Posts made by crustachio

OK there's a barrage of posts coming in so single quoting them is getting tough.

I just want to step back for a moment. I REALLY appreciate the logical replies and hard-hitting questions. It's giving me a lot to think about. What follows will no doubt sound like excuses, but I promise I am trying to keep an open mind here:

I don't think anyone can just declare that we should eliminate Windows from our environment. You don't know our budget, current applications, workflow, user count, or anything else. Yes, I am sure that through application of simple logic and analysis you could make a convincing argument against staying on Windows, especially looking at the situation as presented through a handful of posts. But that in no way allows for all of the soft-costs associated with this. For a team of our size to engineer and execute an organizational wide replacement of an entire operating architecture would be a near impossibility, to say nothing of our other ongoing projects. Recent organizational investments in Windows-based 3rd party applications (including ongoing painful user training and data migrations) alone would kill this in its tracks.

We're kind of getting down in the weeds here. I recognize that there are most likely BETTER ways to do IT in this organization. But the fact is, we are where we are. It has been decades of decisions made over a long time based in the old IT world, and we are very well entrenched. But I can't change that overnight. I am not the key decision maker. My job title is simply "Specialist", and while I can advise and implement, in the end it isn't my call. There is absolutely zero chance of me selling a "take off and nuke the site from orbit" scenario here.

What I am trying to do is make the best of the resources and influence I do have. We're not leaving Windows, at least not during this administration. But yes, maybe Hyper-V is a realistic alternative that we should stop and take a harder look at.

Looking forward to getting this post shredded in 3...2..

@scottalanmiller said in Windows Server 2016 Pricing:

@crustachio said in Windows Server 2016 Pricing:

Organization has no opinion; he is director of IT and has sole discretion over IT infrastructure budget/purchases.

Who does he report to in the command chain?

City Manager.

Again, I didn't say we "can't afford 2016". I'm just having a hard time justifying the price increase as applies to our current environment.

@Dashrender said in Windows Server 2016 Pricing:

@crustachio said in Windows Server 2016 Pricing:

@scottalanmiller said in Windows Server 2016 Pricing:

@crustachio said in Windows Server 2016 Pricing:

If we're comfortable riding 2012 into the sunset....

I have a rule of thumb about this... anything that something like this is said, it probably means you should not be on Windows

See above... We have an unsettling amount of 2003 servers in production.

The question is why do you have those 2003 servers? Lack of funds to upgrade, lack of motivation to upgrade? lack of support from vendors to upgrade?

That's a question but it's not the question. Pick a reason.

@coliver said in Windows Server 2016 Pricing:

@crustachio said in Windows Server 2016 Pricing:

@scottalanmiller said in Windows Server 2016 Pricing:

@crustachio said in Windows Server 2016 Pricing:

If we're comfortable riding 2012 into the sunset....

I have a rule of thumb about this... anything that something like this is said, it probably means you should not be on Windows

See above... We have an unsettling amount of 2003 servers in production.

Perfect time to move away from Windows then. You've made it clear you can't afford it so investing the time and money into moving away from it would pay off in the long run.

The other thing is, your boss doesn't thing the organization can afford Server 2016, does the organization (management) feel the same way?

Organization has no opinion; he is director of IT and has sole discretion over IT infrastructure budget/purchases.

Moving away from Windows... We have numerous 3rd party vendor applications that are Windows based. Not to mention Exchange, AD, terminal services, etc. If switching to Hyper-V is an uphill climb, what do you think my odds are of pitching a MS exodus?

ETA: I don't think I said that "we can't afford Windows". My reason for bringing up the 2016 licensing bump was to point out that it is a big cost increase for some organizations, not to say that we outright can't afford it. The question is, is it worth it to us? Running a Windows environment has huge value to us. Running 2016 Server in particular, I don't see the value yet. IF we made the decision to switch to Hyper-V, then 2016 would be a no-brainer.

@scottalanmiller said in Windows Server 2016 Pricing:

@crustachio said in Windows Server 2016 Pricing:

If we're comfortable riding 2012 into the sunset....

I have a rule of thumb about this... anything that something like this is said, it probably means you should not be on Windows

See above... We have an unsettling amount of 2003 servers in production.

@scottalanmiller said in Windows Server 2016 Pricing:

@crustachio said in Windows Server 2016 Pricing:

I agree in hypothetical terms, but as the environment is already bought and paid for, save Windows licensing, what's the "lot of down the road cost" we risk suffering?

This is where expensive software sucks people in.... they have two really good emotional ways to get people to keep spending money. First is the sunk cost fallacy, which you mentioned above, that feeling that since you paid for it, you should use it. The second is the belief that it is already paid for and incurs no more cost.

The second is untrue for many reasons. For example, with VMware you have to license it year after year, you pay continuously. And when you want to update, you pay even more. You pay for tools and invest in designs and other decisions around the fact that you have it - indirect investments in whatever design you have. In your case this includes VSAN and Windows licensing investments, at the very least, and probably a few others that haven't been mentioned.

