Muli Ben-Yehuda's journal

I want to update this thing more often, but there’s so much going on, the days filled with action and counter-action, that before I know it it’s past midnight, and I have to wake up at 5 AM for a workout, and updating the blog is left on the TODO list for yet another day. Like, today.

So, content?

I’ve been a manager for a month and change now, managing the virtualization and systems architecture group at the lab. It’s an interesting challenge (which is why I agreed to do it), often frustrating, occasionally exhilarating. To my surprise, the part I like most is dealing with human beings in their myriad forms. To my non-surprise, the part I like least is the bureaucracy, but I figured I’d wait a couple more months before I start tilting at wind-mills. I still write code (well, debug code, mostly) and conduct research, but it’s no longer the most important part of my day.

On the research front, we had two papers accepted to ICAC 2009 (one full paper and one short paper/poster), both in the general area of treating virtual machines as black boxes and inferring useful things about them—performance bottlenecks and boot-time–via statistical analysis of their inputs and outputs. Another paper, on the DMA mapping problem in direct assignment, was not accepted to USENIX ATC to my disappointment, and we are now revising it while looking for a new home.

I am continuing to work out twice a week with a private trainer who is seriously kicking my butt. It’s rare when I don’t finish a workout on the brink of exhaustion, drenched in sweat. I *love* it. Twice a week is no longer enough—I crave the endorphin rushes and sore muscles—so I’ve also re-started going for long walks, and hitting the punching bag in the back-yard like I really mean it. The kilograms are coming off, too, an added bonus.

Last but not least, SYSTOR 2009 is coming up next month, with a great program combining academic research and real-world systems. See y’all there!

It’s 5:40 AM. I am is sitting in an empty room full of half-assembled furniture, waiting for the personal trainer to arrive and whip my ass into shape.

There will be a half-day workshop at the Technion’s EE department on Thursday afternoon on “Technology Transfer – from Academy to Industry” which looks mildly interesting. I am on nominally on vacation this week and flying to Italy that night, but perhaps I’ll go anyway. Anyone else planning to go?

Our new paper is online: “Scalable I/O—A
Well-Architected Way to Do Scalable, Secure and Virtualized I/O”,
by Julian Satran, Leah Shalev, Muli Ben-Yehuda, and Zorik
Machulsky. This is an overview paper showcasing the main ideas
underlying a system we’ve been working on on and off since 2004. It’s
not as detailed as I would’ve liked due to the space constraints, but
hopefully it will be followed by more detailed papers. The
slides
I’ll be presenting later today at WIOV ’08 are also available
and go into a bit more details in areas.

Today in both virtualized and non-virtualized systems the entire I/O
functionality is based on device drivers. They are central to any
system structure; both anecdotal and informed evidence indicates
device drivers as a major source of trouble in the classical OS and a
source of scaling and performance issues in virtual I/O, due to
“trusted intermediary” required for the shared I/O. We propose an
architecture which virtualizes the entire I/O subsystem rather than
each I/O device, and provides device-independent I/O at higher level
of abstraction than the traditional I/O interfaces. In our suggested
architecture the system robustness is increased by isolating drivers;
efficient and scalable virtualization becomes possible by a complete
separation of the I/O and compute function and introducing a
protection model that does not require a trusted intermediary for
I/O.

New online for your perusing pleasure: “Direct
Device Assignment for Untrusted Fully-Virtualized Virtual
Machines”, by Ben-Ami Yassour, Muli Ben-Yehuda and Orit Wasserman,
IBM Research Report H-0263.

This is a short paper describing and evaluating our work earlier this
year on direct device assignment in KVM, using Intel’s VT-d IOMMU. Not
much new here if you’ve read our other IOMMU papers, but it does make
two contributions. First, it’s the best (and only) available
description (IMHO) of KVM’s direct device assignment code, and second
it’s yet another data point on the relative performance of device
emulation vs. virtual I/O drivers vs. direct device assignment. As
always, comments appreciated. The abstract follows.

The I/O interfaces between a host platform and a guest virtual machine
take one of three forms: either the hypervisor provides the guest with
emulation of hardware devices, or the hypervisor provides virtual I/O
drivers, or the hypervisor assigns a selected subset of the host’s
real I/O devices directly to the guest. Each method has advantages and
disadvantages, but letting VMs access devices directly has a number of
particularly interesting benefits, such as not requiring any guest VM
changes and in theory providing near-native performance.

In an effort to quantify the benefits of direct device access, we have
implemented direct device assignment for untrusted, fully-virtualized
virtual machines in the Linux/KVM environment using Intel’s VT-d
IOMMU. Our implementation required no guest OS changes and—unlike
alternative I/O virtualization approaches—provided near native I/O
performance. In particular, a quantitative comparison of network
performance on a 1GbE network shows that with large-enough messages
direct device access throughput is statistically indistinguishable
from native, albeit with CPU utilization that is slightly higher.

This is not serious, I’m supposed to remember what I was doing three days ago? I can barely remember what I had for breakfast this morning.

I started walking again in the mornings. Today I was up before the crack of dawn for a brisk walk on the sea shore, and when I got back home, I even had enough energy left for a few rounds with the boxing bag. Finished reading Haruki Murakami, and now re-reading Living the Martial Way. It’s a funny little book, so earnest it’s hard to take it seriously, but with nuggets of wisdom nonetheless.

The WIOV 08 website is up, including the full program! See y’all there.

Thursday: just another day at work . In the afternoon, went to meet an amazing carpenter (US: cabinet maker). Spent three hours going over the plans in minute detail, making lots of changes, and then he told us how much it was going to cost. Staggered to Noga’s cauldron for a late dinner.

Friday: BBQ with old friends at Ira’s. Once upon a time it would’ve been all Linux hacking, all the time, but now business and what the kids are doing is that much more interesting. Progress, of a sort.

Saturday: a day of rest and recuperation. In the evening off to Mika’s 1-year old birthday party. I still remember the sense of accomplishment we felt at Yael’s 1-year old birthday party, that we actually managed to raise her and she is fine. Resumed reading Haruki Murakmi’s What I Talk About When I Talk About Running.

A day of odds and ends. The WIOV schedule should be going up today or tomorrow. Continued looking into the feasibility of a new project which will require coordination with an inordinate amount of people. Worked on a bunch of new patent disclosures. Hacked a bit on a new idea for IOTLB design until the serial port server stopped giving me love, and then went into paper reading mode.

Last year I helped conceive and write a proposal for an ambitious EU project called Reservoir: Resources and Services Virtualization without Barriers. IBM Research Report H-262, 2008, “RESERVOIR—An ICT Infrastructure for Reliable and Effective Delivery of Services as Utilities” is now available. This research report summarizes the (172 pages…) proposal and describes the key ideas underlying Reservoir.