Hyper-V and Virtualisation memory management - memory-management

Hypervisors and Memory Management
I have been using virtual machines for years and never really had any issues. I have primarily used VMWare's free single ESXi host and had nothing but success. Because I have never had any issues I have never delved in much deeper. I have however always been very wary of loading the system up and get a lot of spare resources handy.
I have recently purchased a new server and we have decided to give Hyper-V a try and see how that goes. We have a fairly small team but utilise lots of servers for testing etc.
My question relates to memory and how much I need to leave free or available for the host machine to run appropriately.
Setup:Dell Server 24 Cores: 48GB Ram
When I run taskmgr in the windows host instance I see the following:
Physical Memory: 49139
Cached: 14933
Available: 17743
Free: 2982
What exactly do these figures mean? What is the difference between free and available?
My server uses hardly any CPU resources ever and has 10 Production servers running on it without a single user complaint ever about speed of the services.
Am I able to run up another server with 2GB ram effectivly leaving 982MB free? or am I starting to push my requirements a little?
Thanks for the help.

You shouldn’t use the host partition for anything other than Hyper-V (although you can run security and infrastructure software such as management agents, backup agents and firewalls). Therefore, that 2GB recommendation assumes you aren’t going to run any extra applications or server roles in the parent partition.
Hyper-V doesn’t let you allocate memory directly to the host partition. It essentially uses whatever memory is left over. Therefore, you have to remember to leave 2GB of your host server’s memory not allocated so it’s available for the parent partition.


Hardware requirements for a Virtual Server

We have decided to go with a virtualization solution for a few of our development servers. I have an idea of what the hardware specs would be like if we bought separate physical servers, but I have no idea how to consolidate that information into the specification for a generalized virtual server.
I know intuitively that the specs are not additive - I shouldn't just add up all the RAM requirements from each machine to get the RAM required for the virtual server. I can't really treat them as parallel systems either because no matter how good the virtualization software is, it can't abstract away two servers trying to peg the CPU at the same time.
So my question is - is there a standard method to estimating the hardware requirements for a virtualized system given hardware requirement estimations for the underlying virtual machines? Is there a +C constant for VMWare/MS Virtual Server overhead (and if so, what is C?)?
P.S. I promise to move this over to serverfault once it goes into beta (Promise kept)
Yes add 25% additional resources to manage the VM. So if I need 4 servers that are equal to single core 2 ghz machines with 2 gigs of ram I will need 10 ghz processing power plus 10 gigs of ram. This will allow all systems to redline and still be ok.
In the real world this will never happen though, all your servers will not always be running all the time. You can get a feel for usage by profiling your current servers and determine their exact requirements and then adding an additional 25% in resources.
Check out this software for profiling utilization http://confluence.atlassian.com/display/JIRA/Profiling+Memory+and+CPU+usage+with+YourKit
The requirements are in fact additive. You should add up the memory requirements for each VM, and the disk requirements, and have at least one processor core per VM. Then add on whatever you need for the host system.
VMs can share a CPU, to some extent, if you have really low performance requirements, but they cannot share disk space or memory.
Answers above are far too high, second (1 core per VM) is closer. You can either 1) plan ahead and probably over-purchase 2) add just-in-time. Do you have some reason that you must know well ahead (yearly budget? your chosen host platform doesn't cluster hosts, so you can't add later?)
Unless you have an incredible simple usage profile, it will be hard to predict before and you'll over purchase. The answer above (+25%) would be several times more than you need for an modern server virtualization software (VMware, Zen, etc) that manages resources smartly. It's accurate only for desktop products like VPC. I chose to rough it out on a napkin and profile my first environment (set of machines) on the host. I'm happy.
Examples of things that will confound your estimation
Disk space, Some systems (Lab
Manager) use only the difference in
space from the base template. 10
deployed machines with 10 GB drives
using about 10 GB (template) + 200MB.
Disk space: You'll then find you
don't like the deltas in specific
CPU / Memory: This is dev
shop - so you'll have erratic load.
Smart hosts don't reserve memory and CPU.
CPU / Memory: But then you'll
want to do perf testing, and want to
reserve CPU cycles (not all hosts can
do that)
We all virtualize for different reasons. Many of the guests
in our environment don't have much work. We want them there to see how something behaves with a cluster of 3 servers of type X. Or, we have a bundle of weird client desktops waiting around, being used one at time by a tester. They rarely consume many host resources.
So, if you are using something like that doesn't do delta disks, disk space might be somewhat calculable. If lab manager (delta disk), disk space is really hard to predict.
Memory and processor usage: You'll have to profile or over-purchase heavily. I have many more guest CPUs than host CPUS, and don't have perf problems - but that's because of the choppy usage in our QA environments.

