Yeah please do

Also, remove the servers in the background that aren't working correctly. Fan noise should be reduced by changing the microphone setup.

@@user2C47 Those are most likely live servers and can not be taken down without the customer's approval. Also, they sound like BBUs failing.

@@Redal24 how do you know

@@Knightfall23 i am a data center engineer

no, having the background setting inside the actual data centre makes it better

@@KaesOner Not for my ears

techno-titlan XD?

That's just load balancing setup, usually they don't go into the trouble of redirecting prior to pulling the server, but if server takes x seconds to respond your request may fail or there will be another attempt to serve your request by taking it to a separate server. But the main idea is your request goes to the load balancer, that decides which server will process it, if it's down, it can just plain fail or be "redirected" to another server.

@@claytoncoleiro7190 thanks for your reply. I imagine, at the end it's the app's responsibility to implement a proper health check, and at the moment it fails (due to app beign killed or server malfunctioning), the orchestrator will spin up a new instance

@@WolfElectronicS nah, a well designed app shouldn't handle it. Thats a load balancer job. It depends highly on configuration

@Wolf ElectronicS If the server is in working condition but just need to be upgraded/repaired by changing some part or whatever, in VMware case you can use something called vMotion - it can migrate a live/powered on virtual machine from one host to another but there are some requirements to make that happen fast, shared storage is a first one although it's not required but recommended so that live migration goes faster, second thing is VMware vCenter Appliance/installation which actually makes vMotion possible. Dunno about other hypervisors, did not have much work experience with them. Most of applications have a load-balancer of some sort that allows to just share the load between multiple instances of such software or just High Availability software which enables to switch between instances in case of a failure of the main instance. Such things are e.g. VMware HA (high availability for vmware clusters) or HP Serviceguard for Linux.

A room full of noise where you will be working, and you say its very distracting? So you're essentially saying to everyone that you find the place that you will work in a distraction from your task? Maybe you should change your career. Also, the white noise is soothing to a lot of people which makes it less distracting than a completely silent room, with just a boring voice blaring info. Go back to your filimg career.

well you missed it, he goes on to mention the thermal paste

Copper suffices in most cases unless you want really high speeds

Don't forget servers that are nothing _but_ disks.

This is a late answer to your question but maybe I can help. I work in a large enterprise DC for a fortune 200 BIG Pharma. Most racks are 44 RU in height, The server he was working on is 1 RU. So in theory you can rack 44 of those in to a rack. That however is rarely the case. Things to consider is power consumption for the MOAs and airflow/heat. With regards to cloud computing, most DCs are moving toward high density computing. (One physical server with 100 virtual servers running on the one physical) Need a new server? flip a switch. BANG new VM. Hypothetically speaking. 1 rack with 44 physical servers with lets say 20 VMs. That's 880 VMs in one rack. Keep in mind this is all hypothetically speaking. Not going to talk real numbers.

Battery is for write-cache memory. The idea is that on some RAID levels, writing data to disk takes time so it's buffered. Battery prevents data loss in case of power loss (data is stored until the system is started again)

The onboard battery backup stores the current write to NAND before it shuts down. In this way in ensures that atleast the last write is completed when the server is powered back on, thus ensuring that you could resume somewhat safely with any previous write without the risk of file corruption etc. Generally speaking its just best practice to have a BBU with your RAID card.

Why not just place a super cap or button cell battery on the raid card instead.

@@ElliottVeares , they do. A battery or a capacitor bank. Still needs logic to make it work properly.

That is a battery backup unit (BBU). This powers the cache in an event of a power failure, so that the cached data will be written to the disk and not lost because the RAID cache is composed of RAM which is volatile (data will be lost if power is cut off). The BBU in older RAID controllers use Li-ion batteries. Now they use supercapacitors.

From my experience it will be RAID 5/6 in case of storage arrays and RAID 1/10 in case of local storage for servers. About the CPUs, most of servers for hypervisors that I install have two 10-core CPUs with HT, usually between 2.4 to 2.8 Ghz base clock. Obviously in case of super new servers like the ones with AMD Epyc you can get as many as 64 Cores in a single CPU.

The heatsink might look weak, but consider the 10000+rpm fans that push air across it. When I start mine up they sound like a jet taking off, as you can hear in the background. Core counts actually depend more on what you run. Some software like databases, etc. have licensing structures that encourage fewer very fast cores, while a lot of web stuff is done with many slower cores - for a million requests a day you just need a shedload of threads running at once, even if they're not the fastest things ever.

Does HP / Dell Server have a noisy fan when its in idle position ?, I have an IBM X3650 M1 server its obnoxiously loud even the server in idle position.

@@sonicthedgehog9473 For my HP server (depending on the power profile and cooling profile) my server can be almost unnoticeable.

@The Lavian The control room, the offices, and the main lobby are quiet environments.

Welcome to data-center life...

Using a headset microphone would help also. (Half the distance = 4 times the signal strength!)

