The story of a difficult experience with Hyper-V
If you follow my Twitter feed you knew this article was coming…
After spending about 3 hours on the phone with a customer trying to help them get their CSV (Cluster Shared Volume) for Hyper-V back online (I was on the phone not because I’m a Hyper-V guy but because I implemented the storage they use for Hyper-V) I decided that maybe I should do a deep dive and learn a little more about how CSV’s work so that I can better compare them to VMFS. VMFS is the clustered file system that VMware uses to share SAN LUNs between physical VMware servers. This article also goes on to explain why Hyper-V is an inferior solution due to some of the other issues that I’ve seen in my limited experience with Hyper-V.
Anyhow, before we get to the technical stuff I need to point out that I am obviously biased to VMware… it’s what I do every day. I will try to be as fair as possible, but lets face it if you are just looking for the take away without reading the whole post it is that VMFS is far superior to CSV’s, and that while VMware might look more expensive on a bill of materials it will probably save you time and money in the long run.
The Technical Stuff
Cluster Shared Volumes have been around a long time, Microsoft has been using them for everything from Exchange clusters to SQL clusters. They adapted CSV’s to work with HyperV so that they could allow virtual machines to more easily move from one HyperV host to another, similar to what VMFS allows VMware ESXi servers to do. Both enable high availability for virtual machines because if a host fails, other hosts can access the virtual machines the failed host was running. Additionally CSV’s are needed because under them is NTFS… which was never designed to be accessed by multiple systems at the same time, because of this something had to be put in place to allow that to happen.
Ok so the first article I come across on Technet has this to say:
“…the Cluster Shared Volumes feature included in failover clustering is only supported for use with the Hyper-V server role. The creation, reproduction, and storage of files on Cluster Shared Volumes that were not created for the Hyper-V role, including any user or application data stored under the ClusterStorage folder of the system drive on every node, are not supported and may result in unpredictable behavior, including data corruption or data loss on these shared volumes. Only files that are created for the Hyper-V role can be stored on Cluster Shared Volumes. An example of a file type that is created for the Hyper-V role is a Virtual Hard Disk (VHD) file. Before installing any software utility that might access files stored on Cluster Shared Volumes (for example, an antivirus or backup solution), review the documentation or check with the vendor to verify that the application or utility is compatible with Cluster Shared Volumes.” Taken from: http://technet.microsoft.com/en-us/library/dd630633%28v=ws.10%29.aspxSo to me, that means that CSV’s are flaky, to say the least… but let’s continue.
VMFS, on the other hand, can store pretty much anything you can upload to it… zip files, iso files, etc etc… VMFS is almost like LVM in linux, it doesn’t care what you put on it.
My Next Point
All HyperV nodes that are using a CSV are at the mercy of the coordinator node for that CSV. Think of it this way, you need to look for something that is in a filing cabinet, but before you can actually get the folder you need you must first talk to the secretary and ask her if its ok to look at the folder. In more technical terms this means that the coordinator node keeps track of all the metadata and file locking on folders on the CSV, after the coordinator node allows you access to the folder then IO to things in that folder happen directly to the LUN. But don’t take my word for it… Microsoft explains it in this article http://support.microsoft.com/kb/2008795
What scares me about this method is that if something gets hosed with the Coordinator node and the Failover Manager doesn’t fail over properly your CSV is inaccessible. And going by the reliability track record of Microsoft Services I would not bet my job on Failover Manager 🙂
VMFS, on the other hand, is a clustered file system that has no owner… there is not anyone node that controls access to the file system. File locking is done at a file level by a ‘pulse field’, and in this field, a host must periodically update its time stamp and let the file system know that it is still using the file. If a host crashes and another host wants to use the file the host can let the file system know that the timestamp hasn’t been updated lately and that it is taking over ownership…. this means that each host can access files in the event of a node failure without waiting on a response from a centralized management node. If you want the in-depth answer check out this article http://blogs.vmware.com/vsphere/2012/05/vmfs-locking-uncovered.html
Next, let’s talk about how to get HyperV to actually work …. and I’m not talking about just getting a VM to boot up. I’m talking about setting up HA and DRS and automatic load balancing etc. With VMware you group all of your physical servers into a Cluster and then check two boxes… one to turn on HA and one to turn on load balancing. Of course, you need to set up network interfaces for vMotion… but other than that your done. Oh, and by the way, this was all done from the VMware vSphere client.
On to HyperV… lets see where do we start…. no really… which interface do you want to start with. If I want to configure virtual machines I’ll need to use HyperV Manager… or maybe Systems Center Virtual Machine Manager (SCVMM). If I want to set up CSV’s to allow for HA to take place I’ll need to fire up MS Failover Manager. On and if I want to actually format a LUN, well then I’ll need to get into Disk Management.
I’m sure you get my point….
We could talk about virtual switches and distributed virtual switches (or the lack there of), or things like NIC teaming and how it is implemented, but I don’t want to write a book.
The take away
OK so normally the number one reason I hear that people are using HyperV is because it comes with Windows … Its free. But is it really free? I could argue that it takes less time to implement VMware than it does HyperV… and isn’t it known that “Time is Money”… it certainly is if you are paying someone to set it up.
I could also argue that 3+ hours of downtime trying to resolve a CSV issue where VM’s are not accessible is definitely a loss of productivity and in turn a loss of money.
Lastly SCVMM is not free and while it is not required, if you want to compare apples to apples you will want it…
So in the end VMware may be an additional line item on a bill of materials, but in the end it may be the best damn investment you will make for your virtual environment.
If you are not convinced yet here are some other fun stories about the HyperV
- http://www.ms4u.info/2011/05/why-you-should-not-running-domain.html <- Failover manager required AD, if you virtualize all your Domain Controllers Failover Manger wont start.
- http://blogs.technet.com/b/chrad/archive/2010/07/15/cluster-shared-volumes-csv-extending-a-volume.aspx <- Expanding a datastore in vmware is as simple as right clickign the lun to extend, selecting expand, clicking next like 3 times…. check out this link for the process for a CSV 🙂
- do a google search for HyperV live migration using multiple NICs…. you wont find much, so if you buy that server with 128 or 256 GB of ram and need to take it out of production for maintenance… better have 10Gbps networks in place or grab dinner and a movie… (1Gbps network can theoretically move 125MB/sec or 7.5GB/minute so to move 256GB that would be about 34 minutes… so if you want to do a rolling outage multiply the number of hosts you have by 34 minutes to move all that ram…. just one more way VMware costs you less)
- Look up how to do nic teaming in HyperV… Then look up how to do it in VMware
The bottom line
If you want a solution that is straight forward and easy to use then HyperV is probably not the way to go. While it may not be a big line item on a bill of materials, what it saves you there it is certainly to cost you in time and effort (not to mention frustration).
I’m now in a VMWare shop. Most clusters are up to 5.5. Prod isn’t due to a emulex driver issue with Cisco 1k and Cisco 5K. Hope a new driver will resolve issue.