Mysql Enterprise Monitor MEM 3.0 Hardware Requeriments

At work, we are planning to have a dedicated box to run MEM 3.0, we will be monitoring about 30 mysql instances. I wanted to know if anyone has any experience working with MEM 3.0 and what would be an optimal hardware configuration for this project. According to the MEM docs:
Recommended System Requirements (if monitoring 100 or more MySQL servers)
4 or more CPU cores
8 or more GB of RAM
Disk I/O subsystem applicable for a write-intensive database (RAID10, RAID 0+1)
If anyone has some experience please advice.
In a previous job we monitored 15-20 MySQL instances using MEM before our needs changed and we went with Monyog(Link). The box it ran on was just a m1.xlarge(4CPU,16gig Mem, 4x420Gig HDs) Amazon EC2 instance(Link to EC2 specs) and we never had any issues. Honestly that box was way overkill for only 15-20 instances. I/O is likely to be the bottleneck far before CPU or Memory. If you are worried about I/O issues you might consider an SSD based setup.
In general with HW recommendations for products like this the recommended is usually overkill, especially if you're on real HW, not virtualized stuff.
If you expect your number of instances to grow I'd go with a HW setup that provides ability to support more than you initially have. At 30 instances you're not going to need a heavy metal type setup, but be aware of scaling for the future.

Cache consistency problem for Virtualization high availability solutions

I took a look at Redhat Enterprise Linux HA and VMWare HA solution. It's easy to understand. Several physical servers can build a cluster. The system monitor the status of all VMs. If one VM is down, it can be restarted on other physical server. The applications on this VM can be restarted, too.
However, I have a question. In Guest OS, when most of applications write data to files successfully, actually the data are sent to OS cache (e.g., page cache in Linux) if direct IO isn't used. If Guest OS is crashed, some data may not be flushed to disks but Guest OS reports they have been successfully written. After the VM reboots on another server, it may find the data in disk is not consistent, and it may encounter data integrity problems. I don't find any information on introducing this problem for VM HA solutions. Don't we need to consider this issue?
I was asking a similar question here:
Found this on VMWare, not sure how much it applies to your situation:
VMware hosted products on Linux hosts always use buffered IO … When using buffered IO, VMware hosted products do not bypass the host's buffer cache to produce crash consistent virtual machine IO. Consequently, if the IO is buffered within the host operating system, an application running inside a guest operating system on VMware hosted products might lose crash consistency.

Oracle DB on solaris utilizing swap memory when free RAM available

We have a weird instance where we noticed our oracle database server swap utilization was 100% and surprised to see that the system had free memory available during that period. To my knowledge, swap memory utilization starts once system runs out of free RAM (please correct me if i'm wrong). Not sure what could have caused this unusual activity. Had anyone else experienced such behaviour?
If you are using solaris 10 or later, system parameters are not in /etc/system anymore. Instead, you configure dynamically managed resources through "projects" at the zone, task or user level.
Here is for instance one of these project config files in one of our zones.
The settings here mean that a user of username oracle will benefit of a specific resource setting.
To set add such a line to the project file you use the projadd command. To modify it the projmod command and if you want to list all the resources currently available to oracle you would
su to oracle
prctl -i project user.oracle
This will list all resources available to that 'project'.
You should provide details on how you get your metrics but there is no direct relationship between swap and RAM utilization. You certainly can get out of swap space with free RAM still available.
There must be enough swap for all (virtual) memory reservations to fit. Memory reservation (i.e. malloc) use no RAM. RAM is only required during virtual memory access.
Oracle DB running on Solaris has a "bug".
It is "reserving" swap on an Operating system where "swap" and "ram" are combined into one virtual memory pool.
Reserving swap just exhausts the disk based swap area for no reason and should not be doing this at all.
All transitioning between ram and swap is handled by the Solaris OS itself.
It took me almost 5 years to get Sun/Oracle to understand the exact same problem with Java.
Latest Java 7 and 8 no longer "double dip" into the memory pool.
What I've found, in cases where the OS runs into a "Swap" limit is to size the memory accordingly, and remove swap altogether.
With zero swap available, all memory interactions are in physical ram and you won't see the double dip issue, and you won't see errors about swap exhaustion as long as you have sufficient ram for everything to run.

