Hacker News Reader

Co-author of the paper referenced in the blog post here.

1. Regarding bugs: There is a lot of code in your monolithic OS kernel. There have been, and will be, a lot of vulnerabilities in that code. Various sandboxing mechanisms notwithstanding, the vast majority of processes running on your system can potentially use all of that code as an attack vector. Unikernels let you switch off access to most of that "host" code, and, through the use of a library OS, contain only the minimal set of libraries needed to run your application in the "guest". Regarding updates: Valid point, but no different from any modern application of substantial complexity, except that now you also have to update (e.g.) the library providing its TCP stack.

2. Unikernels don't make that problem any worse. In fact, if you run on something like Muen (https://muen.sk/), it'll mitigate a bunch of these attacks by giving you a less precise RDTSC at the subject ("VM") level.

3. I don't follow. Implement what?

If the unikernel is running as a normal process, all that potentially vulnerable code is still on the box, and still potentially executable in the event of a vulnerability. The security selling point of a standard unikernel is that the all the potentially-vulnerable code doesn't even exist, because only the libraries that are actually used get compiled and linked into the app, and it has no code relevant to functionality that it's not actually using.

The paper suggests using system-call filtering via seccomp or similar to block off this attack surface, but if you trust seccomp to have zero bugs, you might as well do traditional process sandboxing a la Google Chrome.

The problem unikernels solve is that traditionally, a vulnerability in C code means "game over" and the box is completely pwned with all memory and storage directly accessible, while with a unikernel, a vulnerability means only that an attacker can access other functionality within the app, and can't rely on eg. a shell or a debugger being available to arbitrarily move bytes around.

> If the unikernel is running as a normal process, all that potentially vulnerable code is still on the box [...]

Correct. Conceptually the same applies to all Type 2 hypervisors. Type 1 less so, but you could still potentially exploit Xen and you have all of the dom0 to play around in.

> The paper suggests using system-call filtering via seccomp or similar to block off this attack surface, but if you trust seccomp to have zero bugs, you might as well do traditional process sandboxing a la Google Chrome.

If you do that, you will have to do a line-by-line analysis of the code you want to sandbox in order to determine exactly which syscalls it's using. With the approach presented in our paper the developer does not need to care about this. For example, she can develop her MirageOS unikernel as a normal UNIX process and switch to a guaranteed-to-work-minimal-seccomp sandbox with a simple change of target (build-time configuration option).

But yes, you are now trusting seccomp instead of KVM. I believe in giving people the ability to easily make that choice.

> The problem unikernels solve is that traditionally [...] can't rely on eg. a shell or a debugger being available to arbitrarily move bytes around.

That part does not change with the sandboxing mechanism changing. The stuff that's inside is still only your unikernel.

> you are now trusting seccomp instead of KVM.

Which do you trust more?

That's a very good question. The answer is, "it depends".

For the 80% case, on x86_64, I consider them more or less equivalent. KVM is used daily in anger to provide isolation (e.g. GCE, and now ChromeOS) and has been around much longer but you need to trust hardware virtualization which is a large attack surface on the CPU itself. Given what we've learned about CPU vulnerabilities over the last year, I wouldn't be surprised to find some lurking in the VT-x/SVM implementations.

Seccomp OTOH is difficult to use correctly for arbitrary/existing applications but exposes less of the kernel (depending on your metric, see our paper) and does not need hardware virtualization.

For the 20% case, where the stakes are higher (e.g. High Assurance), I would use something like Muen or SeL4 and run a disaggregated system on top of that.

I feel like (1) might be easier to explain as akin to the JS bundling+tree-shaking process.

I'm aware it's not an entirely accurate metaphor, but it might turn out to be a more accessible one for opening a conversation.

How does access control work? There are plenty of APIs that need to validate access (e.g. ping or raw IP sockets can require sudo). If it's all running in-process it seems like there's no security you can implement, or the security mechanism would end up being super-complex like inspecting the TCP packets (which may solve networking use-cases but doesn't do so well for all the APIs).

