RouterOS works ok using Hyper-V R2, but using Legacy Network Adapter.

Using Legacy Network Adapter you have less performance than using a synthetic adapter, and most important, you are limited to only 4 adapters/virtual machine.
I've used Linux IC on an CentOS installation and everything seems to be working ok (installation went smoothly, perfroamce is good).

Is there any chance that Linux Integration Components be adapted for routeros use, maybe Mikrotik guys could create a package "Hyper-V IC" based on the official IC's?


i know it's a long shot.

thanks.

if hyper-v "syntetic" interface was working in previous versions, just create supout.rif file of the RouterOS where this interface is not supported. And send it to us.

hello janisk,

i dont think it ever worked. it cannot work without integration components.

'synthetic' interface means that the integration components are installed in the guest operating system (ther eare synthetic drivers for the disk io subystem, network interface, hartbeat detection, time sync, etc).

Microsoft developed the Integration Components for linux: http://www.microsoft.com/downloads/en/d ... b5b9b64551

I've tested thoose Integration Components on an CentOS installation, and it seems to work just fine.

Could you adapt them, so that we could use it when virtualizing RouterOS using Hyper-V.

I suspect the situation is the same for other virtualization solutions. without some sort of driver/integration component the guest operating system would use only emulated hardware.

and to make matters worse, microsoft emulated a DEC21140 network interface.

"hardware" components used to work, like, ethernet interface. Last report is that legacy works but specific Hyper-V interface does not, but used to work. If you can make supout.rif file with that specific interface and send it to us, please do so.

About tools that should be installed on guest os - they will never be added to RouterOS, as there are a lot of implementations that require them, they all work work well without them. Yes there are some limitations that can be overcome with some network monitoring tools like the Dude, that is free download and if required can be run on Linux using wine, without requiring expensive server OS licenses.

edit: in short - if interface is not working but has decent linux driver, we can look at it and add it. Guest tools will never make it into RouterOS.

I added an synthetic network adapter, and it doesn't even show in interfaces tab.

it takes a while now to get past "starting services" message.

would support.rif or debug logging help?

supout.rif file would help

I just sent it to support linking it to the forum thread.

Hope it helps, would be really nice to use native network adapter in Hyper-V.

Just so sum up as I understand it:

1) Xen and Hyper-V can both use fully emulated hardware (meaning the guest OS is not aware it's inside a virtual machine), only if you have support for hardware virtualization (AMD-V or Intel VT). Virtualbox and VMware do the same thing, but do not require hardware virtualization support. Mikrotik will support this mode.

2) Xen and Hyper-V can also run in paravirtulized mode (Integration Components). This means the guest OS (aka RouterOS) must include tools that run inside the guest so that it is aware that it's in a VM. Mikrotik will NOT support this.

Also, please do not ask Mikrotik to support option 2, they just won't do it at this time. It would take much more development time and resources. Mikrotik should focus on make a solid system that IS the OS, not software that runs inside VM's. If you want other solutions, they do exist, however that is not where Mikrotik shines.

I hope this is clear, and helps others coming in from search engines also.

I would agree but microsoft allready developed Linux Integration Components, that work perfectly on many linux distros.

I suspect Mikrotik would not have to write them again from scratch, jut include them as a package somehow.

It would help a lot on VPN solutions. Routerboard hardware (except rb1000 which is not in production anymore --- and it's expensive) cannot handle high badhwidth VPN (ipsec/pptp) so i (and others) have to use x86 hardware to create VPN servers/clients. Since 1 modern server is already pretty powerful, i, and maybe more people will use it as a platform for virtualization, and run many virtual computers on it.

i currently have 1 dell 2u server with 16 virtual cpu's, 64gb ram and 14 hdd's and i allready have 15 vm's on it. It's a waste of money for me to buy a new server and not use the current one. Even if i buy new hardware routeros would not use it properly, yet.

I'm not trying to say Mikrotik guys should prioritze this, i'm just asking if the could use what Microsoft already developed.

Thanks.

problem with all these implementations is that vendor forgets to support them. With current kernel it works, it is not clear if future kernel releases will work with "Vendor X" sources. It was bad experience with XEN and their integration with Linux kernel. That is reason why Mikrotik is avoiding adding "Vendor X" code to RouterOS, because then we have to support it even if "Vendor X" drops support for it.

i understand that. i also work in software development and i am faced with similar problems.

can you add it as a package named "Hyper-V R2 Intergration Components", with forum support and limited official support?

Sorry that's against our company policy. We don't make packages that might not work, or which we can't properly support.

this is sad news for me.

this is the second time i have to give up on using mikrotik and switch to something else.

i hope there won't be a third.

As I said before, Mikrotik is not targeting the VM world heavily. They make RouterBoards with RouterOS as a single operating system.

However, Mikrotik is not so proprietary like other vendors, so you can still integrate using protocols like SNMP, NetFlow, etc... etc...
Mikrotik's Dude can even monitor other devices' status through standard network protocols.

I support Mikrotik decision on this one.

i won't comment any futher on this.

i am not in a good mood now. i just realized i bought a few things i have to ..... throw away.

I understand you. But I recommend you to ask in the forum about such things, before you make any purchases.

problem with all these implementations is that vendor forgets to support them. With current kernel it works, it is not clear if future kernel releases will work with "Vendor X" sources. It was bad experience with XEN and their integration with Linux kernel. That is reason why Mikrotik is avoiding adding "Vendor X" code to RouterOS, because then we have to support it even if "Vendor X" drops support for it.

If I could get Microsoft to commit to supporting their released code for Linux, would Mikrotik come to the party? It's worth a try, I mean simply asking them that is...

