96 MacBook Pro’s in one server rack(simbimbo.wordpress.com) |
96 MacBook Pro’s in one server rack(simbimbo.wordpress.com) |
You need to buy an expensive thunderbolt adapter. I can understand removing it from the ultra portable super-thin models but not from the top end professional model.
http://www.amazon.com/TRENDnet-Ethernet-Chromebook-Specific-...
The rare use is what makes it a problem. It's not worth the investment to buy and always carry an adaptor for that one time you really need it.
I guess USB-C is supposed to help with this?
Who do we need to talk to for that to be the official collective noun?
Is there some reason why Apple doesn't promote WiGIG?
Here's an overview over the technology:
http://ultrabooknews.com/tag/wigig/
http://www.slashgear.com/intel-wigig-docking-station-in-2015...
http://www.cnet.com/news/60ghz-tech-promises-wireless-dockin...
Personally I wish they used Ultrabroadband-Radio (>500MHz Bandwith). There are numerous reasons fellow RF enthusiasts will recognize in UWB radio. But I am happy with whatever technology allows me to have a cable-free desktop 😌 This is the reason I loved the original Ubuntu Phone (with it's powerful specs).
But at least you'll be able to hide all the dongles behind some wall and just have a single USB-C cable running from your laptop to a powered hub.
£25 for Ethernet and then another £25 if you want FireWire too. And £25 for VGA and £25 for DVI (though to be fair you can get this from the HDMI port). But you can only have 2 of these plugged in at a time. Other laptops of this price come with all the adaptors in the box if the ports aren't integrated.
I like having the option of Ethernet available easily. For example, BT have been having big internet issues for the last few days and it's handy to rule out the WiFi. Also with a previous ISP (sadly not with BT) the WiFi was the limiting factor and I had to plug in to get full speed. Then the HDD was the limiting factor!
I easily reach 60+ MB/s now over WiFi, so for the times where I really need the additional power I don't care about having the additional dongle with me.
Also, the profit margin on those adapters is pretty astounding.
http://hengedocks.com/pages/vertical-macbook-pro-retina
I've been doing quite fine with the DOCK as my 'need ethernet at my desktop' solution:
http://www.landingzone.net/products/for-the-macbook-pro/
I'm sure there are other options; it is a hassle that Apple removed it, but on the road I rarely need Ethernet, and at the desktop, its kind of easy to just plug-in.
Is it mostly about spectrum getting full or security concerns?
I just noticed that I haven't used Ethernet cable for the last few years.
Attempting to provide wireless coverage for a densely populated open office gets pretty expensive pretty quickly. It also can cause degraded signals for everyone in the office. The ethernet port just becomes more feasable way to get stable internet at your desk.
Air circulation makes sense. There is a vent in between the screen pivot and the base, and also the gaps in the keys allow for some heat to escape.
Agreed that the screens should be turned off, though. Could just turn the brightness down to 0.
$ man caffeinateYou can even disable sleep with closed lid[1], but addmiteddly that might intefere with ventilation, since there are vents in front of the hinge.
[1]: https://developer.apple.com/library/mac/documentation/Darwin...
https://jamfnation.jamfsoftware.com/viewProfile.html?userID=...
and the products http://uk.mathworks.com/products/?s_tid=gn_ps
For example, you may have a macbook pro, but it's not powerful enough to run tests in a few minutes while you do other work on it at the same time. In this case, you'd want another Mac like a Mini, Pro or Xserve to send jobs to.
Anyone using X Code has to run it on a Mac and often, you're wasting precious development time when your jobs are taking up your Macbook's processing power for an hour while it tests your latest build.
Travis-CI, the continuous integration platform, will send your jobs to a cloud of Mac servers when they require OS X to build against.
I think Steve has done a great job on his MacBook Pro rack design, but I do wonder about the advantage the MBP has over the Mac Pro for their workload.
It may be that the GPUs of the Mac Pro don't matter to them, which would reduce the value proposition of those systems by a fair amount. The MBP also has an advantage in that it was just updated, and the current Mac Pro is a bit older. That advantage is hopefully just temporary, but it is pretty likely that the MBP will continue to receive more frequent updates in the future.