And last, you invest in knowledge. You put your time and effort into learning what you are running currently. Right now, Hyper-V has little downside compared to ESXi. But in three years when your whole team has built all documentation, policies and procedures around ESXi that will be a very different picture.

Running systems always incurs debt. Putting your debt into your future rather than into your past is an important thing to always consider in IT.

Thank you, there's a lot of good stuff to chew on in that post.

If I am completely honest and stop making excuses to rationalize our predicament, then yes, you're completely right. It is not too late to stop and rethink this. It would be a MAJOR uphill climb for me vs my manager though, as he is pushing for the VSAN deployment to be operational "yesterday", and I'm 99% there except for licensing.

I would also need to go back to square one in terms of research and education. I am comfortable administering a Vmware environment, I know VSAN pretty well and I've engineered the hardware and network (new 10GbE TOR switches) around VSAN. I have almost no comparable experience in terms of Hyper-V. Could I? Sure. But it's a factor. Like you said, investment in knowledge is a factor. And Vmware is something we already have a comfortable investment in knowledge and experience with.

Side point: Switching horses in midstream to Hyper-V would require using Starwind for the local shared storage component, as I don't think 2016's Storage Spaces Direct is the way we'd want to go. I have no idea what Starwind licensing is like, but wouldn't that be a similar argument about future costs as sticking with Vmware? Either way we will have to continue paying licensing.

@scottalanmiller said in Windows Server 2016 Pricing:

@crustachio said in Windows Server 2016 Pricing:

@scottalanmiller

@scottalanmiller said in Windows Server 2016 Pricing:

@crustachio said in Windows Server 2016 Pricing:

But it's not just the sunk cost fallacy at play. Our storage and compute needs are contingent on using at least 3 hosts, and since we're running VSAN, 4 is the true safe minimum, to say nothing of disk groups and future storage growth.

Why are they contingent on three? And the investment in VSAN might be part of the problem, why choose a product that doesn't match your licensing needs? VSAN is great, but it's a high cost option, on top of VMware ESXi another high cost option that then incurs a higher WIndows cost... it seems like the approach assumes high cost anyway, is the Windows licensing really a problem then?

I totally get that the investment was "just made", but wouldn't alternatives like Hyper-V and Starwind from the MS camp or XenServer have not just eliminated the VMware and VSAN costs, but potentially the Windows ones, too?

We're local government and get VMware products for very, very low cost vs list. My manager is not confident in Hyper-V as a solution so never gave it a full shake. In hindsight, sure, Hyper-V + Starwind might be cost competitive, especially with Windows licensing in view. Not sure if there's anything I can do about that now, however.

So he believes in Microsoft enough to depend on them, but not enough to depend on them all the way

Something like that, yes. We still have 2003 DC's, so......... no comment.

Hyper-V is definitely not proven like ESXi is, nor is it "as good." But Xen is more mature and just as proven as ESXi and totally free as well.

VMware cheap is good and makes it hard to beat. But if it is free, then where is the real issue? Just modify the hardware slightly, reduce compute load to two hosts. You have a light workload, right? Why the need for so many compute hosts?

What's your definition of light workload?

We're looking at 48 VMs out of the gate, including a couple not-insignificant SQL servers, plus an immediate Exchange server.

@scottalanmiller said in Windows Server 2016 Pricing:

@crustachio said in Windows Server 2016 Pricing:

It's really too late to change the trajectory of our ESXi deployment at this point, and if we did so just for the sake of this licensing cost we would spend more re-engineering the solution than just eating the licensing bump.

How much engineering is required in that small of an environment? And wouldn't good re-engineering now, even if it has a cost, that saves a lot down the road potentially be well worth it? Cut the technical debt ASAP rather than later once you have accumulated more of it?

I agree in hypothetical terms, but as the environment is already bought and paid for, save Windows licensing, what's the "lot of down the road cost" we risk suffering?

If we're comfortable riding 2012 into the sunset, and can afford Vmware licensing, I'm not sure what the future risk is.

@Dashrender said in Windows Server 2016 Pricing:

@crustachio said in Windows Server 2016 Pricing:

We're local government

And... We're out.

Eh?

@scottalanmiller

@scottalanmiller said in Windows Server 2016 Pricing:

@crustachio said in Windows Server 2016 Pricing:

But it's not just the sunk cost fallacy at play. Our storage and compute needs are contingent on using at least 3 hosts, and since we're running VSAN, 4 is the true safe minimum, to say nothing of disk groups and future storage growth.

Why are they contingent on three? And the investment in VSAN might be part of the problem, why choose a product that doesn't match your licensing needs? VSAN is great, but it's a high cost option, on top of VMware ESXi another high cost option that then incurs a higher WIndows cost... it seems like the approach assumes high cost anyway, is the Windows licensing really a problem then?

I totally get that the investment was "just made", but wouldn't alternatives like Hyper-V and Starwind from the MS camp or XenServer have not just eliminated the VMware and VSAN costs, but potentially the Windows ones, too?

We're local government and get VMware products for very, very low cost vs list. My manager is not confident in Hyper-V as a solution so never gave it a full shake. In hindsight, sure, Hyper-V + Starwind might be cost competitive, especially with Windows licensing in view. Not sure if there's anything I can do about that now, however.