I’m getting used to VMware. Frankly I just think hyper-v is easier. My last environment used HP. HP EVA’s for storage and DL580’s with 4 10 core CPUs and 1TB of ram. Had four DL580’s per EVA 4400 with 40TB of disk. Making a fully redundant solution. Everything was clustered that could be or had another VM running on the other SAN&nodes. Could lose a SAN and everything was up.
Anyway, we didn’t have to much trouble getting everything working and everything was working very well. But I think getting the right equipment was part of the equation. I keep seeing hyper-v folks running cheaper crappy gear. I think that contributed to bad performance in the past.
When you have a host with 1TB of RAM and your not oversubscribing anything stuff works right.
CSV were maybe tricky to setup but not impossible. Once you get it it’s easy to replicate.
I don’t like the networking in VMware. I don’t see it as useful. I much prefer just add HBA’s and NIC’s and let Cisco switches do the network magic. No idea what people are talking about teaming issues. We did it in 2008R2 on IBM m2 servers… And did teaming with HP servers as well. Had a nightmare with teaming on an IBM bladecenter 4 years ago but that’s behind me.
As I recall network were dual 6509 with 10gig Connected virtualized supervisors that allowed for a whole chassis failure and servers stayed up. We were able to move massive amounts of data in a short amount of time.
Now I do remember live migrations to take a while to move. If things were like exchange we maybe would just shut down a VM and bring up on another host if we were in a hurry. Nobody noticed because another mailbox server or CAS was on the other cluster (host and SAN) and everything was A-ok. It was about service uptime… Not server uptime. I don’t care if its a SQL, Exchange, or Lync there is a way to get maximum uptime. It don’t matter if you use VMware or Hyper-v that’s just the tool you use.
What I will say is that vsphere is easy pane of glass to manage…
But here is where I’m impressed with hyper-V. Shared nothing live migrations. That is way fast in my world and way fun to do. Where is vmotion of shared nothing storage? That’s where hyper-v is going to turn heads if you ask me. Shops that can not afford a SAN should look at hyper-v and give it a shot.
With server 2012 R2 hyper-v is told to be even better. I have yet to upgrade my lab for it yet. I plan to do so in the near future. I just need some free time.
Shared nothing migrations have been in vsphere since v5.1
Hey Sean! Thanks for the note… As for shared nothing migrations, if you check on one of your vSphere 5.5 clusters … in the web interface you will be able to do a vmotion+storagevmotion at the same time from not shared storage to not shared storage.
Good analysis. Hyper-V 2012 sucks and does not even come close to VMware workstation version from 5 years ago in stability.
LOL, well I dont know if i would go that far. But I appreciate the enthusiasm!
Even if I´m definitely NO Hyper-V fan at all I wouldn´t go that far either. It is actually possible to get a stable running hyper-v environment but it is a hell of a job. 🙂
Depends heavily on the hardware compatibility and a bit of luck but it can actually work.
The last time we tried to implement a hyper-v environment it was to be deployed on a Dell VRTX. Sounded easy enough and as good as made for it but we stumbled over the most common things to VMware for years in the very first seconds of the deployment. First, Windows doesn´t accept shared SAS storage without implementing a registry key by hand and after that we had to realize that the storage concept of the VRTX wasn´t meant to be used for the shared quorum volume (forgot the correct name already, sorry for that) because it needed a dedicated LUN that was not possible to configure on the VRTX at all. That peace of junk can only deploy whole RAID Groups (and really I tried everything I could) so we had the option to waste a single disk or even RAID 1 double disk to that totally useless configuration step or just try to implement the quorum device externally into the configuration.
Vmware would have been up and running in a few minutes, Hyper-V took us DAYS to get it running on that “meant for virtualization device”… One more reason to stay with the solution that has +10 years of experience in that topic…
LOL
We’re using both VMware and hyper-v depending on the customer/deployment and while there are certainly differences/pros/cons (and sure VMware is definitely more feature rich) it sounds to me like your hyper-v issues stem mostly from not knowing how to implement it. Some hyper-v evangelist would probably say the same nonsense about VMware… If you are having a hard time getting either working than you don’t know what you’re doing, sorry.
I do wonder how using vmm or hyper v manager is conceptually different than using vsphere though.
‘Windows doesn’t accept shared SAS storage without implementing a registry by hand’ – wtf are you talking about?
Josh, I wont disagree with you. My HyperV knowledge is limited. But I have to think that out of the box vSphere is much easier. Hell When I was working with Xen it even seemed much easier to get going that hyperV
Sorry, I’m with Josh. And justin in my world I found hyper-v to be easier to implement over VMware.
However SCCM and Virtual machine Manager were a little bit of a pain to get working. Without those hyper-v is all manual, hands-on.
I dislike the virtual network switch in VMware. It’s more complicated than I need it to be. Hyper-v is easier to me.
I found hyper-v to be easier to work with fiber-channel SAN’s over iSCSI ones but with 2012/r2 that doesn’t seem to be the case. El-cheapo SAS HA storage solutions did not work very well with hyper-v. Didn’t try those with VMware.
VMworkstation seems like garbage to me, hyper-v role in windows 8 works Much better to me. So I’ve enacted test VMware servers to test with, workstation was so slow it was unusable after 4 VM’s started. That was kind of a waste if hardware for testing.
VMware is slow to boot and come up. Much slower than hyper-v.
I don’t like that you can’t live migrate between clusters of VMware. (Different vcenter’s)
I have other small complaints.
Hyper-V easier to implement? Really? I’ve worked with earlier versions of Hyper-V (up until W2k8R2) and, unless you use it as a souped up version of VMware Workstation, it’s way more difficult to get to work. The fact that you need SCCM and SCVMM says it all and let’s not get started with the additional components like the cluster service.
If you dislike virtual networking in VMware, it’s time to attend a training. Do you really think you can configure (virtual) networking if you don’t know what you’re doing? I think you’re suffering from something I’ve seen in the early course materials – MS took the top-down approach (hey it’s a server and now it also does virtualization) versus VMware the bottom-up approach (what’s virtualization, how do you properly configure networking& storage, now we can do neat things with VMs).
In order to not create a mess, you need a solid foundation in networking, storage and server configuration in general. Microsoft can try hard to shield users from reality, but that will not work. On the contrary, if something doesn’t work the way it should, you have a hard time figuring out what is going wrong and why.
Not being able to move VMs live between different vCenter environments (totally different from clusters) is logical and clearly shows you don’t have a clue about how VMware works and how it is used. Customers are using this type of setup for different environments, which are not allowed to integrate, like Production versus DMZ or Production versus a VDI environment (the latter also has licensing implications). If you wanted to live migrate VMs, you would run them in the same vCenter Server instance.
drexciya: I take your comments as a little harsh so you know. I think you should soften your writing style on a public blog, telling people they don’t have a clue isn’t very polite and certainly not professional. Do that with someone like me in a professional setting and you would regret it.