Dude, he's recording from inside a DC. Chill out

CAN YOU TURN OFF THE FAN OR AIR CONDITIONER? YOU HAVE A LOT OF BACKGROUND NOISE! 😂

@@jaywalkra Actually in my opinion they don't need to record inside the DC especially this video and simply put DC image on the background :-)

Cpus no my college teacher did have a old server in the classroom thats specs said hot swap ram never got to see how well it worked. Not sure if tye pic was hot swapable I thought I heard of it b4 but cant b completely syre

@@mousejjt2 Never heard of hot swap ram. Current servers don't have that, i know that much.

@@1960bosman Oh, nice to know

no, that proliant processor replacement is NOT hot swap, only fan, psu, disk are hotswap

you cant do hot swap while replacing cpu. u need to unrack the server to do it.

Given how many clients have faulty equipment in our data center for months, nobody seems to care so it doesn't matter here. Maybe in America people care.

Probably an SNMP alarm, since monitoring systems and sensors use SNMP

You getting that US healthcare level of pricing?

@@BichaelStevens it's waaay cheaper than US healthcare.

You can not broadcast a website. You can only broadcast within a broadcast domain. So you'd be able to broadcast to the computers in your local network that are attached to the same router, but you can't broadcast to the internet. Also, broadcasting has nothing to do with DNS servers. Do you possibly mean host? Of course you can host without a DNS server, but user will have to enter the IP of your server.

@@jort93z Thanks for replying. When I put the ip in the browser do I have to attach a port to it?

@@KhandkarAsifHossain you can attach a port to it. If you don't, your webbrowser will default to port 80(for HTTP). So, If your webserver runs on port 80, you don't have to specify the port in your browser.

@@jort93z Thanks again. Now that I have connection to my server via internet. Can I use this server as an application server / Database server to give aacess to both android app and web app?

was hoping someone commented on it lol

All U's are connected to a router called a TOR (top of rack). These TOR's are called a T0 switch that is about the size of 1U. Now these servers are in rows as he explained, somewhere in this row is another switch, an every bigger and more spicy one, to compare to the 1U size switch, these are about 6U or even bigger. This would be a T1 switch. Now take all of these T1 switches and send them to another switch where all T1's data comes together, the T2. These can handle so much data traffic its insane. Usually there is an entire row required due to their size, and hold multiple T2s. This basically is the internet access point for the colocation. Now in some cases these are then connected to a network room, or a regional gateway, from these rooms data leaves the datacenter and goes into the global network There is ALOT going on in these network rooms, to much for a simple youtube comment :)

I don't believe so, shutting everything down for the video does sound nice, maybe they'll do it next vid

Sure, would that be for here or to go?

To those that didn't get it. He was making a joke. Of course you can't shut down a live datacenter

hahhaa youd shut the whole business if you do that. he prolly needs to move shooting location.

Sure Ivar just shut down the entire data center and tell clients that you'll be back in a jiffy!

That's the air intake temperature.

As far as I understand, servers will have multiple CPUs for processor load balancing/optimization and for redundancy if one of them fails.

It's a Data Centre...... Not a quiet room.

Not so bad, feel authentic server environment :) But he could certainly try to use software for noise suppression in the audio.

Yes, they should turn off the whole data centre for a better video experience for you :)

It feels good to hear those background buzzes

It’s very easy to hear him

You must be deaf AF

@@TheRealJohnMadden maybe but I blame your ma ma for that, she knows how to scream

@@edgargarcia209 that’s your fault, shouldn’t be hitting it so good

hes a person talking about what he likes, not a news anchor stfu

Read this literally as something disturbing was forced in my face

Server alarms are normal from out-of-date firmware to failed disks. They just get ignored until someone's at the DC next. Wrist straps aren't typically used as the entrances for these rooms have anti-static mats, plus the flooring is designed to avoid any ESD all together.

James Schmidt you don’t see alarms being ignored when detected in any of the data centres I work in. Poor work style by this gentleman in multiple videos

@@crazyboy2006cashier These are customer racks, DC staff don't have permission to solve issues, some don't even have permission to report them

James Schmidt that’s fine but why then are they opening their customers servers, could have picked a room with no alarms.

@@crazyboy2006cashier Dude, did you even watch the video. He literally showed the wriststrap and explained why he didn't have to wear one. 7:55 And that is not a server of a customer, but an old one of theirs. 1:00

Its still correct for most single socket CPU's in the server space. Mainly Intel Xeons.

Early 2000's would be 4GB. Hard to find 64-bit in those days.

@@pederstrmKollenborg it's not been correct for quite a number of years. All modern Xeons can handle 64GB+.

@@aidanthetaylor for crying out loud an old Ivy Bridge E5 Xeon can take 768GB for cpu that's 1.5TB for a dual socket configuration....newer Xeon can take a Lot more ram

@@aidanthetaylor E5's and E7's sure, E3's only started catching up post Ryzen. However, he does mention that you can get more with other configs in the very next sentence. Hell my motherboard and CPU combo can do 1.5TB or more if I allow it.

Ya gotta keep that hand sanitizer handy IJS