This seems like an incredible shame otherwise. We miss out on things like running RouterOS in their HA clustering and not only performance on the v-ethernet. With VLANs there are amazing things that RouterOS does for me, the sky is the limit and clustering would complete this. This would also completely nix the whole issue about not supporting RAID btw. Not to mention, you already support SMB as of late.

You just need to take one more small step!

Counting on you,

R

@Mikrotik: what would it take for these features to be integrated?

From what I can see, their Linux integration support has been good. They have just recently released v3.2.

Is this issue about losing support the only obstacle, or are there others?

Wow!
Did I just read in a previous post "Don't ask MikroTik to support ....."
This issue appears to me to be pretty straight forward..
The argument of "future support" and "vendor abandonment" is ridiculous.. We could say the same thing about MikroTik and thier products. Who is more likely to be around and support thier products in the future, MikroTik or MicroSoft?
The bottom line is this, there are *Many* of us system integrators who are committed to using Hyper-V. We want an integrated router system without the appliance for various reasons.
MikroTik *Should* be ecstatic that we are considering thier RouterOS over the likes of Cisco and others. But we have to be able to run it in a Hyper-V VM, PERIOD!
So, If I am understanding this correctly, MikroTik does not think there is a Hyper-V market to make it worth the (I am betting rather small) investment in time to support it.
Maybe they are right and there is not enough of us Hyper-V/MikroTik fans out there.
With that type of thinking, this is a self-fullfilling prophecy.

it is not about how many of you are using it. At the moment it is broken and support from original vendor is weak. When that changes, most probably that code will become part of official Linux kernel and their supported code base. But now i do not see how that could happen.

it is not about how many of you are using it. At the moment it is broken and support from original vendor is weak. When that changes, most probably that code will become part of official Linux kernel and their supported code base. But now i do not see how that could happen.

Boom baby: http://www.networkworld.com/news/2012/0 ... 57916.html

Is this enough to get MT to support it?

You have my vote, if that counts

Boom baby: http://www.networkworld.com/news/2012/0 ... 57916.html

Is this enough to get MT to support it?

You have my vote, if that counts

you should be more careful when reading.
What i red in the article - after months of users moaning about problems Microsoft suddenly woke up and committed huge amount of code.

Boom baby: http://www.networkworld.com/news/2012/0 ... 57916.html

Is this enough to get MT to support it?

You have my vote, if that counts

you should be more careful when reading.
What i red in the article - after months of users moaning about problems Microsoft suddenly woke up and committed huge amount of code.

Wow@janisk, you mustn't be a Negatron, you should be an Optimist Prime

From where I'm sitting: What we know about MS and their Hyper-V code is that they test it, quite a bit, before releasing. This accounts for the longer release cycle I'm sure. Besides, once the code is polished, then there won't be much left to do and it is also open source code after all.

@MT dev team: Did the Xen guys really ruin this for the rest of us?

This seems like such a shame: If the code is in the up-stream Kernel, then u're disabling the modules in MT's release of the kernel and we're simply being denied access to something that's already there? Strange that debian, redhat, suse, ubuntu, mint, slack, gentoo and others all allow the use of these features in their kernels. Is MT really that worried that they'll go as far as to set themselves apart like this?

In the end, all I really want is the NIC drivers. VirtIO drivers for the other stuff is less important, although it kills me to say this.

So after all that, here is my solution....

You need code maturity and quality, this we understand. These are two things that are not an issue from where I'm standing. Test the code in beta, only release when ready. Code maturity is a moot point, it's 2012 now and this was first released in 2009. If the code doesn't pass rigorous testing (just like other vendor stuff has to) then don't release it. If it does pass, then make your clients happy and bring MT into the new world of cloud based routing and firewalls.

I implore you to reconsider your stance, in light of the above comments and reasoning. I'm sure others agree with me too.

Warm regards,

R

don't bash the messenger

anyway, XEN was quite disastrous experience when in the end only good solution to the problem was to go KVM way even with HW support required for it to work. That is fully supported and there have not been any issues that has not been resolved and it is working beautifully.

I want to wake that up because I am now in that position trying to get a x86 RouterOS installation working. I have a fabric built on Hyper-V as routers, with which I am very happy with 2 small nudges - first the lack of a PPTP server, which I rally could need (connecting an ovffice to a data center using 3 internet providers, and PPTP would handle packet distribution fully automatic compared to the current setup) and the queues when bandwidth fluctuates (they get seless when athe provider overloads, they should handle traffic high priority first, only dropping low priority traffic).

Anyhow - Hyper-V. Got delivery of a new server yesterday that has to be integrated into the fabric. It is located somewhere else. SOmewhere VERY else, no´physical access, extreme hosting fees (400 USD per rack unit). I run Hyper-V on it to get a handfull of machines working in that bridge point. I need to tiee it into the fabric for remote desktop and data transfers.

I would love to install a nice, powerfull, not resource killing RouterOS licenses there and to tie that into my fabric, to keep things consistent.

Sadly I can not because RouterOS does not consider one of the two large distributed and enterprise level virtualization platforms (yes, Hyper-V and VmWare, sorry, no KVM) worthy integrating, albeit MS provides the full components for quite some time now (just released v3.3).

That will mean bad integration, bad control, no controleld shutdown etc. The alternative would be hipping a preconfigured 450G there and paying 400 USD a month for the used space.

