A DAY in the LIFE of the DATA CENTRE | FULL CUSTOMER "RACK & STACK" with ASH & JAMES! 

did that last week. went through all of them then I bought 3 servers.

Haha weak! Same here.

Same here, i work in a small DC and am hooked on this channel

@@MsFxtrdr I bought an entire datacenter just because of this video now I am poor and and have a useless datacenter . Jokeing

@@petermeisenstein3752ohh thank god you were joking, thought u were serious for a second😅

It's fun having watched many of Linus' to now see how a server room is supposed to look

Not only a great idea, it's code for anything cat 6 and up, as the zip ties trend to squish the space between pairs in the cable, which increases crosstalk and reduces heat dispersion.

yes omg I switched to velcro recently and it improved my life so much xD

Anyone that uses cable ties should be shot. Especially anyone that doesn't flush cut them (looks at scars on arms)

Yes, I totally agree. Very neat and tidy, which makes any form of maintenance easy to carry out.

Nice job, you should just buy cables already in the lengths you need. Working in the US Postal Service data center we had cable in .5 foot increments up to 20 feet. You waste a lot of time terminating RJ-45's yourself.

Dude exactly. All that cable mgmt and then you terminate without testing? Yikes if a cable is improperly terminated.....no labeling either. Still a very cool video tho.

That's pretty much exactly what I do, just sit at my desk with a box of cable and terminate and test. That feeling you get when the tester is like bink, bink, bink, bink, bink, bink, bink, bink. Feels so good. The worst is when it's the eighth wire and instead of bink it's bork! Ugh lol.

you should look up fiberninja ;)

honestly ive used tons of different systems and I have yet to find one I like for copper, particularly in the type of racks in the video. In fact the only one I think works as intended are the big fibre managers. The rest are frustrating to deal with for repairs or decommissions.

They didn’t make it. They’re it’s bodyguards

Great tip! Agreed.

This! I was like wait you aren't going to do all the cable management and then test are you....oh no....

Welcome to paid-by-the-hour remote hands service. I agree though, you can buy patch cables from most decent places in approx. 4U increments. Also, you have to disconnect these wires to pull out the server and they had a ton of waste, with CAT6/7, that adds up quick. I'd leave at least the depth of the server in slack so you can pull it out, swap RAM etc I personally like to keep power cords on one side and network on the other, it seems they went with all wiring on the same side. Do like the loom tool though, may want to invest in one of those.

@@GuruEvi he actually examined why he put network cable on same side as power. There will be fiber too and he wants fiber and coper separated.

Thought the same thing. Security studies kicked in there

My last job several of us had access to the DC but if a group needed to go in one person would swipe and then we'd all go in. I guess if the cameras went down for some reason and there was an incident then I could see why it's important every single person swiped. Is that what you were getting at?

Okay now do 568A. No cheating...

​@@user-rr3fo6hy9qW/Green, Green, W/Orange, Blue, W/Blue, Orange, W/Brown, Brown

@@user-rr3fo6hy9q Haven't used 568A in 20 years.

Same sentiments lol and how will they be able to get an inventory of the port assignment from those wire without any signs/tagging

@@bonsky899 presumably from LLDP

@@donutchucka8275 Yeah, we just buy pre-made the tiny amount more mess because of not having perfect length cables is worth it

Yeah that's what I noticed as well.

@@MrMrRubic it's pointless, costly extra cable lengths just to avoid unplugging a couple of cables in order to slide it fully out? Most don't bother with cable slack and concertina arms these days. You've got to power the thing down to do most maintenance tasks anyway so no need to keep the cables plugged in.

Thats what you get when you use in all these pre-term cables and service loops everyone else here is insisting is necessary. Giant pile of spaghetti mess.

yeah I've learned to always keep enough slack for that. Fiberninja is a really good channel to watch too :)

always keep a service loop less downtime

Wouldn't that cause inacurrencies in length though? Not to mention the economics of what is actually needed would be bad because a cable may be a tad longer then what's needed. Therefore a little money wasted?

@@travisnelson9104 cable arms are used to tidy up different length cables no big deal, it's used everywhere. yes premade will be more expensive but given the fact that you have to do a lot of crimping (fill a rack), the time needed usually costs more in the long run.

25-50 cm pre-made cables are cheaper than 2 rj45 connector...

I think that Cat6 is ok, but Cat5e not anymore. So i think that they are ok with it

@@eoeoeovideos 5e is ok for short runs, say from a PC to a wall jack, or for patch panel cables. Anything longer I'd use 6 or above.

No they can't. They are going to have to unplug the server to pull it out. If a fan fails the server has to be unplugged, which is no big deal if you have high availability and replication... but its not ideal.

Clearly no cable testing being done before installation, nothing like trying to find a faulty cable later and having to pull apart that loom again 🙁 Do it properly the first time!