Virtualization and why it is good for programmers

Why does it help to know about virtualization from a programmer's perspective? Except testing and developing on several different platforms without the need of switching between operating systems is there a particular reason why virtualization is important for a programmer? Are there any details that must be kept in mind before developing on virtual instances?
I use it for testing our installer, because it is important to check whether the application will work on a clean installation of the operating system.
I used to do these tests by keeping a hard drive with a fresh operating system installation and making a copy of that disk for (almost) every new test run. This was very time consuming, and the virtual machine solution has saved me a lot of time. Note that this even allows you to do remote debugging as easily as when using two non-virtual machines.
Note: If you're interested, I'm using VirtualBox, which is a very good and free virtualization tool.
If you develop a driver or something very close to the hardware with a high risk to crash the machine, you will be glad to be working on a virtual machine.
Reverting to an old state is easier than to repair a damaged OS.
One of the main advantages is having your entire development environment as a single image file. I have a perfectly configured version of Windows Server, Visual Studio, ReSharper, etc. I can easily try a new version of something on a copy of this virtual machine without worrying about it causing problems.
I can also back up my entire dev environment to transfer it to another physical machine very easily. I've been through 3 machines in this office alone so that was a lifesaver in itself.
The only real trade-off I see is performance. You generally have to use less physical CPU cores than you actually have and less memory. With a sufficiently powerful machine this is not much of a problem though.
Edit: As nader said, I/O is obviously important for most projects as well. Although developing on a virtual machine does mean a fairly large I/O penalty compared with a native OS install, in practice I rarely find it to be a problem. The superior random access capabilities of SSDs are helping to mitigate this drawback as well.
Being able to completely reset the state of the system is very useful to debug applications which modify their environment - If the actions are repeated after a reset, and they're constrained to the sandbox environment of the VM, you are pretty much guaranteed to get the same result.
We have a large number of different versions / customer customisations of our software, and its not possible for 2 installs of our software to coexist on the same machine. Virtualisation allows us to replace the 50-60 physical machines that we need to maintain for testing and problem reproduction with 2-3 virtual servers - it takes around 10 miniutes to make a copy of a VHD template we have and create a new virtual machine, and as long as you allocate 1-2Gb of RAM the performance is comparable to that of a (slow) physical machine.
Virtual machines are also great for build machines.
Personally I do all of my development on my deskop machine for best performance, and remote debug into VMs. I dont run virtual machines on my desktop as it uses up too much RAM, we have dedicated virtual servers for that.
Good for developing, because you have same server configuration in virtual machine like on production server.
From a user space application there should be no difference developing for a virtualised OS versus a normal OS. There may be some gotchas if your code makes explicit assumptions of the machines memory size and number of processors and believes what the hypervisor tells you.
I'm surprised no one has mentioned the ease of deployment. All you need to do is get the build down on the virtual O/S and then you can copy the image to as many new servers (running some kind of virtualization solution [like VMWare]) as you want, easily scaling your application.
Record the state of a bug in a program, and send it to the developer (along with the entire "machine").
Testing your code on various O.S, some of which you don't have.
Working in a more protected environment, making sure that the code doesn't harm your system -useful for understanding dangerous programs, like viruses, and developing security against it, for writing potentially wrong hard-drive programs, and anything that can have catastrophic effects on your system.
Easily Write your own O.S without the need to write on 'real' boot sectors, a potentially harmful act (Hope this is not new...).
Quickly use tools and programs not found on your own O.S.
Demonstrate a program at various times, by restoring a virtual machine,
quicker and less prone to failure, than trying to recreate the state at the minutes before the demo.
Less directly connected to programing, but surfing vie a virtual machine (for example to see documentation) has the added value that your own important system (and code) is less likely to be harmed by malicious programs.
From my experience in most cases the answer is typically "no" (When testing and targeting multiple platforms is removed) Both are huge reasons to be familiar with "desktop" VM solutions. Others have done an excellent job of listing rare exceptions like debugging kernel codes.
There are some quirks one must be aware of when running on a virtual machine. This is hardly an exhaustive list:
Loss of precsision or even time reversal in high resolution timers due to emulation of hardware resources (depends somewhat on the vm platform and operating system)
Virtual network interfaces ususally bridged. We've seen some extremely odd behavior in the host system with an application that sets up its own bridge between virtual interfaces -behavior which logically should not effect the host in one of the leading VM solutions.
Usage models - If your product has orwellian licensing codes or records state dependant behavior when interacting with remote systems you should account for what would happen if a system were "paused" and "restarted" or restarted from an earlier "state". Normally this kind of thing would be taken into account anyway in a robust implementation.
If you are developing in a virtual environment you will want to make sure you know what specifications were used to create the environment. If you have say a 4 Gig machine and create a virtual environment with 1 Gig you will want to make sure things in your development do not grow to a point that it overruns the memory. This will cause slight performance problems. I personally ran into this and it was a pretty tricky thing to track down. The scenario was that I was fixing a bug and testing it in a virtual environment. I did not setup the virtual environment by the way... The application took a performance hit because of all of the memory swapping that was taking place.
A very good use for a virtual environment is when you are developing applications that mess with the Windows Gina. It's much easier to reinstall a virtual environment than an entire PC....(been here done that too).
I do all of my development on a virtual XP instance under VMWare Fusion so that I can use a Mac for everything and still write .NET code ;-)
Sometimes they are necessary, because the platform you are programming doesn't support the standard developer environment. One such example is Sharepoint. As of Sharepoint 2007 you still need a server OS to install Sharepoint 2007, WSS, and the Visual Studio Sharepoint Extensions (VseWSS).
Thus for Sharepoint I have to use a Window Server VM to do my development work. As for Sharepoint 2010 they are supporting installations on Vista and 7 x64, but I will still use a VM, because I don't want to have Sharepoint on my main machine slowing everything down. Rather I want it in a VM where the services are on when needed and off when I don't without having to manually turn off/on each service. This in addition to the many great answers posted above.