Wow!
Did I just read in a previous post "Don't ask MikroTik to support ....."
This issue appears to me to be pretty straight forward..
The argument of "future support" and "vendor abandonment" is ridiculous.. We could say the same thing about MikroTik and thier products. Who is more likely to be around and support thier products in the future, MikroTik or MicroSoft?
The bottom line is this, there are *Many* of us system integrators who are committed to using Hyper-V. We want an integrated router system without the appliance for various reasons.
MikroTik *Should* be ecstatic that we are considering thier RouterOS over the likes of Cisco and others. But we have to be able to run it in a Hyper-V VM, PERIOD!
So, If I am understanding this correctly, MikroTik does not think there is a Hyper-V market to make it worth the (I am betting rather small) investment in time to support it.
Maybe they are right and there is not enough of us Hyper-V/MikroTik fans out there.
With that type of thinking, this is a self-fullfilling prophecy.

when Microsoft Hyper-X driver code will be made as a part of kernel, it will be added to the RouterOS the same way as Intel drivers are.

The answer to my feature request was as i have stated before - either code is added to the kernel and hence to the RouterOS or it is not. It is not that just code from some obscure company is left out.

More about the state of the code itself you are interested in - there is a similarity with reiser4 code - kernel team had certain requirements for the code - when these where met code was added to the kernel, there are no exceptions, code has to be of good quality, maintained in timely manner so it continues to work if changes to the core is made.

Since not all requirements are met, code is in staging where developers can get used to code support and to meet coding standards that are required.

On the other hand - don't kill the messenger.

when Microsoft Hyper-X driver code will be made as a part of kernel, it will be added to the RouterOS the same way as Intel drivers are.

The answer to my feature request was as i have stated before - either code is added to the kernel and hence to the RouterOS or it is not. It is not that just code from some obscure company is left out.

More about the state of the code itself you are interested in - there is a similarity with reiser4 code - kernel team had certain requirements for the code - when these where met code was added to the kernel, there are no exceptions, code has to be of good quality, maintained in timely manner so it continues to work if changes to the core is made.

Since not all requirements are met, code is in staging where developers can get used to code support and to meet coding standards that are required.

On the other hand - don't kill the messenger.

Well, this obscure company is no the 17th biggest kernel patch submitter.

When do you plan upgrading the linux kernel to 3.4?. Because according to yor own information just given then you would fully support Hyper-V...

http://mythoughtsonit.com/?p=1155
http://www.h-online.com/open/news/item/ ... 98601.html

That is supposedly the last drive entering the kernel. Which means the others should be there already, notable network.

Now, is that somehow a bad interpretation of your actually very straight statement? That the drivers WILL be there when they are in the kernel (i.e. you do not remove them), and as 3.4 has all the drivers, Hyper-V would be a fully supported system, from a driver point of view (not yours, I understand - I meanfully supported as all drivers that are part of the kernel are there)?

That said, given per 6.0 beta you currently run ancient kernels (linux-2.6.38.2) saying "when it is in the kernel" Is kind of a child's game. The Kernel you are now putting into the 6.0 beta is released more than a year ago.... so obviously it is not "when Hyper-V drivers are in the kernel" but "when Hyper-V drivers are in the kernel and we somehow magically decide that the kernel with a Hyper-V version is now outdated enough to go into a new build of RouterOs

That also explains the issue some people have on another thread to get RouterOS running on a modern X86 platform that was not around a year ago (x79) Any plans to upgrade during the beta to a current kernel? You would make some people very happy....

In RouterOS v6 we will have this:

*) upgraded drivers and kernel (to linux-3.3.5);

In RouterOS v6 we will have this:

*) upgraded drivers and kernel (to linux-3.3.5);

Which beta?

Just checking the download and it reads:

updated drivers and kernel (to linux-2.6.38.2);

which is not 3.3.5 Is that beta 3? When can we expect that?

I did not say beta.

I did not say beta.

Oh, come, stop playing child games

http://forum.mikrotik.com/viewtopic.php?f=19&t=62109

has the information - Beta 3 is the new kernel, and looks like it is quite close to release Time for a little waiting.

Kernel version is not set in stone, but it is likely that next beta will have new kernel. I cannot promise you what kernel version will be in v6 final. It might be v3.3.5, it might be something else. v6beta3 is expected this or next week. I highly doubt there will be Hyper-V support in any upcoming version of RouterOS.

Kernel version is not set in stone, but it is likely that next beta will have new kernel. I cannot promise you what kernel version will be in v6 final. It might be v3.3.5, it might be something else. v6beta3 is expected this or next week. I highly doubt there will be Hyper-V support in any upcoming version of RouterOS.

Why?

I mean, seriously.

The statement was "if it is in the kernel, it stays in". Drivers in the kernel = support for Hyper-V, because regardless how you turn it this is quite trivial - some drivers, that is it.

You just say you would remove the drivers from the kernel? You are aware that there is a hugh market waiting for someone to actually provide a working version with even minimal support?

Unless you downgrade the kernel a LOT (way before 3.4) Hyper-V should work. Why? Because 3.4 was adding the LAST driver (Storage) which - funny enough - is the one we wont really care about (enlighted ISCSI - given how IO heavy Mikrotik is, everyone is suppposedly fine running it on a simulated IDE controller). The relevant ones were added way before (network). I can understand technical reasons for a downgrade, but taking this out - you mean you would take out kernel drivers to make sure Hyper-V is not supported although the kernel happily does so in stock configuration?

Ok, got my hands on a 6.0 b3 current build.

Does not even install. The iso is stuck on booting - seems compatibility with IDE drives (as this is what Hyper-V no simulates) got downgraded to not important

I'd say fine, don't support it. Just don't turn off the drivers in the kernel.

I'd say fine, don't support it. Just don't turn off the drivers in the kernel.