*EDIT: Unikernels are neat as replacements for VMs since the abstraction layer can be much less costly/higher performance. They cannot be used as regular applications running on the OS due to how security is implemented.

>There is a lot of code in your monolithic OS kernel.

This code is usually there for a reason and a unikernel will have to implement large swaths of this code anyway.

Networking isn't trivial and I don't think anyone will be reimplementing the TCP/IP, UDP/IP, ARP, DHCP and more stacks without breaking in a few bugs themselves.

Example, modern TCP congestion control requires a very precise packet pacing and timing.

Another example, TCP retransmission will be required at some point, the code doing this will have to run side-by-side with the app. This will also be some amount of code.

And all this just piles up and up.

The only real advantage I see for unikernels is that because all the hard work is done by the hypervisor they don't have to bother implementing device drivers and task scheduling but end up either replicating or using the hypervisor's own networking stack and a bunch of other subsystems.

> This code is usually there for a reason and a unikernel will have to implement large swaths of this code anyway.

The largest part of a monolithic kernel today is devices drivers, by far.

> Networking isn't trivial and I don't think anyone will be reimplementing the TCP/IP, UDP/IP, ARP, DHCP and more stacks without breaking in a few bugs themselves.

Sure. And then more people will use those stacks, and they will get better. The more the merrier, we have too much of a software monoculture anyway.

> The only real advantage I see for unikernels is that because all the hard work is done by the hypervisor they don't have to bother implementing device drivers

This. People continually underestimate the amount of work required to support the hardware ecosystem. This is also why rump kernels (note, not the same term as the unikernel known as Rumprun) are such an achievement, also very much underappreciated.

Well, yes, a lot is device drivers, but not most. You can make the Linux kernel, for example, minus almost all drivers. That usually slims down the kernel image by a few megabytes. More when you drop various other drivers and subsystems you technically no longer need. Though on most modern systems almost 80% of the drivers are modules and won't be loaded if not needed. The diskspace they consume is irrelevant for most intents and purposes (below 100MB on my distro).

> The more the merrier, we have too much of a software monoculture anyway.

Linux and some other kernels allow userspace apps to have their own network stack in userspace, latest kernels allow even larger sections of the networking subsystems to be entirely in userspace.

I think this approach should be favored over a unikernel since it uses the natural x86 privilege seperation between userspace and kernel.

A few thoughts:

1) (Putting on my black hat) Attackers don't care about bugs or exploits. They care about running their code on your system. Whether that is as simple as mysqldump or wget'ng a monero cryptominer to run on there it all is based on the premise that the monolithic operating system (whose design is ~50 years old and linux is > 27 years old) is explicitly designed to run multiple programs by multiple users. Keep in mind this design pre-dates commercialized virtualization (eg: vmware) and pre-date "cloud" (eg: aws). If we assume that you are already utilizing VMs (and you are if you are on any public cloud and you are in most private on-prem deployments) the VM then becomes your isolation model. Can you still attack the underlying infrastructure? Sure - but if you can root GCE or AWS I'd say we all have some serious thinking to do on the current state of cloud infrastructure. Contrast and compare that to all the ridiculous headlines you see every single day and the fact that every single RCE that is worth doing entails forking/execve a new process. It's one thing to have the instruction pointer - it's quite another to launch a shell that doesn't exist, link your program to libraries that don't exist, as a user that doesn't exist, download new code when you can't.... etc.

2) Not to belabor this point but side-channel attacks affect everyone and Intel has been taking the hard (in terms of market) approach of simply disabling hyper-threading on some of their hardware.

Security is the number one selling point for unikernels imo.

> We usually see wide-spread vulnerabilities in the middleware like OpenSSH

There has been exactly 1 remotely exploitable bug and exactly 1 information leak bug in openssh in more than a decade (I believe the exploit is actually more than a decade old, need to check though), it is by far the most secure daemon ever. It is hard to overstate just how much of an accomplishment that is. NOTHING matches it in security. This is even more impressive given how widespread it's usage it.

