A blog covering Azure, Hyper-V, Windows Server, desktop, systems management, deployment, and so on …
I’ll be on the road again next month on behalf of my employers. This time around the session is aimed not at techies, but at sales, account managers, and pre-sales. The topic: how to get you and your customers ready for 2012 … there’s a lot of stuff coming and you don’t want to lock yourself and your customers out of those opportunities!
Below is the info for the Dublin event. Information about events in Limerick, Galway, and Belfast should be out soon.
Way-back-when, I deployed Microsoft Operations Manager 2005 just after it had RTMd. My boss, the IT infrastructure manager, decided that we should do our initial agent deployment in our DR site. We had Windows and HP ProLiant management packs imported. The DR site in question was rarely visited: pretty much whenever we needed to install something new or when we had a test invocation scheduled.
Within minutes, the agents started reporting degraded hardware: memory DIMMs, RAM, and PSUs. That won my boss over. Once the network team were happy, we started deploying agents 17 worldwide sites.
This week I’ve been involved with a proof-of-concept deployment of OpsMgr 2007 R2 CU5 in a VMware environment. The customer wanted to see how it would handle monitoring of a critical service that had received significant investment and attention from the business. The management group was built, and agents were deployed to the application servers (any consultant who deploys agents to hundreds machines at once is being negligent because the customer will reject the un-tuned full of noisy alerts monitoring solution). A handful of management packs were imported, including Windows and SQL Server. And within minutes we had detected an issue. The SQL log file for the application was not able to expand and the critical LOB app was about to fail. Nice timing 
(I swear I didn’t cause it!). The customer’s IT staff were on it and the problem was avoided. Then a day later, once the data warehouse was populated with some info, I ran some performance reports and identified a vCPU bottleneck in the SQL server VM and a recommendation was made there.
To quote Charlie Sheen: WINNING!
It would be easy to think that SysCtr is irrelevant to VMware. Sure, in my opinion System Center + Hyper-V exceeds the alternative. But, elements of SysCtr + vSphere easily exceeds vSphere by itself or with some overpriced point solution with a “v” badge stuck on it.
We know the business values applications (or services). They couldn’t care less about vSphere VS Hyper-VS any other virtualisation. Now this customer has SLA monitoring/reporting on this particular LOB application and thanks to the early warning from OpsMgr, it’s sitting nicely at 100%, and the IT department’s customer is a happy camper.
I was contacted by journalist Julie Bort a few weeks ago to see if I would be willing to chat a little about Windows Server 8 Hyper-V (or Hyper-V v3.0) for a new article on Network World. I was cool with that and did some prep. There is an incredible amount of new stuff – I quickly had an A4 sheet full of notes with items prioritised for the conversation. We talked tech and we talked business. And of course, the subject of Microsoft vs VMware came up.
The article came out yesterday.
I’d say I’ve had more than a dozen people ask me what my Windows Build (Windows 8 DevPrev) “tablet” was like. I’d show it to them and they’d give me the usual it so big compared to an iPad response. There is clearly some confusion because it is a slate PC, not a tablet. So let’s clear it up:
Slate PC
What we got at Build was not a tablet. It was a slate PC. A slate PC is pretty much like a slim laptop with a touch screen and no keyboard. Characteristics include something like an Intel Core i3 or Core i5 CPU, it will have a relatively short battery life (compared to an iPad), and it probably has 2 GB to 4GB RAM … and it can run Windows at this time (Oct 2011). It is not a tablet, and Microsoft (to their credit) hasn’t tried to call it a tablet.
Tablet
The typical Android tablet and the iPad fall into this category. The Build slate PC does not. The key traits of this device type are driven by the need for long battery life. There is relatively little storage, not much memory, and the CPU is some low power thing like an A5 (iPad) or the Nvidia Tegra (as seen in Motoroal Xoom Android tablets). Traditional Intel/AMD processors are very different beasts with different instruction sets.
Microsoft currently has no offerings in this market. Their future in this market is based on the ARM “system on a chip” processor. This will mean that sometime in 2012 we get the usual x86 (Intel and AMD), x64 (Intel and AMD) media, along with with a new Windows 8 compile for ARM processors. Various vendors build ARM processors including Nvidia and Snapdragon. It wouldn’t surprise me to see MSFT set out a standard design for a Windows 8 tablet, much like they have done with Windows Phone 7.x handsets to avoid market fracturing as has happened in the Android world.