You can definitely squeeze more systems into a rack with the MBP, but my preference would still be to go with the Mac Pro:
No batteries on your datacenter floor
No power bricks
Better airflow
No screen burning electricity (although this is minimal)
The batteries particularly worry me. They're a potential fire hazard, and over time you'll need to replace them (even though they don't matter) before they expand and deform the system chassis. They also make power outlet control over a system annoying, since you'll have to wait for hours after the outlet is cycled off and the battery has discharged to cycle the outlet back on.Not only is this waste of money but also non eco friendly buying all that hardware
Well, no. The guy putting that rack together didn't know what he was doing and insisted on using the stock AC adapter for the devices. You NEVER use the stock AC adapters in a cluster. They are (usually) made to be cheap and not operate at full load. Maybe 5% will fail under continuous full load. Put 16 in a cluster and now you are looking at a 60% chance of a single failure.
ALWAYS ditch the bundled AC adapter and use a single, good quality, high power PSU that branches out to all the boards. 5 volts and 40 amps, for example. These PSUs are more like 99.9% reliable, and as a bonus output much cleaner power.
The only reason that this guy had success with Chromebooks is because laptop PSUs are typically sized at 3x capacity (for battery charging). Running a stock PSU 24x7 but only at 30% output greatly reduces the failure rate.
edit:
http://www.systemcall.eu/blog/2014/06/trashing-chromebooks/
https://news.ycombinator.com/item?id=7876235
Besides ignoring the power supplies, the original author used SD cards on the HK boards instead of eMMC, which is another reliability no-no. Though HK is pretty good about shipping quality PSUs with their hardware, so I suspect it was uSD being flaky in this guy's case.
No, i ask myself why dont you just use the Mac Pro (ie the one which isnt a laptop?????)
If the software they're testing is used mainly on Macbook Pro's, then they're wanting to test on Macbook Pro's. Things like GPU and CPU differences may make a difference to how their software performs.
Seems safe to say this is not a small company
- Can you say who you work for?
- Sorry, unfortunately no.
For example, with mac Mini's in a server environment, you need to plug in a hdmi enabler to get full GPU performance front the system while using it over remote access.
Here's more information: http://www.macstadium.com/blog/osx-10-8-10-9-headless-gpu-en...
http://www.apple.com/shop/browse/home/specialdeals/mac/macbo...
Refurbished 13.3-inch MacBook Pro 2.7GHz Dual-core Intel i5 with Retina Display
$1,099.00
Save $200.00
15% offAlso, no one buying in this quantity should be paying list price to Apple, which cuts into the refurb discount considerably (if not outright eliminating it).
This way I only have to plug in two cables instead of an array of cables. This is why I really don't miss the Ethernet port directly on the machine. For me personally, the gains in portability due to smaller size (and weight) are way more relevant than the theoretical ability to plug in an ethernet cable which I never need.
If you ask whether they'll keep regular USB or thunderbolt, perhaps but just for a few releases more. E.g. both could be gone by 2018.
That said, MPB will of course have > 1 USB C ports.
Just think of the potential loss of equipment if the sprinklers were to activate in a datacenter suite -- and if the flammable material is located in your cage, it's going to be you or your insurance covering that.
But yeah, there's plenty of
NO CARDBOARD ON THE DC FLOOR
NO CARDBOARD ON THE DC FLOOR
NO CARDBOARD ON THE DC FLOOR
At every multi-tentant dc.I can see that this is very much country dependent (we have truly unlimited plans that allow tethering), but at least for me, the smaller size of the machine trumps the ability to plug ethernet cables due to the general lack of available cables to plug.
Why, they don't have the skills to create a private wi-fi network for guests?
I never use guest networks for that exact reason. I know that my traffic and my connection is my own, and works. Way easier imo.
The Thunderbolt widget is less than ideal, but it fits. Hopefully USB-C will introduce more options with a lower price-point.
I welcome not longer having a gaping hole in the side of my computer. Wifi has come a long way and if people want to use ethernet it can be run through a multi purpose more modern port with a smaller footprint.
although adding the 3.5mm headphone jack to an auxiliary board was super smart. people keep dropping them with headphones inside.
Providing reliable wifi for multiple people with multiple devices is a hard challenge.
Probably never heard of professional office routers. You can have several times that.
Even if working in a 100+ person company, all such companies I've been already have all the company wide wi-fi and APs the need, so I don't see why the parent's company couldn't.