I’ll comment on why I want to Live Migrate VM’s between vCenter environments: Where I work right now we have to do it often and would do it more often if we could. We may want to move between hardware setups. (PROD/DR/UAT/DEV/TEST are all on different Storage, networking and server hardware and sometimes in different physical locations. Yet we have the need to move a VM from one place to another without downtime. We can do that on Hyper-V.
I also prescribe to the “old school” thought that network folks should do networking and not server folks. So I don’t exactly want to be doing networking if I don’t have to. In hyper-V you take network interfaces and create a Vswitch and connect your vm’s that need to be on that switch and your done. The network team manages the network. If you have multiple 1GB or 10GB or even CNA great. The Cisco guys can do more with their 6500’s or Nexus switches then I care to try to do with the 1K or VDS in VMware. (5.5’s VDS is certainly better than previous versions)
Saying Microsoft is shielding users or that it won’t work is crazy talk. It certainly will work.
In my last life, I had 4 2008R2 nodes on it’s own SAN and another 4 2008R2 nodes on it’s own SAN. Everything was connected to two cisco 6500;s that were virtualized so a whole chassis could be lost and nothing would stop. Everything that could be clusters (Microsoft clustering) was and services would stay up even if we lost 1 SAN, the 4 hosts and 1 6500. infrastructure had dual power sources and paths for everything. I have to say it was very slick and all hyper-V. It worked really well and it wasn’t very hard to build. My previous DataCenter was with 2008 and that was a little harder and previous to that was 2008 and was for a remote office and ran into issues with SAS storage and hyper-V. Yes teaming was a big pain in the ass till 2012/r2. but that’s fixed now.
My point is that I didn’t need to have networking in the Hypervisor and everything was great. I didn’t create the virtual networking I simply created a Vswitch and made sure it was on the right interfaces and life was good. By the Way performance rocked!
I don’t see a need for it (networking) in VMware but it’s there. Why have a complex setup in VMware and have expensive networking hardware sitting right next to the servers? then you have networking folks saying they want to manage the switches in VMware and Server teams that don’t want them in their VMware… we could talk hours about all that… What works for your shop is what works…
Then your going to bash SCCM and VMM? IF your not running SCCM already then I don’t know what your doing? Most large shops are running SCCM already and are VMware shops. Maybe VMM is an extra cost but its not very expensive if you ask me. There are a few alternatives to SCCM but SCCM use is growing not shrinking. So don’t complain about SCCM.. that’s silly if you ask me.
Look I work with VMware today and I don’t see that changing but I don’t “love” it and I doubt I ever will. It gets the job done… but I see Hyper-V as the product with a better future.
Don’t like my opinion? Fine. But don’t act like Socrates and try and belittle my opinion by slinging mud…
happy St. Patrick’s day.
Just in case that you still want to know which registry key needed to configure the shared sas perc in a Dell VRTX.
That one:
1. Open the registry using the run prompt.
2. Type REGEDIT and press Enter.
3. Navigate to HKey_Local_Machine\SYSTEM\CurrentControlSet\Services\ClusDisk\Parameters.
4. Right-click in the right pane and from the context menu select New > DWORD (32bit) Value.
5. Type AllowBusTypeRAID in the Name field and press Enter.
6. Right-click the new entry and select Modify.
7. In the Value Data field, type 1 and press Enter.
8. Close the REGEDIT window.
Without that, the config is not running well at all. I wouldn´t call the Dell VRTX a cheapy crappy shared SAS system as well.
And the thing with that different clusters and migrating VMs between clusters is actually two different things between VMware and Hyper-V, my opinion.
VMware allows to build the different “Clusters” in one vCenter environment just because it HAS the networking feature. In Hyper-V it doesn´t really make sense to do that kind of configuration so you compare apples and peaches VMware is actually placed on another layer. Hyper-V sits on top of the Windows OS layer. VMware sits on top of the hardware layer so it´s important to have control over anything beneath it. Hyper-V does not need to have control over the network layer because windows already did. I hope I could make my point of view clear without insulting anyone. 🙂
…one addition to the clusters topic…
You could actually build that different cluster environment within one host. All you need are enough nics and or vlan config.
We have multiple vcenter’s to segment environments as well as for managment reasons. One being, not having one vcenter in case that environment fails and you loose your vcenter more multiple environments.
As far as one host that would not satisfy HA requirements.
Thanks for the ResEdit steps for the SAS storage. The SAS storage I was working with was crappy. It was part of a bladecenter “datacenter in a box” project for a remote office in Australia to house services for the PAN rim. That SAS solution was craptastic, 🙂
I’ll read up on the dell VRTX, lately we have been buying Equalogic for gold tier and emc VNX for Platinium tier.
The moving VM’s that I’m excited with came in 2012 hyper-v. Called shared nothing live migrations
You´re welcome.
Shared SAS works quite good for small VMware Environments but is a real pain with Hyper-V. Shared nothing migrations is a Hyper-V 3.0 and vSphere 5.1 feature, so no unique feature for any side. I think that Microsoft has made a good step forward but is still not yet where VMware is. It is a different approach to virtualize and with upcoming new releases of windows once a year, I think MS might catch up pretty fast. Sooner or later it will be just a question of personal preference and not because of features.
Sean, I’m sorry I was a bit harsh. I was a bit annoyed about certain things which were touched upon in your post. I’ve worked with both Hyper-V as well as VMware and in my experience virtual networking in VMware isn’t easy, but it’s more powerful then in Hyper-V.
Also, and this was the thing that really inspired me to take up virtualization; VMware more or less forces you to get into networking and storage as well. Okay, you won’t become an expert, but you have to know “enough” to be able to talk to experts. This makes things way more interesting and you really have to know about the interaction between the various components to make it work.
This is where Microsoft is lacking in my opinion; by hiding complexity, you don’t remove the need to know your stuff. Especially in the beginning, Microsoft had a tendency to downplay the required knowledge for Hyper-V.
As to easier or not; I guess it depends on where you come from. I see the need for any additional components (SCCM, SCVMM, SCOM) as an additional burden and it’s actually hardly easy to get competent people for managing that. You see this with VMware as well; if you really start adding more components (vCOPS, vCloud), things become much more complicated. The fact that you have most functionality with vSphere and vCenter without anything else is a big plus.
As to having multiple vCenter Servers; separation of production and test environments is a design feature and if you would be able to live migrate VMs, I’m not really sure whether you are breaking some separation rules or enabling actions which can result in problems. People I talked with, who worked in environments that had such a separation in place (finance, IRS), would NOT allow that at all (it’s a very strict organizational rule in this case). In other scenarios, like a separate VDI environment, there’s also a clear preference for separating virtual desktops from production systems.
That’s why I was rather surprised by your desire for cross vCenter vMotion actions. If the need for physical separation isn’t that strict, you might merge environments and bring about separation in another way. That would be something like the management versus resource cluster model. Separate clusters within a single vCenter server instead of having multiple vCenter Servers. When properly configured, you can get separation (permissions as well as networking) but still have all the functionality available in one vCenter Server environment.