I’ve seen (but not used or touched) Windows 8 tablets. They are about the same size as Android tablets or iPads. They are nothing like slate PCs.
Ultrabook
Here comes Intel to confuse us some. I guess they are irked by the arrival into the Windows world of a new low power competitor in the form of ARM. The ultrabook is kind of laptop. Quite simply, it is to laptop what MacBook Air is to MacBook. It’s Intel’s attempt (with OEMs) to supply a lighter, thinner, sexier laptop to compete in the lightweight, long battery life market.
Like I said, it’s a mini laptop. It has a Core i5 or Core i7 CPU, GBs of RAM, and an SSD. Typically they are on the smaller side, but bigger than netbooks, at 13.3” (with some smaller variations). And advantage is the presence of a keyboard, which can be detachable. This could leave a touch screen device that might be easily confused with a slate PC.
Here’s some of the events I’m either attending and presenting at. I’d recommend attending these event is you can, despite me being there 
Once again, experts in all kinds of virtualisation technologies will be gathering to share their knowledge at this super-economic mini-conference. I’ll be talking Hyper-V as usual with a 45 minute and a 900 seconds session. Some other MVPs and Microsoft virtualisation experts from around Europe will be there presenting. And as usual, there will be lots of Citrix, VMware and common virtualisation technology sessions.
I’ll be doing two presentations. One will be a brain dump on what’s coming in Windows Server 8 Hyper-V. I say "brain dump" because this topic could do with an entire day and not just a session. My second session will be on the subject of CSV design and backup.
Hyper-V Community Event Amsterdam
Lots of Hyper-V and System Center people will be speaking at this event run by the Hyper-V.nu crew in the Netherlands. Speakers include me, Hans Vredevoort, Ronald Beekelaar, Jaap Wesselius, Peter Noorderijk, Maarten Wijsman and Robert Bakker. Hopefully the place won’t blow up cos that’s a lot of the Hyper-V online community right there!
I’ll be talking about the new networking features of Windows Server 8 Hyper-V.
I finished reading, Kingpin, an account of how a teenager grew up to become the second biggest “carder” hacker in the world. The book is written by a Wired.com journalist, based on the true story of Max “Ray Vision” Butler. His story is an interesting one, covering nearly 20 years of Internet history and Butler’s role in it.
The books is not a page turning thriller book. But so much of it was familiar and that’s what kept me reading. I’d heard of whitehat.com, I remember the huge warnings about a VNC vulnerability about 5 years ago, and I recall MSFT screaming to us about patching an IE graphics processing vulnerability. This book tells us how 1 man made use of unpatched machines to commit cyber crime in the multi-million dollar range. So even though this one won’t have you distracted in the office, it does bring home the reality of crime on the Internet, and how old school hacking for ego has been replaced by hacking for profit. Instead of being the usual fiction, it is based on real history, most of it from the last 10 years.
I haven’t found anything similar to read, so I’ve moved on to reading the “A Song of Ice and Fire” box set which kicks of with A Game of Thrones, the book behind the HBO TV show.
Answers on a self-addressed postcard … or just comment on this post.
I would like to thank Altaro for their continued support. You will be seeing the Altaro advert for another 12 months.
If your virtual machine has 4 vCPUs and it’s running on a multiple CPU host, how many SQL Per Proc licenses do you need to buy (if you’re not doing Server/CAL)? Well this question just came up at work and we wanted to get it right for the customer.
There are two sources of information. This blog post distils it down and this whitepaper explains it on page 5.
Right now, the answer is usually 1 per Proc license per virtual OSE (VOSE or virtual operating system environment – aka a guest OS in a VM) in the Hyper-V world. And here’s why.
Take a 4 vCPU VM running on a host with 2 * quad core CPUs. 4 vCPUs = 4 logical processors. With hyperthreading disabled (or enabled in this case) this VM never runs on more than 1 physical CPU. We can license SQL Server by pCPU. So if it never runs on 1 pCPU then we can buy just 1 copy of SQL Server per proc.
What if this VM runs on a Hyper-V cluster? Do we need to buy 1 per Proc license per host? Nope. We buy the licensing for the VOSE (VM OS).