Amen. And maybe make sure that not only YOUR part, but also the little loader on the ISO file is not totally outdated? Let the Linux people handle it

I've investigated Proxmox 2 for hosting my Linux virtualizations, it is most impressive. Proxmox 2 touts features like HA, iSCSI, AD auth, automated backups, VLANs and support for VMWare disks. (There are more)

RouterOS seems to like the VirtIO based NICs and disks. Bandwidth is much better than on the crappy 100mbit legacy NICs on Hyper-V.

At long last I have an answer for the no-RAID-support issue via use of iSCSI. Until now, my core router RoS x86 installation has always had to ride on a single drive! I'm going to enjoy snapshots and easy restore points if this progresses as it is now.

R

Link: http://www.proxmox.com/products/proxmox-ve

virtio is part of KVM, and virtualization on x86 is done through KVM in RouterOS. In our test KVM with virtio was faster than VMware.

virtio is part of KVM, and virtualization on x86 is done through KVM in RouterOS. In our test KVM with virtio was faster than VMware.

True, it simply works right out of the box like a normal interface and/or disk.

R

I've made up my mind, I'm moving to Proxmox because RouterOS simply rocks on KVM with virtio devices for disk and ethernet.

R

If this happens,

In RouterOS v6 we will have this:

*) upgraded drivers and kernel (to linux-3.3.5);

and this > http://www.phoronix.com/scan.php?page=n ... px=MTAzOTk is also true, then the future looks good for Hyper-V support.

I'm still moving to KVM, but I'm going to do so with virtual disks that I can easily move back to Hyper-V, just in case.

R

MT should be concerned that RouterOS is working well with Hyper-V cause then you can create High Availability Solutions.
In a project we're switching over from physical servers to Virtual Servers with SAN storage solution and Fail Over Clustering for High Availibility.
One of our goal was to virtualize Userman (Radius) to be used organization wide as Radius server for our Visitors WiFi's.
But as RouterOS can't be used due to the not recognized ethernet adapers we can't use it and we have to switch over to another system.

In our case it's only for Userman but for others they're willing to use RouterOs for other purposes too.

So this is a pitty.
Regards

MT should be concerned that RouterOS is working well with Hyper-V cause then you can create High Availability Solutions.
In a project we're switching over from physical servers to Virtual Servers with SAN storage solution and Fail Over Clustering for High Availibility.
One of our goal was to virtualize Userman (Radius) to be used organization wide as Radius server for our Visitors WiFi's.
But as RouterOS can't be used due to the not recognized ethernet adapers we can't use it and we have to switch over to another system.

In our case it's only for Userman but for others they're willing to use RouterOs for other purposes too.

So this is a pitty.
Regards

It is also for people using clouds that may run Hyper-V (Azure - it is not like MS is not pushing that heavily, and rightly so with the public/private approach).

I find this whole argument quite chidish from the Mikrotik side, especially because at least I would have no problemepaying around 250 euro for a Hyper-V version and all the drivers are in the current Linux Kernel anyway. THe complexity of trying to introduce a Mikrotik device into a data center is posibly significant, especially if you can not go there and do it manually as the data center is on the other side of the planet. Also a physical Mikrotik is a weak sport due to hardware failure issues, so it means having everything twice - remote. Given that a data center I plan to move on charges remote hands at 500 USD per hour... having a damn network cable plugged in costs more than a router. I rather stick to virtual here, but like so many I wont change my virtualization platform because the router vendor does not support it. I wont change my car because I can not get my favourite radio manufacturer to allow me to install his radio either - that would both be kind of wag the dog, no sorry.

It starts smelling a lot like a religious war to stick with "open source" instead of providing customers a working solution. Arguments done are in general on a very low level ("vendor support" etc.) when the drivers are in the current kernel. Maybe Mikrotik is afraid of polluting their sacred network by installing a Hyper-V machine (0 license cost) for testing? By now I do not know, but the arguments are in general weak. Mostl because it all is there in the kernel anway. Must take effort to remove it.

Correct,

I hope MT will do something about it.
Anyway if the network interface is not working we can't sent a support file to MT to take a look into it and fix it.

Correct,

I hope MT will do something about it.
Anyway if the network interface is not working we can't sent a support file to MT to take a look into it and fix it.

Well, if they use the later Linux Kernel in the new RouterOS v6 beta 3 as suggested above, then all they have to do is leave the Hyper-V modules enabled. It should simply work then, like on KVM as virtio does.

ok,
Lets hope.

Related, yet not... but worth a chuckle: http://linux.slashdot.org/story/12/07/1 ... nux-kernel

I've just given RoS 6 beta 3 a go and it still doesn't see the NIC.

From the changelog,

*) upgraded drivers and kernel (to linux-3.3.5);

@MT: Have you guys disabled the kernel modules?

I'm also evaluating the possibility of using MT in a purely MS environment using Hyper-V as hypervisor. It'd be really nice if I were able to use ROS 6 there.

I'm also evaluating the possibility of using MT in a purely MS environment using Hyper-V as hypervisor. It'd be really nice if I were able to use ROS 6 there.

Well, at this stage, you can already. It's just that the interfaces are legacy interfaces and only run at 10/100Mbit.

1Gbit and 10Gbit won't work without those modules in the kernel.

Let's hope MT didn't disable them and maybe just forgot to include them in this beta.

Unfortunately, since version 5.x, even the "'Legacy NIC" in Hyper-V does not work. RouterOS detects the NIC, but it cannot pass any traffic at all. No ARP, no IP, nothing. :(

I have tried this on multiple different servers/PCs, with Hyper-V 2.0 and 3.0 (Windows 8), and it simply does not work.

you can check vmware offering. It is known to work in some configurations.

