psydii

Trad three hosts, fibre channel to a low-end (typically a rebranded quantum) SAN, with *lots* of disks, groups up and raid6 with a hot spare per 8 or so physical disks. Hosts may or may not have a raid1 boot SSD. Extremely simple and extremely easy to get 3rd party support. They typically can get 8 years of 24/7 manufacturer support/warranty too. Got 9 out of the last one. Run basically this same set up since 2008. Occasionally I look at some sort of hyper-converged roll-your-own thing but it never seems worth the hassle. If it throws a fit on a Sunday afternoon, I want to pick up the phone and have an engineer on site by 6pm. If it loses a disk I want one out with a courier before I’ve even noticed the alarm. Multiple RAID 6 plus hot spare because I want to be able to leave the country for a couple weeks a year and be (almost) certain that even three disk failures won’t mean I need to stay abroad for the rest of my life!

I shall just observe the following: The site team office here is the second most secure room in the building. I have keys to almost everywhere but not the site office or the Exam ‘cupboard’. Ok, I have an opinion too. I think the DPO is a stakeholder in a security review. They should perhaps be a useful [emoji639]rd party who should be sharing their perspective on this?

I thought they do, but then I haven't read the spec in the last six months. On 2.5G to APs Can we just be clear - in an environment with 20+ devices on a single AP all active syncing cloud documents/printing etc you see greater than 1GB/s over wireless through a single AP? I've taken the lack of whitepapers boasting *real-world* throughput since early .11ac era as indicative that throughput still collapses under load, but perhaps to a higher value than it used to?

Not recently. But a DfE funded installation here, and we have two routes from each network switch room back to the network core. Even then, it violated the standard because 1) ultimately all cables come together in a service shaft run of 30m, up and along the same route to get to the core. There was no other reasonable way to the core, even when we were remodelling the entire building. 2) cabling actually is not direct from switch room to core. Each hub room is physically linked Core ->Room1 ->Room2 ->Room3 ->Room4->Core. Each Switch room has a patched link (violating the current standard) round the building eventually getting back to the core, and another patched link going the other way round again to the core. There is also is a switch in each location that has a layer 2 link to the two rooms on either side of it. The top switch in the stack in each room has the 10Gb link to core, the bottom switch has the two 1gb backup links to neighbours and the second 10Gb to the core. Each link to core lands at the core on independent blades in the core so the infrastructure can survive *two* module or fibre failures at either end. Subsequently to the initial implementation we stood up a new building, and to provide service to that building a new external route was created from the core direct to the new building, and also from the new building to another network switch room, thus providing a physically diverse route that would survive should there be an incident that damaged the cabling in the aforementioned 30m run. At another school, with (again) a funding stream that is no longer available, when remodelling just the network (not the entire building) a couple of new links were put in to provide a similar level of resilience, but we were lucky because the cabling was so old there that it followed the route of the old old old thin ethernet/z-net single bus topology. A new fibre from either end to the core produced resilient/redundant topology cost effectively, if the existing fibres had been is a hub and spoke arrangement we would have needed many many more new runs to provide the resiliency described by the current standards. l

Just out of interest, does anybody see more than 1Gb/s the Ethernet side of APs while they have 20+ client devices? I know that you can get high throughput on a single client, but what about when you have 20/30+ clients - can APs these days get anywhere close to using more?

Apple for the longest while used Nuance technology under the hood, which is why their in-box dictation was so good. Microsoft purchased it (the company) recently so I don't know where that leaves anyone in the long term. With Microsoft's recent history of pulling features and re-releasing them with the AI/CoPilot branding as paid extras I am not at all hopeful.

Nebo.app might be what you are looking for. Cross-platform too.

Salamander can update this field, so if you have their system they might be able to help?

We have a Sharepoint site that acts as a root/anchor/hub for all other sites that are not departmental sites. Its content is managed by the Head Teacher's PA. We use GPO to set it as the default page for Staff.

12/13th gen i3 in the Photoshop/Illustrator suites has been absolutely fine. 256Gb nvme is too small in a OneDrive world unless you go Intune/shared device where magical things happen. (we've got 512 in all our media suites) Photoshop/Illustrator classes do need 16Gb ram, but everywhere else is more than fine with 8Gb. ML/AL/etc is allegedly the thing that kicks up specs, but ai cores are only on the higher end of Wintel CPUs so really that's moot. Probably more cost effective to go i3 equivalent, save between two to three hundred pounds and refresh after three years if it turns out AI-on-device is actually important for apps you need. If you go mid tier (i5 equivelent) you still don't get the AI cores, but you need to sweat out the assets to 5-7 (whisper it, even 10) to balance the budget, and then if AI really is important, you're 10 years behind the curve. Oh and 24" screens are fab, but watch out some are extremely fragile (Philips), while others (IIyama - same spec, same price) are quite robust.

We're at the "if we turn it off, we far from certain it will start again" stage. I've got a config set up to allow us to re-purpose a "modern" edge switch (a 2920!) to get us back up and running if the worst were to happen.

Maybe. There is a lot of investment locked up in ESS, this doesn't feel to me like Serco/Advance/CMIS/PRogresso just yet. I'm pretty sure the plan (for all the MIS providers) is to start raising prices prices once the market levels off. Bromcom have demonstrated it is possible to develop and maintain a profitable cloud MIS, so I'm pretty sure SIMS can be a going concern (with a tough management decisions having to be made) even if its market share collapses below 20%. If they can get their cloud MIS upto the "minimum-viable-product" standard for their remaining customers quickly enough they may even arrest their decline and begin to recoup their investment.