As for "ensuring they do proper hand-off as you move around the office", it's much easier with APs than plugging and unplugging ethernet cables -- which was the alternative we were discussing.
I don't think they're opening up each row of MacBook Pro's continuously for they're software though. I think they've just got the lids open to make sure the GPU is enabled for better performance in remote sessions.
The big requirement for keeping the lid open is probably to control the temperature. You can run a MBP with the lid closed, but it will get pretty toasty, and these aren't being used for a typical laptop workload one presumes.
Anecdotally, most desktops I used lived much longer than most servers I worked woth. I know a Compaq desktop that worked as a server for 11 years. Still does. One hardware failure in that entire time. (Power supply.) Could be a result of lower workload, but still...
Putting all of this aside though, the real secret to the superiority of hardware meant for the datacenter often comes down to the practice of binning. Manufacturers have different tolerances for the products they produce; hence why Intel has a million models of CPUs, Seagate produces so many different hard drives and Samsung sells DRAM chips to other vendors and makes their own DIMMs.
The products that perform the best are binned in the server-y bins, the others are moved down the list until they fit another bin. No manufacturer wants to discard parts if they can possibly avoid it.
Sometimes you're actually getting a deal, but a lot of the time you're just trading reliability for cost when you use lower binned items like desktop hard drives in a server environment. Sometimes that trade-off is worth it though.
ECC RAM and redundant PSUs alone should be unquestioned advantages over standard desktops.
Not to mention hardware raid controllers, IPMIs, designs that allow fans and other parts to be serviced without downtime... :P
Working with docker containers, virtual machines, streaming, and VOIP/hangouts... bandwidth becomes the main limiter to the number of users per AP, not the memory or other capabilities of that device.
> it's much easier with APs than plugging and unplugging ethernet cables
For the user, and only when the handoff actually occurs smoothly. When it goes wrong, folks have to power cycle their WiFi, or their entire machine. Compared to picking up a nearby Ethernet cable and plugging it in, this isn't that challenging usually. Of course, having to also pull out a dongle can get aggravating, but we were discussing that as well.
I did see one system that used HI-FOG sprinklers. Smaller water droplets are supposedly able to extinguish a fire with significantly less water. You're still sad, but maybe you can salvage more equipment.
dry pipe systems are probably the most common though.
Until bitten by a problem, yes, that's exactly the thinking. In theory, a single AP can handle two thousand plus simultaneous connections. Most users have found this number, and don't think beyond it.
However, most APs don't have the memory to manage more than 40 or 50 keys for encrypted connections. Throw in shared bandwidth on this maxed out AP, and suddenly the (again theoretical) 1 Gb+ connection is down to less than 20 Mbps per user, or 1/5 that of a wired connection of ten years ago.
That's obviously not acceptable, so the best way to bring up the average bandwidth per user is to have fewer devices on each access point. And when each user is bringing 2-3 wifi capable devices with them everywhere they go...
The math is not very forgiving.
Are people so used to working in inept and el-cheapo companies that they cannot fathom a company wide Wi-Fi with APs (professional, not your home router free with your cable connection) serving "a couple hundred devices"?
Is that considered IT voodoo in some companies?
Heck, your $30/night Motel 8 can manage that...
Compared to using plain Ethernet drops, the cost is hardly negligible, to the point where it isn't worth it for many companies.
My company has been going through the process of implementing this for a new office to hold around 100 people, and the amount of planning required to provide good coverage at a reasonable price has been enlightening to observe.
> Heck, your $30/night Motel 8 can manage that...
Not in my experience. At least not well enough to support bandwidth above the 100Kbps range.
ECC RAM will both work-around and warn you about the start of RAM failures before complete DIMM failure.
And sure, consumer drives have the same MTBF as enterprise drives in practice[1], but: when RAIDing consumer disks, make sure that you get models that immediately report read/write errors, instead of retrying, which is the main (or only, if SATA) difference between enterprise and consumer drives.
If you have a nice RAID setup, you DO NOT want a drive to retry a failed block read 15 times before reporting a problem, and causing latency: you want the failed read to be reported immediately, so your controller can pull the same block from a different disk, and start diagnosing possible faults.
1 https://www.backblaze.com/blog/enterprise-drive-reliability/