@scottalanmiller said:

I wonder how much of this came from me because I came from a background of "you never call the vendor." Not that you literally would never call, but you only call when it is their fault (bug in code or whatever.) You never ask them to do the IT job for the IT people - if you do, you aren't the IT person but just a vendor coordinator. I've never worked in a place where it was culturally acceptable to turn to the vendor for IT tasks that are the IT department's responsibility. I've carried that on. You lose a lot of advantages when you do that (delays, lack of cohesive vision, etc.)

It is only since joining some online communities that I even learned that there were vendors that would participate in this and would do IT work or handhold in the way that they do.

It also depends on the problem. I've worked in Student Information systems that were buggy as all get out. Sure, IT can work aroudn the issues with a few SQL Queries and patch the problem until it happens again... But instead, we should call the vendor and see what their recommendations are first. If they say "We'll fix it later"... Then "A patching we will go... "

@thanksajdotcom This is why it is helpful to be friendly with those higher up than your direct supervisor(s), lol.

@thanksajdotcom Give it a chance... Maybe you'll get a new GM who is also halfway decent... and one who can fire the Sales manager....

@thecreativeone91 said:

Dual PSUs are for failover. You can't draw more than the wattage of one PSU. And even then anymore than 80% of it's rated load is not something you want to be drawing.

Yeah, I kinda figured that, but my questoin about the power consumption would have been answered if I had watched the little popups in GPU-Z, lol. It's the power usage on the card it self, and it is per card.

When something graphics intensive, such as 3D Mark or Unigine's Heaven ar erunning, it runs about 50 or 60% with general remote desktop usage, it hangs out between 10 and 20 percent.

And it's not steady... it goes up and down depending on how hard the GPU is working.

Sadly, this is a state level job, so yeah, I don't see that happening any time soon, lol.

As much as having everything the same would help, if you are an MSP, as @JaredBusch said, there's no need to change for sake of keeping things the same especially if the client is not having major performance issues.

@thanksajdotcom That is the situation I am in -- kinda. At my last job, I had my hand in all the cookie jars... At my new employer, I am a Systems Administrator only... We have a lot of nice Cisco Networking gear that I could quickly fix issues on, but I am a System Administrator, not part of the Networking Team. (Sadly, all of our IT Functions are siloed like that... it breaks my heart).

I do like my bosses and folks that I work with, so that is a plus!

Perhaps they just don't want those of us that are not afraid to think.

@Danp I got it activated again last night at 10:45PM... IOWait was still low then... 0.03 %

This looks interesting.

Looks like they fixed mine. My box is totally idle at the moment, just restored my OwnCloud instance last night.

Scratch that, I just checked the order sheet, and it has 2 x 1100 Watt power supplies. I'll be over in that server room later this morning, so I will check it when we are there.

Yeah. I just updated the drivers a couple of days ago this week, 347.88.

Is there any way I can check the current power draw or PSU size remotely? I know the system has dual power supplies, and according to GPU-Z, power consupmption is ~20% on each of the 4 Grid K2 Cores(?) that show up. Is that percentage applied as a whole -- 20% total, or 80% total power draw?

I was wondering if anybody out here had any experience with configuring and tuning Windows Server 2012R2 and Hyper-V + RemoteFX for a VDI setup?

I've got the VDI configuration done, and RemoteFX configuration at least working, but the performance of it leaves... a lot to be desired... I know the RemoteFX is working because the Unigine Heaven benchmark app that I'm using won't even start when I remove the RemoteFX adapter from the VM.

Performance of the standard desktop apps is fine, and even our copy of Adobe Premiere is usable. However, any 3D application or even something as mundane as watching videos in full screen on youtube just simply do not work well at all (choppy video and graphics rendering).

The Server in question is a Dell R720 with 256GB RAM, 6 x 200GB SSD Drives in RAID 6, and 2 x NVIDIA Grid K2 cards and runs Windows Server 2012R2, etc. The Client is WIndows 8.1 Enterprise

If I monitor the server's usage with GPU-z, I can see the memory usage go up and down depending on what I am doing, so i know things are configured at least partially correctly, but the video has me baffled. The systems are connected to the Network with Broadcom 10Gig cards (and I have VMQ disabled, as it really destroys performance).