> We need authority if we are sharing peripherals.

I think you'll find that large websites ... just don't do this. The only real purpose of multiple processes on such webservers is administration.

> f we are not going to save memory and implement that as a static library ...

The reason you want to do this is that the slowest thing in modern processors is context-switching. Going from user-space to kernel space. If you have, say, a networking stack in user space you can avoid such context switching entirely. The speed gains are enormous, especially for webservers.

Ridiculous performance in networking also enables many more applications that just won't work without. What do you think about cluster-wide disks, for instance ? SQLite like databases that run against Petabyte-sized files, concurrently, safely. You're just not going to do that efficiently by having it done in the OS.

Just to touch on this point - you hit the nail on the head. Microsoft and others have measured this cost to be north of 30%.

Re: 1 - that might be true, but you still probably wouldn't want to co-tenant a sandbox'd process with a total stranger. The overhead of a total Linux kernel is still larger than a unikernel-native application, so if you're a cloud provider, encouraging and using unikernel technology makes a lot of sense.

Re: 2 - this is managed by the hypervisor. SEL4 can be run as a hypervisor and has been formally verified to be correct [1]. Trusting a proven system like SEL4 is a far-cry from trusting Linux's isolation primitives because we can make hard-guarantees about the behavior.

Unikernels have the advantage that they are basically backwards-compatible (you can run any VM on the infrastructure you develop, even if that infrastructure is tuned for unikernel-native applications). With Unikernels, you can achieve VMs that are lighter than containers [2] thereby increasing your customer-per-server ratio.

[1] https://sel4.systems/Info/FAQ/proof.pml [2] http://cnp.neclab.eu/projects/lightvm/lightvm.pdf

I'm not an expert in unikernels by any means, but it's not the case that it's the equivalent of a lightweight Linux distro plus some neat debug tools. I don't think the linked article does the idea justice.

Here [0] is a good discussion of unikernels (TheRaven knows what he's talking about). ctrl-f for "You absolutely can not do the same thing with a conventional *NIX OS"

https://soylentnews.org/comments.pl?sid=11888&cid=296337

I agree. One of the appeals of unikernels in my opinion is that (I think) they will eventually give us the things that Docker currently gives us, but with one less layer of abstraction (unikernels running on a VM orchestrator instead of containers running on a container orchestrator running on a VM orchestrator). I imagine that this is inevitable as a matter of efficiency, but it's probably such a small practical gain that we won't see much traction in the next decade.

Or you could just run containers without the abhorrent mess that is hardware virtualization...

The post discusses running unikernels without "the abhorrent mess that is hardware virtualization".

> I imagine that this is inevitable as a matter of efficiency, but it's probably such a small practical gain that we won't see much traction in the next decade.

I agree that if the advantages boil down to saving a hundred MB of RAM of two, and enabling additional compiler optimisations, it won't be compelling to that many folks.

Taken to the extreme though, we can start spinning up instances 'just in time' to deal with incoming requests. Doing this, you could run (and hopefully pay for) your instances for only a tiny fraction of the time your service is in a working/available state.

This has apparently already been demonstrated [0], but their demo has been broken for as long as I can remember. I presume it worked at one point. I don't know if any commercial Xen-based offerings can spin up an instance as quickly as their configuration did, but the idea is there.

[0] http://erlangonxen.org/zerg

I think container orchestration offerings are pretty much at the pricing model you describe. You can spin them up just in time and only pay for what you use. The advantage to unikernels is that cloud operators don't have to operate two layers of orchestrators, so they could pass those savings onto unikernel users; however, I don't think that will be the lowest hanging fruit for a long time.

My VM is lighter and safer than your container:

https://news.ycombinator.com/item?id=15610155

Unfortunately the tooling story is poor for VMs. There isn't a good analog for the Docker toolchain (build, pull, push, etc) or container orchestrators (e.g., Kubernetes). These stories are at least as important as "lighter and safer".