I have been working with Hyper-v for the last 3 years…don’t really have too much experience with VMWARE…
I agree with most of the things you said…
in my experience Hyper-v Clusters aren’t too bad if you don’t have too many Vm’s…. If you are going to have 100’s of VM’s then prepare to be frustrated, angered and stressed dealing with stability and backups related issues.
I’ve worked with VMware since 2006 and love the product. It’s not only been a transformative product for IT, but for my career as well taking me from a Linux admin in 2006 to a consultant specializing in virtualization working for a VAR today. Back in the 2008 and 2008 R2 days, Hyper-V was a joke and I didn’t think much of it. When I heard about the features coming in 2012, especially SMB3, Live Storage Migration, and so on, I decided to take a closer look and get it running in my lab.
Many of my clients have started asking me about Hyper-V and, having run it in my lab at home and the lab at work, it really is a cool product. VMware is the feature leader and is without a doubt easier to set up and configure, but many customers are taking the “good enough” approach when they see they can save money on licensing costs. However, I’m quick to warn them that there are hidden costs as well: OPEX to train or hire personnel to administer it, risk of downtime since Hyper-V is more sensitive and, in some ways, more complicated than VMware (AD requirements, permissions, etc), the cost of the entire System Center suite if you want VMM, and the variable support costs that come with MS software.
For me, I see Hyper-V as a really good product that has come a long way in the last 2 years and I welcome some competition in the hypervisor market rather than having one company dominate it.
We completed our migration of Hyper-V VMs to 2012R2 cluster before few months ago. Now new issue came up, VMs loose network connectivity randomly. If we vmotion them to another host it works for some time (may be for some days) and issue is repeatig. We run VMware on same hardware/firmware with zero issues. It seems Justin’s (and some of us) frustration with Hyper-V is never going to end…:)
Problem details: http://www.hypervtx.in/archives/mvaneijk/2013/11/vnics-and-vms-loose-connectivity-at-random-on-windows-server-2012-r2/
OK so the first rule about Beta programs is … dont talk about the beta program…. but let me just go on record now as saying that whatever ground hyperV made with 2012 R2 … well they are about to get lapped again when vsphere .next comes out 🙂
The customer i was referencing in this article was using 2008 R2… in fact i think at the time of writing 2012 wasnt out ? (or maybe JUST out). Maybe read the date on the post ?
unfortunately the ground that hyperV 2012 has made will be decimated by vsphere 6 features…. all sorts of awesomeness in store.
I am well aware of the fact that HyperV isn’t as good as VSphere, yet, but still. It sounds more like a bashing HyperV article (and replies to the article) than a real fair and objective one. You are comparing a product that has been on the market for years already with a product that is just getting a little more mature. I don’t think it is a real fair comparison. It would be the same a saying: hey the Junior Consultant doesn’t perform as steady and good as the Senior Consultant. Imho HyperV deserves a little more credit than it does now in this article and that HyperV will get there eventually, but VMWare will always have a head start as they are around longer. Maybe HyperV will catchup, maybe it won’t. Just give it a fair chance.
Furthermore I think HyperV which comes Windows 2012 R2, has got some mayor improvements versus its predecessor(s) and is pretty nice alternative to VMWare/Vmware. (No I am not claiming it is better than VMware).
Thanks for the Comment EWB. My article was supposed to basically show my frustrations due to clients only buying on price. Namely because if you add up all the stuff that HyperV hasnt (and may still) not do as well or straight forward as vSphere… you end up coming to a wash on price after paying for all the extra consultants (be them vSphere or HyperV consultants). So in the end I think that if clients were to have your view, which I translate into “Well vSphere is ahead in the game, but hyperv shouldnt be totally ignored”… which i agree with. Then maybe they would understand that if it is business critical and you dont have the knowledge or support to run hyperV…. then choose VMware. If you want to run tier 2 apps on hyperv so you have a smaller vsphere cluster I’m cool with that, heck it will even hone your skillset so that if you decide to go 100% HyperV you will know how to do it inside and out.
Honestly I foresee my lab being a mixed environment at some point, but I dont think that will happen until maybe the next version of HyperV when things get a little stronger.
This article seems to me as very unobjective.
I am certified on both VMware and Hyper-V with System Center/VMM and have been using VMware for about 8 years and Hyper-V in combination with SCVMM for about 2 years.
I do agree that VMware and Vsphere is easier to get started than Hyper-V / VMM
But, (and now I must make a reservation for me not beeing really up to date with VMware since ver 4.1)
with SCVMM I would say you have a lot more possibilites, like Services and Service templates.
You also have builtin features for using virtualized networks, you’ve got baremetal deployment of hosts(Think you have this in Vmware also) and more….
Another great “feature” is that when you buy SCVMM you also get the rest of the System Center family (SCCM, SCDPM, SCSM, SCORCH, SCOM) which makes a great “toolbox” for managing all of your infrastructure. also all of this costs less than a VMware license.
I would say that once you learn how to use these Tools the right way you do have a cheaper and more flexible Environment than with VMware.
Thanks for the comment Michael. I think you would be impressed with the new stuff in vSphere 5.5, however you are correct service profiles and such are reserved for things like vCAC and vCloud Director.
Virtual networking exists in many different flavors so id have to understand more about SCVMM before commenting.
As far as price what does SCVMM cost for something like a 3 or 4 host cluster ?
I agree that vSphere is the superior product, but Server 2012 R2 is just finally able to make Hyper-V a product I can be convinced to switch to. But only from a cost standpoint. To virtualize Windows on vSphere, you still would need to buy a Datacenter license for the host. In our environment, we were spending close to $50k/year just for vSphere maintenance renewals. Now we don’t.
Speaking to VMFS, I’ve dealt with plenty of datastore/host issues where due to network or storage saturation/contention, the hosts and VMs stop responding and we can’t vMotion anything. VMware’s answer is “reboot the hosts” which means they have no clue. It is exclusively a VMFS issue as NFS does not have the problem. The newest versions of VMFS finally mostly solve this problem with a major reduction in LUN lock requirements.
So long story short, choose the product that fits your environment. For example, if you have lots of Linux VMs, stick with vSphere, as Hyper-V has terrible Linux support.
I know this is an older post but I just had to comment on how true this really is. I have personally run into this same issue with CSV’s going offline due to underlying issues with the CSV owner forgetting it was the owner.
We have numerous customers running both platforms and I don’t have nearly the issues or frustrations with VMware like we do with Hyper-V.
If it isn’t the fantastic SCVMM 2012R2 telling me my 2012R2 hosts are not responding there is the bug with NIC teams in 2012R2 that duplicates the physical MAC address on the vSwitches. Oh, don’t forget about using storage migration and bricking your VM because ODX is enabled and has corrupted your vhd. Want to upgrade that cluster? Better plan for some downtime, new cluster names/ip’s, etc.
The problems were even more comical with 2008R2!
You always have to keep in mind, that VMware has almost 15 years of experience in virtualization and more than ten years in server virtualization and they came a very long way.