At 4:3 resolutions, like 1280 x 1024 (and even at 1600 x 1200) the video seems to perform acceptably for watching a video, and not so great for using the Heaven Benchmark.

I guess I'm just wondering if anyone has some pointers or some settings I could check and tweak.

I've googled and read articles about changing the network settings and such, and I have done those and tested with them, but still not getting good enough... I am open to any suggestions or things to think about / watch out for.

Or maybe buy one Unifi unit a month until they have all been replaced?

@Dashrender said:

If the controller dies, we're sunk. I can't get another, Cisco won't overnight me a replacement. I quite literally will be running to BB to buy some home routers to get us up and running that night, then ordering replacement equipment and installing it ASAP. 80% of our use is wireless, and we have a huge single point of failure (the wireless controller). While management (and I) were willing to deal with a day or less of downtime when a new one was being sent overnight, we no longer have that situation.

That seems like a good reason to upgrade to me, lol... That would fall under preventative maintenance.

@Dashrender said:

@dafyre said:

I do see you point... but then comes the question of how do you justify using Wireless N now? Won't wireless G still work? At what point does the technology become "cheap enough" to upgrade? [I realize that answer is usually "it depends"]

If I would have done that 8 years ago, and purchased N-Draft APs, my bill instead of being $25K would have been $40K+. Here I am 8 years later, and sure I could use N, but it wouldn't matter because the Wireless Switch and APs are EOLed from Cisco (primary reason to get rid of them) and I'm still running along mostly fine on G, and with my 100 Mb switches, N wouldn't really give me any real advantage. so I would have wasted $15K+.

Like I said -- it doesn't always make sense to do it that way.... That is when you go back to the drawing board and ask is it cheap enough to upgrade now?

You may decide, "No, the time is not right now" and just upgrade your switches to gigabit or get new wireless N access points... or both. There's no pressing reason to upgrade to N access points unless the wireless devices will be transferring files around the office.... Especially since you only have a 10 meg internet connection, right?

@scottalanmiller said:

Check the cost, though. How much money do you lose going to AC now? How much money is "being ready" going to cost? Keep in mind that likely, in about two years, you can buy AC for probably half the price that they are today. And you can upgrade one at a time as appropriate. The flexibility, time value of money, unknown future principle and other things make investing in technology you can't use or justify yet generally pretty bad. Especially stuff like this that has a pretty predictable cost dropping curve.

I do see you point... but then comes the question of how do you justify using Wireless N now? Won't wireless G still work? At what point does the technology become "cheap enough" to upgrade? [I realize that answer is usually "it depends"]

I usually find that if I have the money to buy $device now, I will buy the latest and greatest thing I can afford. Yes, I know it will be obsolete the next $interval, but what I buy should last mey well beyond the time it took for me to purchase it. For instance, I was allowed to build my own office computer at my last employer. That was 10 years ago. With a refresh cycle of every 3 years for most faculty and staff at ~$800 per year (for a single computer), my $3,000 computer is still trucking along, and still being used by the new network admin, and it will likely be good for another 3 or 4 years.

So yeah, there's a big premium for buying the latest and greatest, but if you keep it long enough, then you can see some ROI on it. It doesn't make sense to do that with everything though... I think it is one of those things that when you can, you should.

What happened to then?

(Nice pic!)

@scottalanmiller said:

Remember, investing today in technology for tomorrow means you pay a premium for that technology.

I will agree with you on this point, for sure.

You are just continuously losing money doing that. You need a tangible benefit to your investments, especially when they are large (in percentage, normally.) AC is not cheap compared to N, it's not like a couple extra dollars. It's real money that you lose, money that could have been banked and used to buy even better stuff when the right time came and a need existed.

That depends on the brand names you are talking about... and at what scale. If you are going to buy 100 Wireless N for a total of $230k, an additional $50,000 is quite a chunk to go from N to AC. However, I think we are starting to get close to the end of life of Wireless N (we're not quite there yet) -- in the sense that most new laptops and devices sold this year come with systems that work with AC, and N (and even G, in dual rado setups).

Since the newer AC Units are backwards compatible with N on the 2.4 gHz band, it makes sense to me, to upgrade to get the extra processing power of the AC access points since they are both firmly planted in that future looking, yet backwards compatible limbo, if you will.