Rails with cable management arms are great for keeping the cables super neat and enabling the server to be slid out. A cable tester could have been used and not added much time, but been another report for the customer to see (a quality tester will be able to have a report per cable). Something like a fluke link runner (I use an AT2000) will do this while one end is plugged into a switch. Qualifying it for gig speed.

@@anthonypolsinelli1179 On the other hand cable management arms can impede airflow quite a lot. I'm currently in the process of removing all the arms from a 5-rack HPC. It's a pain, but in this case the arms are quite full (redundant power + redundant network + redundant Infiniband + IPMI = 7 cables per 1U), and the machines do really run several degrees cooler without them - which means they will run at higher clocks. I'd recommend arms only for 2U+ servers.

they probably tested the cables off camera, thos it's weird they didn't mentioned it..

Especially since they are using cat5. Alot of data centers use cat6 minimum with already made cables. Only did crimping in college and home use haha

Taking up way to much unnecessary time?

So for my little (~1000 node) network we buy pre patched too, but running exact length is by far cleaner. If we werent shutting down this network I would have insisted we cutover to manual and align with the rest of the company standards. The other issue I have with prefab is when the server guys run cables and they pick up a 20 ft instead of the 15 ft they needed. Especially with big builds space is at a premium. And decommissioning becomes a nightmare. But for small enterprises with a dozen servers than yea dont waste time patching. Just depends

I think the rails are still there for ease of maintenance. I mean depending on the server type things that you can hot-plug like e.g. disks are often accessible from the front or back directly. All other parts you typically cannot maintain while the machine is online anyway. So you'd plug out electricity and ethernet and then start to replace/add components to the server. However you can still do this without completely taking the server out of the rack. Just pull it out the server on the rail, open the chassis and replace the parts.

During maintenances we often need to power down entire machine to open box and do whatever is needed. I liked their job leaving the cables as it is is much better for air flow and less visual noise

@@areyounow_2023 visual but not practical and if you need to shut down an entire production rack to work on 1 server you’re doing it wrong 100%.

@@CustomComputing I understand your pov

@@CustomComputing If you don't have redundancy so you can power down a device and do maintenance you are doing it wrong 100%. What happens when a device fails during production hours or you have to do troubleshooting/maintenance? Sure you can do it after hours but some customers work 24x7. So you better have redundancy.

I agree with you but I think in a Datacenter that doesn't apply really. First of all let's agree on the fact that a server in a DC has to have an uptime of 100% (except of course for the be weekly windows update or some patching that requires a reboot, or maybe a dedicated software that has some issues). Then, if you don't have any connectivity and you need to have access to a serveur, you can go plug yourself on the back or if there's a KVM you can use it too. So beside that, the only time you will pull those servers is in an EOF cycle. Also, if you suggest a hardware issue, the "most common" issue is regarding the SSD/Hardrive having an issue or a raid controller having an issue, in those cases, no need to pull the server you can hot swap those) Remember that those servers are top of the line servers (full RAM, bi processor, dual power supply etc). I've honestly never seen a top of the line server have a critical hardware issue and it's even true in the recent years. That's what I think, thanks for your comment.

@@Loutripe most common issue we have is having a "RAM Error" which can be remedied either by popping all the dimms out and pushing them back in, or if that doesn't work then replacing one. The other thing we find ourselves doing is changing pci-e cards for NIC upgrade or downgrade to 10GE or 100GE. Servers get repurposed a lot around here so the 100% uptime comes from migrating VMs around. But dealing with bare metal at the Colo really sucks if you can't slide the box out on it's rails.

personally i would zip tie the power cables to the server as possible, way too many accidents with power when doing repairs otherwise

Apparently nobody here has heard of redundancy. You should have that in place so even during production hours you can shut it down and do maintenance without disrupting end user experience.

I didn't see a NetScout either...

Agreed. I was very surprised they didnt at least put a tester on each cable. Anytime i make a cable, i always throw a tester on it.

@@Bob-TheTechGuy eee, managed switch shows you if its gbps duplex capable, or you just floodping or run test on each machine for some time. these guys look like they know what theyre doing. i also met shitload of out-of-box not working cables

This is insane to me. My last job we did all the cable runs, termination, and testing ourselves. We rarely hired vendors for cabling and never bought cable to the exact length. Terminating your own cable and testing it is a standard task for any network engineer. Plus it makes the cable management look way better. But hey to each their own I suppose.

I wish I could share videos and photos from inside a facebook data center on how this is done, as well as all the fiber dressing. everything is pinstriped, and perfectly measured including the fiber cables.

They are a datacentre, not a ISP.

Matt Ford Seems like Neil is a network engineer so he thinks he’s “smarter” than DC Techs. I’m sure he knows he can find a bunch network focused channels but it makes him feel smarter requesting it here.

I don’t think its up to them. They are mounting the equipment according to customer’s specification/design

Most of the time it's simply added effort. The cone makes it much easier.

Start at one end and work your way down. I was taught at like 18yrs old with power wires