stuff. Over the course of a trip, the ETA is constantly updated each time the product moves towards destination. By the time the trip is over, we know how long it actually took. So the goal is to find a good way to measure the ETA accuracy by lane. The times from each point it moves are pre-calculated ahead of time, then we add that time to where it currently is at that point to get the ETA.
I'm having trouble coming up with an appropriate statistical method or measure to describe how accurate (how close an estimate is to an actual measured event) - is that an appropriate SE site to ask for help with it? I have a bit of stats in my background but its been years.
I've run a DBCC CHECKDB and it found allocation errors. When I run it with REPAIR_ALLOW_DATA_LOSS - will it need to take the same amount of time to scan for errors, or will it attempt to use the last CHECKDB I ran and go faster?
(SQL Server 2000. Yes its ancient, no I can't upgrade.)
If I've run a DBCC CHECKDB and it found errors. When I run it with REPAIR_ALLOW_DATA_LOSS - will it need to take the same amount of time to scan for errors, or will it attempt to use the last CHECKDB I ran and go faster?
@AndriyM my understanding was the for postgres text is actually faster than a fixed length, because fixed length was essentially implemented as text+constraint behind the scenes. But i could be wrong.
@Phil would the SCN be different than TIME ? because the backup finished at 11pm, I set the TIME to midnight but thats when it complains. Running that now.
@Phil the comments section there says you have to use an older backup and restore. So that means recent backups are screwed if you add a datafile? That doesn't sound right.
I've duplicated our production to our UAT environment several times in the past but due to various reason, this time has been a serious PITA.
Oracle 11.2.0.3.
We have a recovery catalog setup. This is the script thats run on the UAT box:
connect catalog rman/cat@rc
connect auxiliary /
run {...
So instead of checking whether each car is in the polygon, we result_cache the result of polygon and location sdo_contains, so that we can satisfy all 200k cars from a cache of 5k locations.
@JackDouglas well. I've logged a ticket. This is actually fairly cool functionality that we're trying to use the result cache for. I work in logistics and we're allowing our customers to draw on a map to support geographic searches on their cars. They might have 200k cars but they'll only be sitting in 5k distinct locations.
(this is kind of a shopping question,so I didn't want to post on main site) I can't find something that I figured would be not hard.....so either there's a reason technically or I'm not looking right. I need to read a series of dip switches (4 total) with 9 switches into a microcontroller. So i figured a parallel in/serial out shift register...but I can't find any that take 9 inputs. I could do 8 in and just run an extra register, but that sounds like a PITA. Do they not make one or..?
Ten fold
Ten fold
I've run a DBCC CHECKDB and it found allocation errors. When I run it with REPAIR_ALLOW_DATA_LOSS - will it need to take the same amount of time to scan for errors, or will it attempt to use the last CHECKDB I ran and go faster?
(SQL Server 2000. Yes its ancient, no I can't upgrade.)