I'd disagree here. The entire public cloud model is built on top of VMs. While I have a strong distaste for most CFM tooling out there that is precisely what you're talking about. Chef, Puppet, Salt Stack, Terraform, etc. all fulfill this and all of these are popular in their own communities.

I don't doubt that there are analogs for individual components, but I don't think there is anything comparable to Kubernetes for VMs. I also doubt the quality of those individual components or their integrations (for example, how do I build a VM image in <1m, what image repo technology should I use, what orchestration technologies exist and how do I tell it how to pull from my image repo, can I run that orchestration tech locally or on prem, etc?). I don't think these are insurmountable problems; there just isn't the same amount of investment (or rather, the investment is comparatively poorly coordinated).

LinuxKit gives build push run for VMs. It was kind if designed to be a half way stage to a full unikernel model.

I've played around with LinuxKit; it's quite difficult and limited. I couldn't get anything off the ground. Promising for sure, but needs more investment before it's ready for general purpose use.

I'd argue the opposite, which is that the future will have even more layers of abstraction, not less, since that's been the trend for quite some time.

My prediction is absolutely not "fewer total layers of abstraction", but only that this one will be collapsed in time. You can collapse one layer of abstraction and add two more, which is sufficient for my prediction to be compatible with the "more total layers of abstraction" prediction. :)

This thread was very helpful:

> Explain this to me then: What's the difference, conceptually, between (A) a VM host running a unikernel capable of solving one particular kind of problem, and (B) an OS running on bare metal running a process capable of solving one particular kind of problem?

They are very similar. The main difference is the shape of the OS APIs. In the former scenario, the host provides things that look like CPUs, NICs and things that look like block devices, with no high-level abstractions (e.g. threads, sockets, and filesystems). If these are needed, then the unikernel will link in device libraries that provide them. In contrast, in the later situation the OS will be providing these even when they are not needed and they will add overhead to those code paths.

Additionally, the amount of process state owned by the OS is far greater in the latter case. Things like file descriptors, thread priorities and so on make it far harder to snapshot or migrate an OS process. Hypervisor interfaces tend to be as close to stateless as possible.

> In the former scenario, the host provides things that look like CPUs, NICs and things that look like block devices, with no high-level abstractions (e.g. threads, sockets, and filesystems).

To add to this, with the approach we're taking with Solo5 [1], hardware virtualization is only used as one possible sandbox/isolation mechanism. If you look into the code, you'll find that the abstractions presented to the unikernel by the hvt tender are much thinner than those presented to a guest OS by a traditional VMM (e.g. KVM/QEMU).

[1] https://github.com/Solo5/solo5

1st. It’s not going to solve that. But, I don’t think solving that is a goal of this approach.

2nd. In a pure-unikernel approach, applications can’t steal from each other because there is only one application. Isolation is left to the VM. The goal is to get rid of the redundant isolation of running a multiplexing OS over a multiplexing VM. In the as-processes model, the Linux kernel still provides sandboxing. The goal there is to be a bridge between traditional and unikernel development.

3rd. The goal is to boil down to just virtualization and only the libraries specifically needed by a single specific application.

We are already seeing exokernel concepts. On a Linix machine, most 3D graphics is handled in the userspace GL library. The kernel driver handles modesetting, manages memory, and arranges for the userspace library to be able to directly submit command buffers to the GPU via memory mapped areas. DPDK works the same way. The NIC ring buffer is mapped directly into the userspace process, which can then perform packet I/O without calling into the kernel.

Well, that's roughly how it's supposed to work in theory for graphics, but in practice the kernel ends up having to do fairly deep parsing of GPU command buffers for security.

See, for example: https://www.kernel.org/doc/html/v4.11/gpu/i915.html#batchbuf...