Well. Bad guys would say, microsoft could take benefit out of the knowledge VMware had to learn the very hard way every once in a while but I think the major problem MS has, is that they try to build up the virtualization layer on top of a running windows with all it´s drivers and all it´s monthly critical hot fixes. You just can´t build up a hypervisor with an OS that is so rebootable like windows.
What makes VMware the (my opinion) way more reliable hypervisor?
An OS that fits on a 1GB SD.
An OS that, once running, does not need to write anything within the OS.
It needs less than a GB of memory.
If it fails, you can reinstall it in less than 10 minutes (depending on the BIOS Boot of the server and the SD class.
What is a pain in the ass with VMware?
The web client!!!
Why couldn´t they just use the vSphere client we all learned to love in all this years??? I don´t know. Maybe it gets better when running on HTML 5…
Hi Justin,
Thank you for sharing your expeprience. We had a very similar critical situation with a 24 hours downtime and a 4 hours (working hard with DELL Pro Support) recovery time with an Hyper-V 2 nodes cluster.
The CSV layer over NTFS relies of the accessibility of Active Directory services like Authentication and DNS in a cluster configuration. The CSV layer runs its service under an account that’s shared through an active directory to access NTFS files so that you need to maintain an active directory outside the cluster over hardware. If for any reasons your active directory remains unaccessible and, you reboot one of the hypervisor, the whole cluster of shared volumes is unaccessible having the hypervisors explorers freezing when accessing the CSV mount points (C:\ClusterStorage\volumeX).
We’ve encountered this critical situation when trying a controler failover (restart) of an equallogic array on production. A few VMs became unresponsive having us no other choice than rebooting the hypervisors. The AD services where hosted in a VM so when the hypervisors came back with no CSV volumes, this AD VM could not start and the whole cluster was down.
Fortunaly, we had another AD DC (hardware) in the same AD domain and with a few configuration steps, we were able to have the CSV layer back with all our VMs.
This is too much to provide a resilient virtualization infrastructure…
Frédéric.
We had VMware 4.1 along side of Hyper-v 2012 for 2 years. Same hardware, same san. The Hyper-V cluster was kept fully up to date, vmware wasn’t. 100 VM’s on each. Vmware absolutely crushed Hyper-V in both performance and ESPECIALLY stability.
Just do not use Hyper-V for anything other than a test environment. Before you think something is wrong with the config, we had 3 paid cases with microsoft, including a premier case. It simply does not scale, and we are in the awkward position of now having 200 VM’s on Hyper-V because we can’t pay for Vmware. You will get timeouts of the cluster storage, the rhs will stop process will crash hosts (event 1146). Hotfix applied over a year ago, still happens.
We had VMware 4.1 along side of Hyper-v 2012 for 2 years. Same hardware, same san. The Hyper-V cluster was kept fully up to date, vmware wasn’t. 100 VM’s on each. Vmware absolutely crushed Hyper-V in both performance and ESPECIALLY stability.
Just do not use Hyper-V for anything other than a test environment. Before you think something is wrong with the config, we had 3 paid cases with microsoft, including a premier case. It simply does not scale, and we are in the awkward position of now having 200 VM’s on Hyper-V because we can’t pay for Vmware. You will get timeouts of the cluster storage, the rhs will stop process will crash hosts (event 1146). Hotfix applied over a year ago, still happens.
Most definitely this is the work of VMware fanboys. We are a cloud solution provider with 5,000 virtual machines in VMware and 5,100 VMs in Hyper-V. We use Hyper-V and VMM 2012 R2. We use VMware vCenter with ESXi 5.5. They are both great products in their own right. Hyper-V has worked for us flawlessly with excellent performance. We have recently started to use KVM on CentOS. While less mature than Hyper-V and VMware, it is also a very good solution that takes much more initial configuration but it also runs very well.
I thought my boss was crazy when he said that we had to use Hyper-V for the set of 910’s we had sitting in the datacenter. They were tired of dealing with VMware licensing and thought it was way too expensive. Coming from the ESX world I did everything I could to dissuade them. Anyways I have grown very attached to Hyper-V and have found it to be very stable with few limitations. I can think of only one issue that I had with it and that was due to my misconfiguration. There are a couple of things about it that annoy me but every product can be improved. I am liking Hyper-V 2012 even more than 2008 R2. I wanted so badly to hate Hyper-V but as time passed and the VM’s remained stable I finally decided that its a great product and it was not VMware that I loved but it’s actually virtualization technology that I am so fond of. If I had to compare them I would have a hard time deciding at present because VMware is very expensive and not every business has that kind of capital. Is VMware really more stable? I haven’t seen any difference at all as we run both ESX and Hyper-V.
There will always be pros and cons but there are (still) some critic points with Hyper-V I just can´t get used to it.
Every first tuesday a month there will be a list of 10-20 hotfixes to install. Recommended or critical. That´s definitely easier with VMware
The CSV meta data communication path is a real pain in the … when you have two datacenters with two active storage systems and you see ISL (inter switch link) flooding all the time if you don´t plan your vm placement accordingly. No problem with VMware as well.
You have three front ends to work with: Hyper-V cluster management, SCVMM or (what was the name again? 🙂 ) and everything you try to do you will have to carefully think about to use which front end for it… that´s easier with VMware as well.
AND these front ends cost money as well, especially with SCVMM.
Nowadays there is no need to buy Enterprise (Plus) VMware liceneses any more for at least 80% of all installations because everything you might want to use is included in Standard and that´s not really expensive., especially when you use the acceleration kits.
Of course, when you need to install 40 or more hosts you really want to use Host Profiles but there is no comparable way to install Hyper-V that easy.
You need to have a physical DC to get access to the CSVs when everything else is down. With VMware you don´t need that at all and can stick to a very cheap Backup solution like Veeam for example – no way to protect a complete Hyper-V environment with such a software…
I think, MS needs one or two more OSes and Hyper-V will be a real competitor. Right now it´s just a “feature” that windows provides… nothing more, nothing less. 🙂
Just my opinion of course.
In addition to the Hyper-V host outages (event ID 1146) and entire storage volume timeouts (ERROR_TIMEOUT)
SCVMM R2 – REMOVED P2V capability (Physical to Virtual Conversion).
GREAT “Upgrade” microsoft…
SCVMM is very buggy software otherwise too, not enterprise VM management by any stretch. Be ready to have phantom “missing” machines until you reload the software.
I deploy both regularly (although more VMware) and I do tend to agree with a lot of this. The CSV is a nightmare – like you say if you have a virtual environment and power it all down and back up, the CSV won’t start (required AD authentication and a domain to function properly). So you need a physical DC in your new Hyper-V environment. You need a Windows domain to enable HA functionality.
With VMware, whilst I always set up SSO and join the hosts and vCenter to the domain for access control, it isn’t a prerequisite of enabling HA and DRS etc.