Let’s change it up. What if the host has 2 dual core CPUs with no hyperthreading? Now we need to use a formula:
VOSE SQL Server Per Proc Licenses = A / B where
In this case the number of SQL Server per proc licenses = 4/2 = 2. And that makes sense; the 4 vCPUs in the VM run on 4 logical processors, and the 4 logical processors are made up by 2 dual core CPUs.
Things are kind of easy right now with Windows Server 2008 R2 Hyper-V (maximum of 4 vCPUs per VM). But what about Windows Server 8 or vSphere 5.0 where it increases to 32 vCPUs/VM? Let’s have a 32 vCPU VM running on 4 * 10 core Intel CPUs with hyperthreading enabled. Ouch, my head hurts already.
The formula gives us 32/20 = 1.6 pCPUs. The VM can’t run on 1.6 pCPUs; it will span 2 pCPUs (we always round up). That means the 32 vCPU VM can be licensed with just 2 SQL Server Per Proc licenses on this host.
Who wants to do division, fractions, and roundups? This might sound like hassle but it’s good because it saves you money and keeps you legal. In our most basic example above, the customer pays for 1 per proc license instead of 4. In the most complex one, they pay for 2 per proc licenses instead of 32.
And that’s how to license a single VM. Things get a whole lot more complex when licensing many SQL VMs and then you start looking at buying SQL Enterprise/Datacenter licensing at the host level, and then throw in virtualisation clustering where your SQL licensing impacts on your virtualisation design so you can save large amounts of money. I covered that one about a year ago and am happy to leave it there for the moment 
I’ll be honest, the number of 64 maximum logical processors in a host isn’t something I’ve worried about too much in the past. I’ve never actually seen a Hyper-V host locally that had more than 2 CPUs/sockets with more than 24 total cores. Other than doing some sizing for a VDI project and seeing a line on hyperthreading in a HP document, I’ve not cared about the 64 logical processor limit in Windows Server 2008 R2 Hyper-V (had to be clear with our new grey area with Windows Server 8 where it is 160/host).
First, what is Hyperthreading? Wikipedia says:
“Hyper-threading is an Intel-proprietary technology used to improve parallelization of computations (doing multiple tasks at once) performed on PC microprocessors. For each processor core that is physically present, the operating system addresses two virtual processors, and shares the workload between them when possible”.
In Hyper-V world, a logical processor is a thread of execution. With a 12 core AMD Opeteron CPU you will have 12 logical processors. With an Intel 6 core CPU with hyperthreading disabled, you will have 6 logical processors. With an Intel 10 core CPU with hyperthreading turned on (2 threads per core) you will have 20 logical processors. Yoiks! And that’s what’s brought be to this subject.
I’m working no a fairly sizeable project, that based on early numbers, seems to indicate that the ratio of CPU to RAM will be higher than normal. After playing with the HP Power Configuration Utility, I found that the cost of owning X HP DL580’s with 4 * Intel E7 Xeons (10 cores each) was slightly cheaper than owning 2X HP DL380/5 servers with 2 CPUs. The big box still cost more to buy/power, but it could require a much smaller network. But all this is meaningless guessing without a MAP assessment/sizing (which is currently running).
But hold on … 4 CPUs, 10 cores each, hyperthreading turned on …. *bang* we have 80 logical processors and we’ve exceeded the 64 LP limit for W2008 R2 Hyper-V.
HP recommends disabling hyperthreading in their monster DL 980 G7 8 socket server. Can you imagine this with 8 * 10 core CPUs? it should be OK for Quake!?!? But that spec would exceed the 64 LP limit so that sucks. In server virtualisation, we’re told that leaving hyperthreading enabled in a modern CPU doesn’t hurt performance and can give a small boost – it can help quite a bit in VDI. I’ll be disabling hyperthreading if our MAP 6.0 assessment says that a 4 * 10 core server is the way to go.
Do you care about this? These big core CPUs are becoming the norm. If you’re using the usual 2 CPU hosts in a 3 host cluster, you won’t care. By the time we get to 24 cores, we’ll probably be restricted by the 160 LP ceiling of Windows Server 8. When I last compared a pair of 2 socket machines to a single 4 socket one, the 4 socket machine cost more to own. That has since changed. If you’re looking at 4 * 2 socket hosts or more, you might consider fewer 4 socket hosts, and then the current 64 LP limit is a factor you have to keep in mind.