I think their cloud product looks (and feels) like it could actually be viable in 5-6 years.

Is it just me that remembers the last time the bbc said "no there isn't going to be a hurricane..."? 36 hours later I was picking bits of the computer suite out of an oak tree.

I will deny everything but we have similarly old and faithful box. I have **no idea** where this screen shot came from.

That seems odd? We have a lot of vlans (and acls and spanning tree), the core is the default gateway for everything (except the edge router/firewall system where it gets more complex), and CPU utilisation is in the 40-60% range, and this does not appreciably alter even when we are saturating our larger inter-switch connects with updates/images etc. What switch do you have for the core? Is it suitable for its role? Is its firmware up to date? Might you actually have some sort of loop on your hands?

"loads fast but might be wrong until such time as we can get the consolidated view of the data ready for you at which point we update and what the user sees completely changes" whatever the underlying technological reasons, is pretty poor ux and only in the cloud era did it start to effect everything. That said, AD has the same underlying problem for similar reasons, it can present you stale data until both the server you are connected has the latest update *and* you remember to manually refresh the view. Usually people aren't making massive changes in AD though, so its less likely to induce the "oh my god where has all the data gone" moments.

For what its worth, we have lots of stuff syncing to/from sims and have had no reported issues with deadlocks on CSWP4. We went to the summer release very late, so all the third party sync products were updated well in advance. We are a fairly large school, and registers are taken in a third party system and sync'd back to sims in near realtime-ish. The upgrade to CSWP4 was the smoothest I have ever seen, but that is probably because it came out so quickly after we'd completed the main upgrade. Hopefully patch 5 will be as easy.

At the risk of irritating DJM, this is mostly a cloud thing. It seems to stem from the fact the hyper-scale databases are only best-effort consistent and this mentality (somewhat necessary at global scales) has permeated into developers and product managers operating at the front-end customer level. As a small/medium/single geography entity we expect/need the UI (and reports) to show consistent, accurate, current data... but cloud principles mean that for the back end this is expensive/impossible. You see this a lot in mobile apps where the additional pressures of battery-life management and unreliable connectivity reinforce the approach of show stale data and update the view as-and-when the data becomes available.

I vaguely recall that the handwriting code was licenced from Evernote... given that all evernote staff were laid off and the new holding company jacked prices last year - this could be entirely outside of CleverTouch's control.

How do Bromcom and Arbor handle user actions that touch a lot of records? For example applying a new timetable mid-year, or compiling the census report? These are the two actions for us that would regularly cause SIMS to momentarily come to its knees. Mostly solved here by people learning to perform certain tasks off-peak (which is a pretty standard technique when doing intensive tasks on a database that doesn't have a 24/7 global foot print). Are these actions ok to perform in hours? The other thing that used to bring SIMS down were some 3rd party data sync tools being too aggressive. By the sounds of it this term ESS have done exactly this to themselves for some of their users. For us a big one was the Document Scanning system - it was way to keen to check sims for changes so its UI was always up to date. I imagine though neither Arbor or Bromcom have scan-to-cloud MIS integrations at all, and carefully rate-limit 3rd parties calling their APIs?

They requested more capacity from the cloud infrastructure provider and did not get it, not only did they not get it, but a process failure allowed (some) of the previously provisioned lower tier capacity to be recycled back to Azure before the new capacity had been confirmed. Capacity constraints in Azure are not unheard of.

I think her approach here was fair. It is actually illegal to cycle on the pavement, and by doing so the cyclist enabled themselves to enter a junction where turning traffic had been given priority by the red light the cyclist chose to bypass by breaking the law. The other bit you describe is one of those situations where there is a correct answer and a practical answer. If a cyclist on the road is coming up the inside, the car should come to a stop and be clearly indicating that it is intending to turn left before acting as though it has priority. On your road it sounds like cyclists and local drivers know what happens there for practical reasons, doesn;t make it right, and one day someone will get knocked off. One problem I have (as a cyclist) is many car indicator light are basically invisible in day light when you are up close and personal... is the car slowing because the queue of traffic just ahead, or is it slowing because its about swing left in front of you, or have they flashed that driver you can't see that was turning right from the other side of the road?

Both Arbor and SIMS are backed by Private Equity and able to (for now) sustain losses of nine times Bromcom's annual profits. These debts must be repaid at some point, and the money can only come from one place. A third company in this space, that is profitable and competitive at the current market rates poses a long term problem for PE backers recouping their investments. Bromcom's problem in this space is three fold, it can be out invested, it is vulnerable to poor (naive) procurement practices on the customer side, and it is vulnerable to poor (potentially illegal) procurement practices by its competition. In an ideal world it would run purely on the merits and flaws of its product(s), but it isn't ideal and it has to be litigious to ensure it is even in the game. Finally it should be noted that in the big expensive case the DfE underwrote the Academy's expenses because it was a useful test of the market the law and the process. 1M to figure out a market worth 100M's annually is really nothing. After that was settled, all schools and Trusts should have sat up and made sure their houses were in order. Now excuse me while I go clear some space on our SIMS server and make sure our link to Classcharts recovers.