I’ve configured FC, iSCSI and SAS attached SANs in both VMware and Hyper-V environments, and VMware is far easier. You’re right that it’s fairly clunky in Hyper-V, i.e. having to use Failover Cluster Manager to create the CSV and enable HA on your VMs.
But I must say, I didn’t have a hard time at all getting my first stable Hyper-V environment up and running and P2Ving servers onto it. Granted it was a massive pain as the only way to P2V is with SCVMM or other tools, whereas VMware supplies the incredibly powerful vCenter Converter free of charge.
Since my title actually includes “VMware” I will ALWAYS advocate VMware solutions above Hyper-V, however it has come a long way since Server 2008 R2 and is being used in a number of “household-name organisations” that definitely do have the budget for a VMware solution.
Graeme, Thanks for the great reply!
What I need to add to Graemes Post and my current encounters with Hyper-V is:
The most pain in the arse is definitely the MPIO plugin in windows. If you have a fully supported storage connected, everything is fine, but if you have to set up the DSM and need to configure active and passive pathes it´s just desperation. The way to find the neccessary setup spots within the DSM is just ridiculous. And if you don´t set it up right and the system does a csv owner transfer you will know it by all the nice blue screens on all the hyper-v hosts connected to that csv. Nice try but microsoft really has to improve THIS crappy plugin very fast!!!
I realise that this post is now a bit long in the tooth but I though I’d give it my two bobs worth…
Whilst I do agree with some of these comments, Hyper-V on Server 2012 R2 is very solid. A lot of the issues that surrounded Clustering in general and Cluster Shared Volumes (going offline etc.) have been fixed through various monthly update rollups since then, so a lot of these are non issues. Secondly a lot of these sort of problems are down to poorly written drivers (certainly in our experience). Take the Emulex VMQ issues that they finally waved the white flag to (although to be fair to Emulex, this was a joint effort between them and MS). Server 2012 however was horrific and if there are still people with Hyper-V running on 2012 in production, I feel for you, really, really I do. So long as you are using good quality hardware, patched and configured to best practice, you will be fine on 2012 R2.
Overall, Hyper-V has come a long way since initial Server 2008 release with some great additional features coming up in the 2016 release. Yes it is still young (comparatively), however its popularity is growing (fast).
We offer managed Private Cloud services based solely on Hyper-V and System Centre 2012 R2 and have had no service interrupting issues to report whatsoever since this platform went live (non that were Hypervisor related anyway, and we adopted 2012 R2 very early on). Yes there are bugs and glitches (part and parcel of the career we have all chosen to work in), but from what we have found, these are quickly dealt with and rectified. Yes there is feature disparity between management tools, however who needs a GUI when most of the Datacentre management leg work is being automated for you? Very rarely do we need to actually dive into Hyper-V manager or Failover Cluster Manager. In actual fact, all of our compute runs in Core. Surely this is what PowerShell is for 😉
I must agree however that the feature disparity between SCVMM and Hyper-V 2012 R2 is a bit of a pain. There are certain management tasks that would be nice in SCVMM, however with the announcement of SCVMM 2016, a lot of this has now been cleaned up as Microsoft have stood up and finally taken note of this.
Both Hyper-V and VMWare are excellent platforms, but too often than not, I hear a lot of misinformation about Hyper-V floating about in the ether. Only a few months back we were in discussions with a client that “Wanted to do VMWare”. Some of the reasons behind their choice were quite uninformed. “Hyper-V doesn’t do VMotion”, “You can only place one VM on a CSV at a time”, “Hyper-V is a Type 2 Hosted Hypervisor and sits on top of Windows” etc. As soon as they were properly informed and more importantly, given the costings for both a VMWare vs. Hyper-V environment (both using identical hardware), they were quickly swayed by Hyper-V from a cost perspective. Who wouldn’t be in today’s current economic climate?
What we see an awful lot of is that upper management mainly couldn’t care less about one Hypervisor adopting a Monolithic Type 1 architecture and the other adopting a Type 1 Microkernalised architecture, one requiring AD Domain Services to run and the other not, one using a coordinator-less storage approach and the other not etc. etc., and why should they? What they are focussed on however is cost and what makes sense from a business perspective (even more so for smaller environments where cost is paramount). This is probably the main reason why we are seeing more and more clients move away from existing VMWare infrastructures to Hyper-V.
I am in neither the Hyper-V camp or VMWare camp. Both are solid platforms and when put together correctly, to best practices, both will provide reliable, good performance virtualisation. Without running the risk of insulting people, one area where things slip is when things are not done to best practice or are not configured correctly, so this is also an area that can introduce unknown variables into an environment, whether it be VMWare or Hyper-V. I think this stands for a lot of MS products that are too often branded as “unreliable” or “flaky” when more often than not, it is lack of knowledge and corner-cutting that causes a lot of issues. With all that said, from a financial perspective, I don’t think there is any getting away from the simply astronomical costs associated with licensing VMWare products (even more so when virtualising primarily Windows workloads).
Happy Virtualising (and everything else in between)!
Kindest Regards
Matt
I agree totally with Matt H.
If you skimp on hardware and the overall design you will be unhappy.
I found CSV with HP EVA’s and HP HBA’s… Have to look up what hba, but it was the one that the Eva 4400 had recommended.
We had few issues with 2012 or 2012r2 on 4 dl580 with 1tb of ram & 4 10 core xeons and the eva’s were 8 shelves of 600gig 15k FC drives.
The CSV’s were setup in the cluster and hyperv just did its thing.
I’ve been in VMware 5.1 and 5.5.0 for the last two years and its the networking that I preferred in hyperv.
The dist switch in VMware is just more work than it needs to be.
Plus it shocks me to see.how few shops setup the networking back end correctly.
Sorry people I feel you need a pair of Cisco 6509 with the best supervisor that supports virtualization. Then your a rock solid setup.
I am clearly capable of reading the date on the post, hence my comment as to how Hyper-V has come on a long way since its early incarnation.
Perhaps it would be beneficial to other readers if you thought about updating the post with some up-to-date, technical relevance? This would enable them to reflect on those bygone days and actually make a better judgement.
As already stated, for our clients, cost is the driving factor behind their shift to Hyper-V, and that really is the very bottom line. Like I said, when it comes to cost, most decision makers really couldn’t care less in the real world about “technical awesomeness”, that one Hypervisor is better because it can scale to 128 vCPUs and one can only scale to 64 as an example (and let’s face it, in the REAL WORLD, do these figures even mean anything??).
Here’s my take away. Hyper-V is certainly a platform that makes much more economic sense, and from where me and the company I work for are sitting, it certainly seems as though a lot customers are beginning to sit up and realise this ( Now I’m just waiting for the “your customers are obviously run by very reckless individuals”, which for the record, we have also heard on more than one occasion, based purely on their choice of virtualisation platform. Very strange).
Again, kind regards
Matt
I had to step in and comment after reading those last few comments from Matt H and Sean from Chicago. Matt H you do seem quite biased towards Hyper-V although claiming you like both – but your setup is all Hyper-V?