Unfortunately not an option for us (6 data centres, all Hyper-V). We'd pay good money to have even the legacy NIC working (as in ver 4.x).

I've just given RoS 6 beta 3 a go and it still doesn't see the NIC.

From the changelog,

*) upgraded drivers and kernel (to linux-3.3.5);

@MT: Have you guys disabled the kernel modules?

Normus, please dude... get us an answer for the Kernel modules. Why are they off or missing?

R

I've just confirmed that RoS v6 RC1 also has the kernel modules excluded or disabled.

MT: Are you deliberately doing this? Can we give up asking?

I was wondering if there was any news on this? I'm perfectly happy with the legacy NIC. We have a large Hyper-V installation and are looking to create a light-weight firewall/router dedicated to customers. We use a lot of other MikroTik hardware so we'd like to keep the same platforms so our support is relatively simple.

I installed the v6rc1 and the legacy NIC's show up but no traffic passes on them. Maybe I'm just missing something simple? Or is the really not working at all?

The NIC is visible, but does not pass traffic. I've spent many hours trying to get this to work to no avail.

There are all kinds of modules that have always been disabled in RouterOS. to enable them, we would need to do extensive testing. Unfortunately, currently other thigs have higher priority than Hyper-V support

We went ahead and setup Ubuntu with KVM to run MT. Works well. We configured an Ubuntu desktop VM running virt-manager to get a simple GUI to the console.

Wish it would work with Hyper-V...but this is a nice option and runs very reliably and fast.

ROS on HyperV would be a dream come true. We use HyperV for a lot of VDI deployments, and virtualizing the routers would be awesome.

All our ESXi deployments have virtualazed ROS, on HyperV deployments we have to use RBs.

Well, I guess I can keep dreaming and hoping

I'm deploying a new Hyper-V server, and due to requirements I am forced to implement a virtual MikroTik router, and because of these limitations, going to be running v4.17... I would happily purchase a new license, even a L6, and upgrade to something new, but it simply isn't an option here until support is added. I have deployed dozens of RoS installs in Hyper-V pver the years, most need license upgrades purchased to go beyond v4 or v5, but until there is support in the newer versions, I won't be spending a dime on new licenses for virtual routers.

There is absolutly no reason this should not be supported by MT, there are dozens of people on the forums asking for this ability, easily a couple hundred licenses already spoken for in this thread when support is added, and surely many (tens of?) thousands more will be purchased over time after word spreads that there is a very stable router platform that works very well in a virtual environment, supports advanced routing protocols like BGP, and has nearly all of the capabilities of expensive cisco hardware.

You have the potential to market to an entire new segment of the IT industry here, and you're throwing it away. There are many IT shops and small ISPs around the world that use Cisco for all sorts of implementations, and many of those could deploy RoS in Hyper-V Clusters much cheaper, and gain much better uptimes and performance, as well as simplify management of the routers, but until you offer full support for Hyper-V networking, none of them will even consider MikroTik as a viable product.

I think yo ucan stop discussing it. This is quite a "intelligent design" type of religious matter. Someone high up in Mikrotik thinks that his balls will rot when they support Hyper-V and that their children will die and they serve a millenia in hell or something.

Given that Linux FULLY supporty Hyper-V now, and given the arguments from Mikrotik, I have decided to assume they see this as a religous matter. And we all know you can not argue with religious fanatics.

I am in the same boat - I have some situations where I can only sensibly deploy a Hyper-V vm to link a router into my fabric. I nicely hear "Use KVM, it is better" but hey, whoever the religious fanatic is saying that, what about you tell that TO MY CLOUD PROVIDER? Which is also note replacable because he is providing a LOT more under the Hood than "just VM's", which is Access to 40 Terabyte of data FOR my VM's.

Willing to pay, but Mikrotik does not want my Money. Well, their choice.

There are all kinds of modules that have always been disabled in RouterOS. to enable them, we would need to do extensive testing. Unfortunately, currently other thigs have higher priority than Hyper-V support

I'm actually interested, what would it take to get HyperV higher on the priority list?

Just tried RoS 6RC2. They've disabled the kernel modules on that release too.

Why disable them MT team, why!?

Who knows? What are you installing on hyper-v now?
I read about http://www.proxmox.com/products/proxmox-ve here, but maybe there is something else?

Just tried RoS 6RC2. They've disabled the kernel modules on that release too.

Why disable them MT team, why!?

we are not disabling them - they were never enabled in the first place.

I hate to be the guy that bumps an older topic but,I to am in the same boat need a way of making mikrotik virtual.

It seems to me that MK is just ignoring the customers it has currently and saying we will never touch it.
The kernel drivers have already been integrated into the kernel and there running out of excuses to not support it.
Sure they may believe that the open source world is key and not to even bother with a "paid" version o wait i forgot hv is free now.. sure its not open source but heck its free never less.

Someone high up is ignoring customers requests and for that my friends i will no longer support the MK brand; I mean there are two full pages of customers requesting this feature and hundreds more viewing it. Customers are how a business are driven based on there wants and needs. If they can not understand that then shame on them. If anyone is still looking for a solution vyatta support hyper-v just fine supports OSPF,BGP,RIP,NAT,Firewall and VPN.

It seems to me that MK is just ignoring the customers it has currently and saying we will never touch it.

Sorry but I have repeatedly said that we have no plans to make HyperV support.

It seems to me that MK is just ignoring the customers it has currently and saying we will never touch it.

Sorry but I have repeatedly said that we have no plans to make HyperV support.

Yes, and like the poster that you answer here I will now also look for alternatives. Your religious war interferes with your business decisions. Checking out vyatta now.

