dmj

Software patents aren't really a think in UK law, having said that, it's complicated. Even so, I'm 99% sure you couldn't patent a database schema like this. It's going to come down to copyright, and I can't see there is a case there either. IANAL edit snip from the wikipedia

It's much more common for certain web companies to charge per user per month, so for this sort of thing we'd use a shared account: hence the bitwarden recommendation for MFA codes shared per team

It's not the standard table that would be copyrighted, its the schema: the relationship between those tables.

IIRC you have to increase the 'bytes per cluster' to one that will fit the larger disk. That's probably going to require reformatting the disk though, and if you are going to that much trouble you might as well use zfs.

The backup file would have the necessary SQL code to reproduce the database, I think that's what they are claiming. I can't see how their legal argument would work, the point of copyright on the schema is so that somebody cannot make a derivative work - but nobody is actually doing that so their tactic is just to scare schools rather than that they have an enforceable legal route to prevent its use.

We use bitwarden when we need to share MFA codes

More info here https://schoolsweek.co.uk/competition-watchdog-pulled-into-mis-dispute-again/

I assumed it was the CEO of Arbor as well with all that talk of ethics.

The clue is in the error message using a sleep is generally a bad way to do this, ideally the 2n'd command should try and read the output and check the job is completed. I'm not good enough with ps to do that, so try increasing the sleep.

The 13" is for sale already. Starts at £849 with a Ryzen 7640

These are the easiest to fix laptops: https://frame.work/gb/en For repairability scores: https://www.ifixit.com/repairability/laptop-scores

I know it's the default, but isn't that just a weird place to put a database. I'd feel terribly dirty leaving it there.

I agree, loads of what we do is from scratch. Some aren't so comfortable with that, which is understandable if they doubt their own ability to make a reliable system or don't have the support of their school to do that.

This is fair, I don't see a reason why something like home-assistant shouldn't scale. You'll need things like High Availability MQQT and that's propriety so even homemade stuff you'll end up spending something on it if you do it properly.

^I'd do that, but most here don't like the homebuild stuff and only want vendor supported systems. I guess it depends on you technical aptitude and the schools appetite for risk if you leave.

That's how I understood it, and why WIFI-7 boards have multiple 10GBE connectors: https://www.banana-pi.org/en/bananapi-router/155.html

Wifi 7 support really only landed in the linux kernel in september 2023. It's going to take vendors a while to build this in. Even OpenWRT (which many wifi routers base their images from) is only beginning to support it in dev builds. If you want to get ahead of the game on this you'll need to build your own AP's and manage them with openwisp.

Can't you just replace the batteries with APC compatible? or is there a feature that you are missing?

Moving databases is difficult at the best of times, adding document storage to this massively increases the issues you're likely to encounter just by the increased db size alone. As noted, storage costs on db's is expensive and means recovery isn't straightforward. We put a link into the database that points to a bucket, the application can then either render the document itself or just point to the (CDN backed) bucket. The link to the bucket is through a url map and is dynamic so you can easily move buckets around without worrying about updating the database. The bucket just scales with load automatically, independently from the database (further reducing costs)

The more I read about Bromcom infrastructure, the longer the barge pole gets. Storing documents in a database is just wacko for a SaaS provider.

I think this is part of the problem, they shouldn't have to constantly 'do' anything - the infrastructure should scale with usage, it shouldn't be anything they even notice.

"extensive testing"

I see. I was quoting bucket prices, access time isn't determined by the type of disk they just scale as needed

Google storage pricing is 1.8p per GB per month or 2.1p per GB per month for multi region. It's even cheaper for coldline or archive I knew azure was more expensive, but five times the price? Are you sure those figures are accurate? @localzuk

Nice. You can get similar functionality in CI/CD tooling. We're using Jenkins for this sort of thing.