@Dashrender said in Windows Server 2016 Pricing:

i thought I read somewhere that you would get grandfathered in if you are running more cores than the new licensing comes with by default. anyone else read that?

This is a new host deployment, so there is NO licensing on these hosts currently. Nothing to grandfather. We're coming from OEM licensed physical servers.

@scottalanmiller said in Windows Server 2016 Pricing:

Anything stopping you from changing how that is? It's not as good as having smart licensing, but can't you consolidate to one or two hosts?

Well, since this is a new deployment we just sunk a whole heap of cash into these hosts

But it's not just the sunk cost fallacy at play. Our storage and compute needs are contingent on using at least 3 hosts, and since we're running VSAN, 4 is the true safe minimum, to say nothing of disk groups and future storage growth. We weren't expecting the 2016 per-core licensing cost increase when putting this project together, we assumed flat rate per-proc licensing as usual. It's really too late to change the trajectory of our ESXi deployment at this point, and if we did so just for the sake of this licensing cost we would spend more re-engineering the solution than just eating the licensing bump.

As a completely on-prem, VMware-invested environment, can anyone list any significant reasons not to just stick with 2012?

@JaredBusch said in Windows Server 2016 Pricing:

Of course anyone running more than 2x8 core will pay more. But then no one in the SMB really needs more than that. They certainly do not generally need data center in the first place.

What workloads do you have that you need so many cores?

ESXi 6.0.2 running VSAN and quite a few VMs.

We're in the middle ground between SMB and Enterprise. I can understand a simple SMB just running a handful of VMs never needing more than 2x8. And then a full scale enterprise with many hosts who can negotiate licensing. But we're in the middle... several dozen VMs spread across a handful of hosts.

@JaredBusch said in Windows Server 2016 Pricing:

@crustachio said in Windows Server 2016 Pricing:

@JaredBusch In what respect?

The claim of more expensive in percent does not clarify that it is more expensive compared to server 2012 pricing.

Any sane person looking at that graph sees a standard linear price line per core.

That graph was taken from the linked article discussing the 2012 vs 2016 pricing. Yes, out of context the graph could seem misleading... but for the purpose of this discussion it is entirely sane and relevant.

Do you disagree with the interpretation that running more than 8 cores per pCPU will carry the indicated price increases?

ETA: Imagine another column labeled "2012 Price" , where every row said "$6,155".

It looks really cool and I plan to use it for my next trip, however it looks like (and correct me if I'm wrong) it does not offer any sort of collaboration or shared access to a trip? In other words, my wife and I can't access the same trip on our phones? Seems like a big miss if that's true...

I have seen so many oddities related to the Win10 start menu... it is confounding. Last week it would not take my keyboard input, so I couldn't open any app using the usual Win + [type] process. Reboot was the only thing that fixed it. Bizarre for such a core feature of the GUI to act out like that.

From the Microsoft Windows Server 2016 Licensing FAQ (PDF Download

See also:

Instead of 2012's two socket license pack, 2016 will use a 2 core pack, with the license cost of each 2016 pack being 1/8th the price of the corresponding 2 socket pack for 2012. Each system running Windows Server 2016 must have a minimum of 8 cores (4 packs) per processor, and a minimum of 16 cores (8 packs) per system.

For systems with up to 4 processors and up to 8 cores per processor, this won't change the overall licensing cost. Above this, however, things get more expensive; although the price for a single processor 10 core system will remain the same, with two or or more sockets populated by 10 core processors, prices will go up; 2 or 4 processors with 10 cores per processor will cost 25 percent more to run Windows Server 2016 than they did 2012.

(Source: ArsTechnica)

@JaredBusch In what respect?

Pricing remains this same if you're only using 2 physical CPUs per host with up to 8 cores each. But with higher core count CPUs (and let's be honest, anyone running a virtualized datacenter will be using more than 8 cores per CPU), you have to license per-core, in 8-core bundles (EDIT: 2 core bundles, not 8 core bundles, my bad).

See here: http://www.vladan.fr/windows-server-2016-what-is-the-difference-between-standard-and-datacenter-edition/

http://www.vladan.fr/wp-content/uploads/images/cores.jpg

We're about to get bit by this. I just stood up 3 new hosts in a VSAN deployment (working on the 4th). Dual Xeon E5-2690V3 12-core CPUs per host = 24 cores per host = $9,232.48 per host for Server 2016 Datacenter licensing. Times 4.

That's 150% of the cost of Server 2012 for our deployment. And it is really making me wonder if we should move to 2016 at all. Almost all of the new features are aimed at cloud convergence and/or Hyper-V, none of which is applicable in our environment. Heck, they're even locking out features like Nano Server to Azure based deployments only. The only reason to upgrade as I see it is to avoid earlier sunsetting/EOL for 2012. No significant feature changes that validate paying 50% more.

I'm curious to see what kind of SLED contract pricing is available. Reports say the 2016 pricelist will be published Oct 1.

I always throw /f in too, just because I don't take no for an answer.