Hey normis, as I asked before, what would it take to get HyperV higher on the priority list? I understand you have no plans with HyperV atm, and I will not question your decision.
Im just wondering if its final, or if you just think there is not enough customer demand, or you simply want to do other more important thing before adding support, etc.

We of course dont want full HyperV support with integration components, etc. That is not availible for ESXi either.
I would be happy with the legacy adapter working, like it did in v4.

Full support oа ESXi is good too
Is there any more ways for me like a customer to vote for that feature?
Mtik make a cloud core router, whithout any idea from customers how to use it, but ignor real cloud problem with virtual routers.

yes there is, attend MUM

I recently was on MUM, and asked question about hyper-v. Answer was - no.
I wasn't alone in attempts to get another answer.

Full support oа ESXi is good too
Is there any more ways for me like a customer to vote for that feature?
Mtik make a cloud core router, whithout any idea from customers how to use it, but ignor real cloud problem with virtual routers.

i apologize mik staff, but this guy is right, from a biz perspective, perhaps your looking at the growth incorrectly
visualization, i dont care that you say there is not any plans, this is what we want, you don't make money from selling crap that people don't want.... this guy is right....

It seems to me that MK is just ignoring the customers it has currently and saying we will never touch it.

Sorry but I have repeatedly said that we have no plans to make HyperV support.

Yes, and like the poster that you answer here I will now also look for alternatives. Your religious war interferes with your business decisions. Checking out vyatta now.

and this guy is correct also

fair enough all the other top vendors have bigger budgets and have released many high end routers
and yes the ccr is surely a contender, but don't let the single focus of making hardware, and passing off the virtualization
train pass you. ignorance will hurt you.

It seems to me that MK is just ignoring the customers it has currently and saying we will never touch it.

Sorry but I have repeatedly said that we have no plans to make HyperV support.

Yes, and like the poster that you answer here I will now also look for alternatives. Your religious war interferes with your business decisions. Checking out vyatta now.

Maybe MT also should "check out vyatta"? They have a hypervisor preference poll on their frontpage...
50% ESX and Hyper-V at 3. place with 12% (of 5500 votes). Guess there are at least some potenisal customers among those +60%..
Seems like I have wasted some money for my MT license, too...

jafo, why 60%? only 12% voted for HyperV. And the first place - VMware ESX, is already supported.

Just based on previous posts in this thread saying something about missing ESXi support (17. & 20. dec)
As I don't intend to use MT on ESX I have not checked whats working or not myself.

Anyway, Hyper-V seems still to have a clear "no" from MT.

Any news? Or still no plans of implementing it?

Any news? Or still no plans of implementing it?

I had originally thought running a few KVM machines just for MikroTik would be ok. Now we're running about 300 VM's and the only ones that aren't on Hyper-V are the MikroTik VM's. It is making us rethink that we may need to move to something else. We're running about 50-60 MikroTik devices (RB's and VM's). Managing the KVM devices is really unpleasant in our environment.

Wouldn't it be possible to enable the drivers only in the x86 version, even if they aren't supported? Or charge more for that version?

Or let people install their own drivers and live with the consequences.

IManaging the KVM devices is really unpleasant in our environment.

Use the Proxmox VE. Is Open Source Virtualization Platform with support for almost everything what is needed ;) And it support OpenVZ and KVM and plays very nice with MikroTik.

Using Proxmox VE isn't really an option. The point is that we shouldn't have to modify our entire VM infrastructure to accommodate a particular vendor. We're standardizing on Hyper-V and System Center so we can use the new Azure recovery manager to simplify data center failover (http://www.windowsazure.com/en-us/servi ... y-manager/).

Just set up vyatta and pfsense since we're using MikroTik as basic firewalls in our VM environment (currently KVM, needs to be hyper-v). Vyatta free open-source version used synthetic NIC for Hyper-V without any install issues at all. Big drawback is the open source lack of GUI but does fully support Hyper-V. pfsense worked with legacy NIC but requires a boot script to cycle the de0 and de1 interfaces to work properly in Hyper-V (noted here: http://www.ern.nu/blog/2013/01/pfsense-and-hyper-v/). MikroTik doesn't support either NIC type (recognizes legacy NIC but no traffic). I should note that I was using Server 2012 as the Hyper-V host.

Just FYI for MikroTik on just two competitors.

I'm using an MT5 instance on a VM too (Proxmox PVE, KVM, virtio NICs). It's working fine. Also shortly tested 6, no problems.

But I can fully understand that those already having a Hyper-V infrastructure in place, would want to run ROS on it. All that is missing is some modules? And MT is not willing to include them on many people's request when it's not even extra work? How frustrating is that? Come on, Mikrotik, you don't need to test those modules. You don't test all of the others included either (it's apparent from the bug reports on the forums). And you don't really need to, they're standard modules on one of the most important platform for Linux which is x86. Just let those who need them test it by including them in your kernel. You can only gain by doing so, winning potential customers and license fees for L5/6. Just say it's officially not supported, they will understand. I don't really get the whole ordeal about this, really... Now the competition eats this piece of cake in the VM world instead of you.

Hello,
I've seen a few folks ask, but didn't see a response. What's the proper way to get a feature request elevated in the queue?

- Tickets?
- Request Count?
- Business impact metrics (i.e. 200 port count deployment)
- ???

I think a compelling case can be made, but understand it all comes down to resourcing.

Thanks,
Tim

I figured id come back and see how the mikrotik world is doing with hyper-v integration... after all its a lot easier to deal with than vyatta, After a year and two months their has been no change from mikrotik?

small bit of info you guys might find interesting or not..

http://blogs.technet.com/b/server-cloud ... erica.aspx

as others have asked how do we get hyper-v on the map, Its not going way and will only grow.