Intel's GPUs aren't exactly known for stellar performance, but I'm pretty sure that's old info even for Intel. Certainly this kind of parsing hasn't been required on AMD and Nvidia GPUs' 3D engines for almost ten years now.

There's still a good bit of it necessary even for new GPUs like Radeon R600.

See r600_packet0_check, r600_packet3_check: https://github.com/torvalds/linux/blob/master/drivers/gpu/dr...

Thanks for the very interesting references and discussion. GPUs are hard, huh? AFAIK, this sort of validation isn't necessary for NICs. (Question: why is parsing necessary to ensure security, instead of say relying on the IO-MMU?)

The IO-MMU only provides device-level isolation for accesses to RAM, but you really need process isolation, and not just for accesses to RAM but also for VRAM where the IO-MMU is not in play at all.

NICs are an unfair comparison because the commands for NICs are all generated by the kernel, which is trusted. For GPUs, commands are generated by an untrusted user space driver that translates OpenGL or Vulkan commands into command packets for the GPU. Those command packets contain memory addresses, so in order to achieve isolation between different user space processes in the face of a potentially malicious user space driver, the kernel used to have to validate those addresses in the absence of GPU page tables.

Anyway, this kind of validation really hasn't been necessary for a long time now, despite pcwalton's outdated information, because GPU page tables were phased in around 2010. (The first AMD chip to have them was Cayman in 2010. Since we're citing kernel drivers here, take a look at https://github.com/torvalds/linux/blob/44786880df196a4200c17... and note how the function skips any parsing when virtual memory is used.)

R600 is not a new GPU, don't know how you got that idea :) It was released in 2007.

Except that on Windows the kernel part is quite tiny. Still waiting for X to survive driver crashes.

I would (in theory), love a unikernel for the work I do. Our bare metal boxes run about 20 Unix processes -- cron, syslog, sshd, etc plus one server daemon. It's nice to have a familiar environment, but we could run on fewer boxes without a general purpose kernel and general purpose memory protection that we don't strictly need, especially for nodes that could easily be tweaked to forgo local disk access (because filesystems are tricky, I'd rather leave nonephemeral storage with something proven)

That said, we're not moving in the direction of unikernels, but it has a clear application in my mind. It's a fair bit of work to migrate, and I've yet to see a practical comparison, so it's unclear what the actual benefits over conceptual benefits would be.

What is the problem domain?

Chat service, in Erlang. I'd love to just boot to Erlang, and have as much of the system as possible be hotloadable, including the tcp stack. But I may be suffering from too much exposure to hot loading. :)

Sounds like a great problem to work on! When you have such a well defined problem it can be a great ticket to do good work since in some ways you can control the stack top to bottom. Have fun!

I'm not actually working on it, unfortunately. There's too much risk, not enough time, and we're in the middle of migrating to a different environment where it would be an even weirder thing. But I think it's a good example of a place it could work, and it would be a lot of fun.

That's a shame, because Erlang as a unikernel already exists (as https://github.com/cloudozer/ling, unfortunately unmaintained). I'd be really keen on seeing an Erlang port to the Solo5 API, which would give you KVM/bhyve/OpenBSD vmm/Muen/others to run on.

There's some evidence of it running with rumprun (which appeals more to me than Ling -- I want to run the standard BEAM, and I want to run it on bare metal, because I can fill multiple machines and don't want to exchange an OS for a hypervisor).

https://github.com/rumpkernel/rumprun-packages/tree/master/e...

http://www.erlang-factory.com/static/upload/media/1474729921...

I'm not sure if there's any news on it since then.

> We usually see wide-spread vulnerabilities in the middleware like OpenSSH

> Correct me if I'm wrong.

OpenSSH is probably the single most secure piece of C network software. It isn't exactly a good example for "wide-spread vulnerabilities in middleware".

Unikernel is about install controls to application writer.

Once developers get into the workflow of writing applixations that can have fuller control of its environment, rarely one would return back to the world of opaque infrastructure and framework.