I’ll start by saying I’ve designed and deployed both VMware and Hyper-V solutions – Hyper-V from 2008 R2 and VMware from ESX 4 all the way to ESXi 6.
Yes, Hyper-V is solid. It’s now a type 1 hypervisor that sits directly on the hardware, even with the GUI-based installation of Server 2012 R2. The OS actually becomes a VM running on the hypervisor, you can turn off the GUI with a PowerShell command if you want.
However, networking in Hyper-V is absolutely rubbish compared to ESXi. You need separate teams of NICs for each vSwitch you want to create, and can’t have multiple networks on one set of NICs. How do I know? Because literally this very week I put in a new set of DL380s with 10GB/s NICs. My plan was to use these NICs in a team to handle vMotion (Microsoft calls it Live Migration, so I’ll call it LiveMi) as well as the iSCSI traffic with the use of VLANs. I’ve done this a million times in ESXi, splitting out the physical connections over a couple of dedicated stacked switches. Well, this didn’t work. I had to then use 1GB/s connections for the LiveMi.
As for vDS (Sean) – how are they difficult to configure? They are as easy as a standard switches just over multiple hosts at the same time. I would suggest maybe you don’t understand them well enough. I will admit that migrating standard vSwitches to vDSwitches can be tricky, but configuring from scratch is simple. VMkernel port groups (for management traffic etc.) is still more manual, because you need to give each host unique IP address settings.
With regards to configuration, I’m a by-the-book kind of guy, I have spent hours of my own time testing and noting what works better and working out percentages of gains. From the OS configuration to the network configuration, using link aggregation and different kinds of load balancing settings. I’m very fortunate to have two dedicated labs at work for this kind of thing.
Licensing costs cheaper with Hyper-V? Yes, if they don’t use SCCM / SCVMM. If they do, it can actually be MORE expensive than VMware. Luckily VMware comes in a few different flavours, depending on requirement often their most expensive variant isn’t required. I’ve also seen some clients paying for Enterprise Plus licencing but not using half the features.
I’m really not fussed with regards to what the world uses moving forward, because I know both technologies and work with both. If you’re a school or charity, yes absolutely Hyper-V is cheaper WITH SCCM / VMM. If you’re a corporate organisation and want a simple Hyper-V cluster without VMM, then yes it’s cheaper as you get Hyper-V with your Windows Server licences. Microsoft have done a great job of vastly improving their product, but VMware are the innovators here. Microsoft is following suite.
If you’re paying for SCCM / VMM, go with VMware. Seriously. And for all the folks migrating to Hyper-V now from VMware for no good reason, usually they find themselves regretting it down the line when they realise they needed to spend money on SCCM / SCVMM to get the same sort of functionality they expect. Managing servers from Failover Cluster Manager / Hyper-V MMC snap-ins doesn’t cut it for me, or any of the IT Managers I’ve worked with.
I find your statement; “can’t have multiple networks on one set of NICs” quite difficult to comprehend. You absolutely can do this, i.e. converged networking (I think that is what you are referring to here).
Without SCVMM, you create your NIC team and then layer a VM Switch on top of that, then simply use PowerShell to create as many vNICs as you want, exposing them to the Parent Partition (Host Partition, Management OS, whatever you want to call it). You can than VLAN tag each vNIC and even apply weighted QoS polices to each if required to isolate each data type as best as possible.
With SCVMM, it is even easier, repeatable and less error prone (unless you are scripting the vNIC creation of course).
And yes, you absolutely can use the converged fabric approach with your storage networks if you are using iSCSI, however I have always been an advocate of dedicating the storage to that very purpose and not sharing it with any other workloads. Just a personal preference.
We do the very thing you have just described as being impossible. A single VM Switch with One vNIC for Cluster Heartbeat, one for Live Migration and one for host management, all exposed to the Management OS and with QoS applied. Works well.
If you need any advice, I’m happy to give it to you, at no cost I hasten to add. (I am not in the business of telling people that they don’t know what they are doing etc. as is often the case on many a product bashing thread like this. That behaviour is just plain unprofessional).
Kind Regards
Matt
I agreed with you that it’s a great platform, I disagreed with Sean that vDS are unnecessarily complex to configure. I am perhaps less tactful than you are, but I see this as a casual discussion and not taking anything to heart – so apologies if my post offended you. I’m just representing myself here, not VMware or Microsoft or my employer. I just happen to agree with Justin on his views on Hyper-V (his view as per his blog above is that it’s unnecessarily difficult to configure compared to VMware).
I didn’t say that the solution you proposed it was impossible. I just said that the way it’s done within Hyper-V is a lot more convoluted than it needs to be!
So what if the client hasn’t gone for SCVMM to manage their virtual machines and networking? What if you’re putting a solution in that your helpdesk will have to support further down the line, will it be easy for them to troubleshoot or will they require PowerShell for that too? Would you say that it’s simple for your typical IT manager to pick things up where you’ve left them off with regards to the converged networking – would they be able to understand it without calling you?
I prefer to have my vSwitches pass untagged traffic to physical switches as it’s easier to understand. If there’s a limited number of NICs, this isn’t always possible. In this case I would go with the solution I described earlier. The servers from my example had one dual port 10GB SFP HBA, along with 4 onboard NICs. With regards to Hyper-V and iSCSI, I’d usually leave my iSCSI connections separate (depending on what type of SAN you’re using) and team my management and VM traffic NICs. Within ESXi I’d create separate VMkernel port groups and team them within the software iSCSI adapter, with one active adapter per port group.
I just have to say that a complex VMware solution can be implemented that’s easy to understand, it’s graphical (if using the desktop or web client) and makes sense if done logically. They say the difference between good code and bad code is that humans can understand it – anybody can write code that a computer can understand. Isn’t that the same when we install a solution – it will be supported by people after all?
If you look back to my first comment, you’ll see that I’m not against Hyper-V at all. I wrote a blog last year after my first Hyper-V 2012 R2 installation – I said that the product has come a long way since 2008 / 2008 R2. But also since I’m specifically employed in my current role as a VMware Consultant (however I also do Hyper-V projects) I will always advocate Hyper-V solutions when I provide consultation to clients.
Since Hyper-V is included with Server 2012 R2 and is free (with virtualisation rights based on your Server edition) it does make sense to use Hyper-V when the budget is tight. You’re right that most finance directors don’t care about what their infrastructure is running on, nor what type of hypervisor it is. I agree that the majority of workloads that are virtualised are Windows-based.
But there’s also other elements to take into consideration. I once had to migrate a Hyper-V cluster to a new domain. It was difficult given the requirements of Hyper-V – i.e. a Windows domain to actually BE in a cluster.
I’m actually doing my first VCAP exam in two weeks, so I’ve invested a lot of time and money into myself with regards to being proficient at VMware technologies. It doesn’t mean that I’m completely anti other virtualisation technologies – just like having a Cisco qualification doesn’t mean I don’t like working with other switches and routers – in fact HP stuff is really good.