I think the Hyper-V graph there is interesting. I guess my question would be why is the only major virtualization platform that isn't supported Hyper-V? Linux distributions have fully embraced Hyper-V by building in the integration components. Apache Cloudstack and other cloud management systems are beginning to support Hyper-V. It seems like the trend.

My main driver for Hyper-V is so the routeros systems become consistent with the rest of our infrastructure. D/R planning and failover to a second site would be much easier if our routeros x86 VM's were on Hyper-V. Instead, they are the only one-off in our infrastructure. We use KVM for routeros.

We have real-time replication and live migration with Hyper-V. Learning to do that on a separate virtualization platform (KVM) just for routeros is not worth the investment. It is more cost effective to find a different router VM and I really don't want to take that step. Routeros has a ton of features and is a great solution for us. I really like it.

I don't even want a supported solution. All I really want is the NIC to work with hyper-v. I'm not sure how hard that would be to include that one component into routeros. We don't need all the integration components...just work with the NIC. Can't that just be treated as another NIC driver?

I realize it probably is a lost cause...but had some time to write a post!

none is supported, except KVM that is used in RouterOS x86 to provide virtualization. Also, even KVM is required to use virtio as virtual devices brake compatibility.

With Ubuntu/KVM, we use the e1000 NIC in our routeros VM's which works well with the gigabit switches we use with the underlying host. We do all the VLAN setup and management on the host OS, not routeros. That's our choice and we have to live with it. As I mentioned, I'm not looking for a supported solution. I don't expect Mikrotik to be the virtualization experts. That's our job. I'm just asking for the driver to be in the routeros (Hyper-V NIC).

We could not introduce the KVM virtualization within routeros into our private cloud. That would be an even bigger departure from the management systems for our cloud which today manage about 500-600 VM's. That's our decision...and we can make that within the context of what is supported and what is not.

A good analogy is modifying Microsoft's Exchange system to do multi-tenant setup (prior to supported configurations). They don't support it...but many hosting providers did it so they could offer a multi-tenant setup. There are lots of resources on the web for how to set it up (non-Microsoft resources), but if it breaks, you're on your own. Microsoft won't help. Each hosting company can make that decision based on the risk/reward. But Microsoft didn't disable the ability to set up it up that way. Eventually they saw how many people were running it and integrated a supported solution into the product. They even used that as the basis for the first versions of Office 365.

routeros is a great OS. It is small, fast, and efficient. I think adding this simple driver to it would only help growth...not hurt, and I don't think you have to support it. Just like you don't support my Ubuntu/KVM setup.

MikroTik cannot promise that RouterOS will work from a release to release. Either for RouterOS or your host virtual management system. If you look into the history you can see how it ended for XEN support in RouterOS.

I'm sorry, but I'm not sure I fully understand your post.

The part I'm confused about is "Mikrotik can't guarantee routeros will work from release to release". Are you saying something about the upgrade paths of the core routeros that you license to customers? Or are your saying that you don't guarantee any of the mainstream virtualization platforms like VMWare, KVM, and/or Hyper-V will continue to be able to run routeros?

Just trying to understand.

this one - Or are your saying that you don't guarantee any of the mainstream virtualization platforms like VMWare, KVM, and/or Hyper-V will continue to be able to run routeros?

Except KVM. That is working and will work as this is used for virtualization on x86 arch.

OK, thanks for clarifying. I'm not sure I agree with that strategy but that is clearly Mikrotik's choice. Personally, I think it is too bad. There is a role for virtual routers and firewalls. Routeros is perfect because of the small footprint and powerful platform. The idea of training staff and support personnel on a single platform for both virtual and physical routers and firewalls has a lot of appeal. But missing out on two of the three main virtual platforms for data centers makes the platform limited in its viability for that purpose.

It seems to me that adding a few drivers to the OS, supported or not, would open up those doors and provide a growth opportunity for Mikrotik. Of course, it isn't our decision as customers...that is yours as the manufacturer.

Thanks for responding to my posts with the answers.

is their any harm in including the driver? Microsoft has made it pretty clear they are not going away...

worse case can mikrotik look into adding routeros support for their Extensible Switch Extension? so we dont even need to have a vm of mikrotik running .


infohttp://msdn.microsoft.com/en-us/library ... s.85).aspx

I do understand, from reading this forum and also Microsoft/Linux documentation, that Microsoft Hyper-V drivers are included in Linux kernel 3.4 and higher.
Mikrotik is using kernel version 3.3.5

So I hope that as soon Mikrotik is implementing kernel 3.4 or higher Hyper-V integration is supported.

The question to Mikrotik development is: "When will Mikrotik implement Linux kernel 3.4 of higher??"

2 month bump on when will Mikrotik implement Linux 3.4 or higher.

2 month bump on when will Mikrotik implement Linux 3.4 or higher.

In RouterOS 7

Some news:
http://www.brocade.com/forms/jsp/vyatta ... gcn=&ggeo=

Brocade make vyatta distrib that makes that what we want from ROS in HyperV\esxi

Some news:
http://www.brocade.com/forms/jsp/vyatta ... gcn=&ggeo=

Brocade make vyatta distrib that makes that what we want from ROS in HyperV\esxi

This is in a completely different league. Look at the prices.

They have community edition (it's free)
And some open source forks http://vyos.net/wiki/Main_Page

it's a pity that Mikrotik ignores virtualization that way

3 month bump.
has their been any change in Mikrotik stance on hyper-v now that is has become an industry leader?

