realMarkusSchmidt's questions

I'm analyzing a problem where performance of CPU-bound workloads inside virtual machines often (not always) is way below what we would expect based on the underlying hardware.

We're using Hyper-V on Windows Server 2012 R2. The server has dual Intel Xeon E5-2643 v2 @ 3.50 GHz.

Here are some figures that seem to be relevant:

• Hyper-V Hypervisor Logical Processor, % Total Run Time, Instance _Total: Average 20%
• Hyper-V Hypervisor Virtual Processor, CPU Wait Time Per Dispatch, Instance _Total: Average 20000 (this number seems to be totally on the safe side, so it doesn't seem like the hypervisor has to "steel" from virtuals CPU to schedule time to logical CPUs of another VM; seems to translate in an overhead of 2%)
• Hyper-V Hypervisor Logical Processor, % of Max Frequency, Instance _Total: Average 34%
• CPU-Z tool shows most of the time around 1200 MHz for Core #0 of both processors (pretty much matches the % of Max Frequency reported by Performance Monitor)

On a desktop with only a few cores, core speed goes up immediately as soon a CPU-bound activity starts.

On our Hyper-V hosts however, core speed seems to go up only if overall system load seems to be high for a few seconds. Now e.g. if you have a VM with 4 virtual CPUs out of a total of 24 physical (with Hyperthreading turned on), and this VM needs CPU power and Task Manager inside the VM shows nearly 100% CPU usage, most of the time the clock speed of the physical CPU won't go up and performance is bad.

Obviously this is unwanted behavior. Think of a database server that needs 3x the time to answer a query because the server has not "enough" load to step up CPU frequency. That doesn't make any sense.

I found a blog post describing the exact same behavior for VMWare and Cisco blades, from 2011. I didn't find information on this anywhere else.

I was actually able to get rid of this behavior by switching to the Windows "High performance" power plan in powercfg.cpl, at the cost of around 30% higher power usage. I actually get better and more consistent performance and Performance Monitor shows lower load figures.

(On an older server, I found an additional setting "Processing power management | Minimum processor state" which could be set to 100% without disabling all other power saving options. The new ones show only "system cooling policy" which is on "Active" even for the "Balanced" plan, so my only option was to choose "High performance".)

Is this really best practice for Hyper-V hosts, or is there any other workaround? If SpeedStep is really a problem, I wonder why they even build it into server CPUs and enable it by default and why I never read about this setting in a Hyper-V configuration guide?

This is SQL Server 2008 R2 running in Hyper-V virtual machine with 4 virtual cores.

Task Manager is reporting sqlservr.exe to use at least 50% CPU, sometimes 100% for minutes. I/O is very low. The top "Recent expensive queries" listed in Activity Monitor are reported in the range of 100 ms/s and pretty low figures for Logical Reads/s as well. Number of threads seems okay (around 180 for around 120 open connections, very few of them running).

Maintenance Plans are running normally: log backups every 15min, running only for a few seconds.

Any idea what could cause this high CPU load?

Update: It seems like I found the issue by analyzing a short unfiltered SQL trace. A stored procedure, repeatedly called by a background service, issued a SELECT statement suffering a typical "parameter sniffing" problem. After fixing this, CPU usage came down.

So I want to rephrase the question a bit: Is it normal, that stored procedures or the statements executed by stored procedures never show up in "Recent Expensive Queries"?

I'm migrating virtual machines from Win2008R2 hosts to a Win2012R2 host. Since there is no direct import/export path due to the WMI namespace change ("did not find virtual machines to import"), I recreate the VMs as new (Gen 1) and attach the old VHDs.

However, even though I set the network adapter MAC address to static using the MAC address the adapter had in the Win2008 config, after upgrading VM additions, the old device is "disconnected" and a new (and unconfigured) Hyper-V network adapter shows up.

I would really like to avoid that since the network setup of some of the VMs is complex. How can I get Hyper-V to present the network adapter in a way that the guest OS recognizes it as the old adapter and keeps all the settings?

I need to connect a Win2008 R2 server to a iSCSI target and I'm supposed to use a CHAP user name that is different from the initiator name.

Technet says:

By default, the Microsoft iSCSI Initiator service uses the initiator node name as the CHAP username. You can override this action by passing a CHAP username to the login request.

Now my question is: How and where can I override this? At least using the GUI, I couldn't find a way to specify a CHAP user name.

Web PI seems to make it very easy to install popular open-source apps and keep it up-to-date, even if based on a non-MS software stack (like Wordpress).

Given I'm only familiar with the Microsoft stack (IIS, ASP.NET, MSSQL etc.), and not PHP, is it a good idea to take a small virtualized Windows Web Server instance, and use Web PI to set up a public installation of say Wordpress and its dependencies like PHP etc.? Does it follow common practices concerning security, stability etc.?

Or is this only intended for developer machines and I better let someone with a better understanding of PHP do a "hardened" manual setup?

I should add that no other software would be deployed on this VM, so I'm not afraid about Web PI messing with my custom IIS configuration.

I’m running a ASP.NET application on Win2003 32bit that tends to allocate a lot of memory. To avoid OutOfMemoryExceptions that start to appear when the W3P process exceeds 1 GB, I have set a maximum virtual memory limit of 850 MB for the this application pool, so the application will restart instead of throwing exceptions.

While not throwing exceptions, these restarts still lead to temporal performance degeneration (low cache hit ratio, aspx compiling etc.), so I’d like to minimize the number of restarts. Watching the performance counters I can see that the ASP.NET cache works as designed (item count and cache hit ratio increasing after start, after I while some items get removed probably due to expiration dates), except for cache trimmings.

E. g. I would expect cache trimming to kick in when e. g. 750 MB are in use and half of that is cache. But it never happens, under load it continues to allocate more and more memory instead of e. g. trashing low prio cache items, until it hits the memory limit.

Does anybody know what would actually trigger cache trimming? Available physical memory? Is there any other setting available to optimize this scenario?

(Yes, I know, 32 bit is so 2004 and with x64 this problem would go away, probably even with the same amount of memory, but for this app I’m stuck with this server for another few months and have to keep it up running...)

I have a database with a size around 100 GB, and I have serious problems expanding the size of the database. This is my setup:

• SQL Server 2005, Version 9.0.4035
• OS: Windows Server 2003 SP2 x86
• Running as a guest on Hyper-V
• Data file and log file on different disks (virtual and physical)
• The disk for the MDF file is 500 GB and has nothing else on it, so no fragmentation problems etc.

At first, I had default settings for auto-growth by 10%. That started to fail with timeouts when the database was around 30 GB big or so, so I changed it to 1000 MB. Now even this failes, so I tried to increase the file size manually. I had to cancel my attempts because they "never" completed until I went down to a increase of 50 MB, and that took over 22mins! :-o Something must be seriously broken...

I've been monitoring the Average Disk Queue length for all disks (system, data and log), they were all idle except the disk that holds the log file, this one had always a disk queue length over 10. Before and after this operation it was idle, too.

That's the command I used to increase the filesize manually (I added 50 MB to the initial size):

ALTER DATABASE [MyDB] MODIFY FILE ( NAME = N'MyDB', SIZE = 103987200KB , FILEGROWTH = 51200KB )

Edit: When I created this database months ago I initialized it with 10 GB, and I can't remember that I had any remarkable delay. I can also e. g. restore other databases on this server with 2 or 3 GB, and it just works fine. The I/O performance on this machine is not superior, but far from that bad.

Any ideas? Thank you very much in advance!