I’m not trying to product bash in the slightest, however I have already said that I’m a VMware advocate. Should the day come when VMware is pushed out of the market by Microsoft (which is highly unlikely) I won’t panic and get upset, I will just make sure I focus on investing myself in the Microsoft Private Cloud path. I think that if people start moving to Hyper-V in numbers because it’s cheaper, VMware will simply lower their prices. We’re a top tier VMware partner and offer our clients huge discounts anyway.
I hope I haven’t offended you.
Kind regards,
Graeme
Sorry – I meant I will always advocate VMware solutions to clients – excuse the typo.
Absolutely no offence taken whatsoever (I wasn’t aware that you had offended me!).
I guess that your situation and mine Graeme are very, very similar in nature, just polar opposites to each other in the sense that in the company I work for, I am primarily a consultant specialising in virtualisation based on Hyper-V and System Centre mainly. That said, like yourself, I am also not just siloed into one product. VMWare certification is something I have also given some thought to, however I do need to somehow find enough hours in the day to actually achieve this! It just so happens that at present, it is the Microsoft solution that I am focussing on for the time being.
vSphere out of the box is great and the ability to get a configuration up and running in no time is brilliant. I do absolutely agree with the original thread topic that the toolset disparity is annoying with the out of box basics, however getting even a simple setup working with these isn’t really that hard in the grand scheme of things. What MS should be looking at doing in future releases is to create some form of basic, cut down version of VMM that will at least allow you to get all of your networking, storage and compute configured in one all encompassing tool. For the moment though, this tool is SCVMM. I guess really this is a similar situation to having to purchase vCentre as an additional centralised management option (obviously with differing licensing models).
I guess my summary statement is that I was really only trying to iron out a few misconceptions about Hyper-V that’s all, and this really was not aimed specifically at you. We are all professionals and the very last thing I would want to do is insult another fellow, skilled individual. In the past, a few issues and incorrect pieces of information have come to light from time to time and this really was down to customer misinformation. The old “Hyper-V does not allow for Live Migration between hosts” is still a classic that even now, in the year 2015, we still hear on the very odd occasion. And yes, virtualising the virtualisation fabrics Domain Controllers and placing those on the same cluster is a really bad idea that we have also seen many a time! This one has even been known to keep me up at night!
When managing Hyper-V with SCVMM, I would say the experience is fairly simple to get your head around (although doing it every day, I am bound to say that). We have however given fairly comprehensive training sessions to our clients with no real concerns raised. Yes there was a little questioning around the Logical Network > Logical Switch > Port Profile > Port Classification relationship, however once demoed and diagrammed out, it soon becomes quite clear. You can generate Visio style outputs in VMM that give you an idea as to the logical networking topology for hosts, clusters and individual services / VMs. This is really for the benefit of others rather than yourself Graeme as I am sure you are already 100% aware of this.
I think really with or without VMM, there is still a requirement for some basic PowerShell knowledge though. If you are going to do the entire host cluster provisioning from VMM, then this is actually pretty straightforward and really powerful (This includes all of the storage management (if your storage supports SMP or SMI-S), networking, host provisioning, library resourcing among many, many other things).
Picking up on a comment I spotted a while ago, I would say performance wise both Hypervisors are much the same from a Guest perspective to be honest (I can only assume this was what was being referred to), and really this is going to be determined by the hardware that is in use.
Keep it light hearted and keep blogging!
Kind Regards
Matt
You’re definitely right in that we’re probably in the same sort of role at polar opposites. I’m sure when you’re working with ESXi you get to a situation where you say “man, I wish Microsoft would do that” and the same goes for me when I’m working with Hyper-V. I am a HUGE fan of Server 2012 R2. It was essentially built to be virtual, unlike Server 2003 / 2008.
I’m not very experienced with Server Core but I imagine it’s similar to ESXi and adding a new host within vCenter when you’re running an SCVMM environment. Most of the time the customers I work with want full GUI based installations running so that their IT staff can log onto the server and troubleshoot if there’s an issue. If using a Hyper-V solution without SCVMM management, it’s necessary to become familiar with all types of MMC snap-ins but mostly FCM.
PowerShell is what drives Server, the GUI is just an overlay. Microsoft seems to be moving towards this more and more, with a lot of the GUI tasks taken out of their products (Exchange 2010 to 2013 for example, moving a database). Definitely agree some PowerShell knowledge is a good idea.
You’re right with regards to making DCs virtual, but with ESXi this isn’t as important as Hyper-V. The virtual estate will come online regardless of the state of the DCs. I’m sure you’re familiar with the different methods of getting communication happening between your servers and UPS. Time synchronisation can be an issue with older servers, especially since the server won’t be able to reach the NTP pool if the DNS is down unless you put a secondary DNS address which is public.
This becomes a huge problem when running multiple sites and having a badly designed domain hierarchy – for example a client I worked with recently had a tombstoned DC in Singapore, the speed and latency between their primary site in London was just too high for two-way replication (and DFS) to work properly. When it came time to migrate them to Office 365, the issue was discovered and they were in a world of pain!
I always create affinity rules to keep DCs running on separate hosts and separate datastores for a truly redundant setup.
Hyper-V does need to follow VMware’s example and create something like vCenter, the licencing will definitely need to change for this to happen. Most of the customers we have would never pay for SCVMM. If you take vSphere Essentials Plus as an example (which is basically vSphere Standard packaged for 3 hosts) it includes vCenter licencing too. It’s a one-off payment for that particular version (i.e. 5 or 6) which gives you rights to update from 5.0 to 5.1 and 5.5 etc. Microsoft is making Windows 10 a subscription rather than a one-off payment, and come 2016 I expect their Server licencing will follow a similar model.
Licencing changes so much that I always tell customers to check directly with their account manager or Microsoft themselves – and even then I’ve seen some cases where Microsoft themselves weren’t sure whether the customer would be compliant.
You already sound quite proficient at ESXi, so I’m sure for you to take the exams would be fairly easy. It’s the VCAP certifications that require dedication – I’ve been non-stop for two weeks now.
I used to be an IT trainer so I know all too well how customers can be. If I can’t explain something in a simple way, I know I don’t understand it well enough so I need to hit the books again. Diagrams are a great way of showing how it works, both from a logical and physical perspective.
Glad to have had this chat with you Matt – keep up the good work with Hyper-V and happy virtualising!
Cheers,
Graeme
Sirs,
We have two hyper-v clustered sharing a CSV on starwind. However, we managed to setup veeam for testing.
Tried to backup an online and offline virtual machine with no luck…
Error receiving:
unable to create snapshot (Microsoft Software Shadow copy provider 1.0; Microsoft CSV shadow copy provider) (mode: crash consistent) . details:
unknown status of async operation.
the shadow copy provider timed out while flushing data to the volume being shadow copied. this is probably due to excessive activity on the volume.
try again later when the volume is not being used so heavily. failed to create vss snapshot / failed to perform pre-backup tasks
Any hint or advice?
I would reach our to Veeam support. They are pretty awesome and should have you running in no time. Even if you are in demo mode right now…