If not is ROS 7 going to be based on kernal 3.4 ?
when can we expect a ROS 7 alpha / beta? So the users of hyper-v and feal some love.

Sorry, no news regarding Hyper-V component integration in RouterOS. Not implemented. Not planned.

is ROS 7 going to be based on kernal 3.4 ?

Hello is there any one who is using Mikrotik RouterOS with VMware Vsphere 5.1.
I am using vsphere but there is a issue that VMNET3 network adapter is not suppoted in Mikrotik due to which we are unable to control CPU.

Yep, only legacy adapter at esxi. No upgrade
But it works... And pretty stable.

I have one more issue that when mikrotik get 100Mbps data its high internal loop back interface response like 40 60 MS.
do any buddy know how to control this ?

Hello,
Wanted to keep this thread alive and see how we can champion feature requests.

Mikrotik support for Hyper-V would allow me to have a single networking architecture across both physical and virtual environments, I imagine many customers are in a similar situation.

Just an update for people trying to get tick to work in hyper-v

installing the latest version of hyper-v 2016 aka tech server preview 2 their is no change.

+1 for hyperv support

so thought this was kinda interesting... figgured id share it here...



has mikrotik made any new decisions related to adding driver support for hyper-v ?
this topic has been requested for 4+ years...

based on the stats and based on what the customers are asking for can we get hyper-v drivers on the supported list? or at the very last allow traffic to pass... as the legacy nic's inside hyper-v do show up just dont pass data.

The company that I work for has been purchased by another. The stack that they use is VMWare based. Again, MT doesn't officially support VMWare even though RoS does work well on it.

I'm forced to reconsider using MT because I'd need them to support me if I had issues. I can't believe that MT is so far behind in the virtualization game.

They're simply losing out due to their ignorance of customer needs. Such a pitty because RoS is awesome!

Please MT, allow us to wake up from this nightmare!

Another bump for this age old topic. I suppose I have a few asks (frankly as a paying MikroTik customer):

- What is your roadmap for virtualization support (and specifically Hyper-V)
- As a customer, how do I properly and formerly request features and shape your Roadmap? (I work for a large software company, customer input greatly influences business decisions in conjunction with industry trends).

Thanks!

Hello,
I was reading the changelog of RouterOS v6.31rc14 and I noticed this...

*) chr - added support for HyperV fast ethernet

It talks about HyperV fast ethernet so I think that the synthetic network adapter is still unsupported but maybe it can be a sign that MikroTik is working to officially support HyperV in the future.

Someone already tried the v6.31rc14 on HyperV?

Thanks in Advance,
Davide

UPDATE: I just tried v6.31rc14 and I can confirm that HyperV legacy network adapter (Fast Ethernet) seems to work properly, the synthetic network adapter is still unsupported.

Hello,
I can confirm too.
After many years RouterOS seems to work on Hyper-V hosts

Thank you Mikrotik!

This is great news! Finally we can retire the specialized VM hosts just for routeros and integrate it into our existing Hyper-V environment!!

This is great news! Finally we can retire the specialized VM hosts just for routeros and integrate it into our existing Hyper-V environment!!

So far my tests on the release candidate are giving very positive results. Performance is amazing on a single Xeon core. I only gave the VM 128MB ram and a 1GB disk but WOW!

With VLANs the sky is the limit now

Thanks again Mikrotik!

Hello,
I hope that MikroTik support team will add HyperV synthetic network adapter drivers since FastEthernet can be a limit. I read that they added also the drivers for VMXNET3 adapter for VMware so I think that support for synthetic network adapter is on their roadmap. I think that MikroTik is working actively to enhance RouterOS to work on the most popular hypervisors since on the forum Normis announced officially the "Cloud Hosted Router" project.

C0ReDuMP

Hello,
I hope that MikroTik support team will add HyperV synthetic network adapter drivers since FastEthernet can be a limit. I read that they added also the drivers for VMXNET3 adapter for VMware so I think that support for synthetic network adapter is on their roadmap. I think that MikroTik is working actively to enhance RouterOS to work on the most popular hypervisors since on the forum Normis announced officially the "Cloud Hosted Router" project.

C0ReDuMP

I missed that note in the change log. I'm so impressed that they added it. Will test on our VMware stack soon and post results.

Thanks again MT!

OMFG. Seriously? AMAZING. This means that at some point this year (not toooo far in the future), I will be able to retire a VM Running a Server 2012 R2 with Routing and Remote Access that has the only sense in keeping a VPN connection to our office alife from our US servers? And replace it with a Mikrotik.

FINALLY. It is good to see that those near religious war attitude has given and that finally Mikrotik starts to accept that as good (or bad - but really mostly good) as their hardware is, sometimes virtual is much better because it better matches the use case

THANKS.

currently, there is a problem in CHR image that non-legacy ethernet interface is not working properly. We will work on this to resolve the issue. Both interfaces will be supported in the future.

thnx for hard work!
I'm w8ing for full support hyper-v 2012 to buy lvl 6 lic.
On level 1 license only legacy adapter works, tests results will be later.

hp dl360

hp dl360

Did you try CHR?
http://forum.mikrotik.com/viewtopic.php?f=21&t=98981

Did you try CHR?
viewtopic.php?f=21&t=98981

nope, how CHR'd help me?

ISP in my city can give me only 100 mbit speed, so I don't need synt adapters so much.
Bigger problem that ROS sees only 1 core, but in VM presented 8 cores (Intel Xeon).
And at 30 mbit download speed this core load grows up to the 100%

Only CHR will have any kind of Hyper V support

Cant see interfaces on hyper-v MT vm anymore. Did hyper-v support get removed again?