Just got my approval... 2nd one for our team on storage features. One on Storage I/O Control and another a tech preview. Yea!

Several influential bloggers have now picked up on the PARDA research paper and its implication to the future of storage resource management. Here a few of note:

• Gartner's Cameron Haight: PARDA the Plan?
• Virtualization Review's Rick Vanover: Next Storage Frontier
• VMware's Duncan Epping: Project PARDA

Many thanks to these individuals for their favorable coverage.

There is an active Call for Papers for the Second International Workshop on Virtualization Performance: Analysis, Characterization, and Tools (VPACT’09). I feel very lucky to have been asked to serve on the program committee (PC) for this excellent workshop by Peter Varman, the general chair. Peter is a superb researcher and I really like the work that he and his students have been doing in the area of QoS for storage systems. PC membership means that I'll reviewing papers submitted to the conference and selecting the best ones for presentation and for publication in the proceedings.

If you have interesting ideas that you'd like to run by the research community in the following areas, please do consider submitting your work.

The workshop is intended as a venue for researchers and practitioners in academia and industry to present their unpublished results in the area of virtualization research. Papers are solicited on topics including, but not limited to the following aspects of virtual machine (VM) execution:
• VM analytical performance modeling
• VM performance tools for tracing, profiling, and simulation
• VM benchmarking and performance metrics
• Workload characterization in a virtualized environment
• Evaluation of resource scheduling
• Models and metrics for new VM usages
• VM energy and power modeling

Take a look at the list of program committee members to see if you recognize any of those names.

As part of our PARDA research, we examined how IO latency varies with increases in overall load (queue length) at the array using one to five hosts accessing the same storage array. The attached image (Figure 6 from the paper) shows the aggregate throughput and average latency observed in the system, with increasing contention at the array. The generated workload is a uniform 16 KB IOs, 67% reads and 70% random, while keeping 32 IOs outstanding from each host. It can be clearly seen that, for this experiment, throughput peaked at three hosts, but overall latency continues to increase with load. In fact, in some cases, beyond a certain level of workload parallelism, throughput can even drop.

An important question to consider for application performance is whether bandwidth is more important or latency. If the former, then pushing the outstanding IOs higher might make sense up to a point. However, for latency sensitive workloads, it is better to provide a target latency and to stop increasing the load (outstanding IOs) on the array beyond that point. The latter is the key observation that PARDA is built around. We use a control equation that uses an input target latency goal beyond which the array can be considered to be overloaded. Using our equation, we modify the outstanding IOs count across VMware ESX hosts in a distributed fashion to stay close to the target IO latency. In the paper, we also detail how our equation also incorporates proportional sharing and fairness. Our experimental results show the technique to be effective.

Ajay Gulati, Carl Waldspurger and I have just finished work on a distributed IO scheduling paper for the upcoming FAST 2009 conference. So I wanted to provide an update. PARDA is a research project to design a proportional-share resource scheduler that can provide service differentiation for IO like VMware already provides for CPU and Memory. In plain terms, how can we deliver better throughput and lower response times to the more important VMs irrespective of which host in a cluster they run on.

This is a really interesting and challenging problem. A bunch of us first started brainstorming in this area 2 years ago but despite several attempts, for over a year, we couldn't come up with a comprehensive solution. For one thing, IO scheduling is a very hard problem. Second, there aren't existing research papers that tackle our particular flavor of the problem (cluster filesystem). To top it off, the problem sounds easy at first blush, encouraging a lot of well-intentioned but ultimately misleading attempts.

Ajay and I first published a paper on our idea to use flow control (think TCP-style) to solve this problem at the SPEED 2008 workshop in Feb '08 and the feedback from the research community was encouraging (this later became the basis for an ACM SIGOPS Operating Systems Review article, October 2008). Since then Ajay, Carl and I have worked out the major issues with this new technique resulting in the FAST paper.

The paper is entitled "PARDA: Proportional Allocation of Resources for Distributed Storage Access".

I published an academic paper at the IEEE International Symposium on Workload Characterization (IISWC 2007) in September that I want to spend some time talking about. The paper was entitled "Easy and Efficient Disk I/O Workload Characterization in VMware ESX Server". Here's the abstract:

Collection of detailed characteristics of disk I/O for workloads is the first step in tuning disk subsystem performance. This paper presents an efficient implementation of disk I/O workload characterization using online histograms in a virtual machine hypervisor-VMware ESX Server. This technique allows transparent and online collection of essential workload characteristics for arbitrary, unmodified operating system instances running in virtual machines. For analysis that cannot be done efficiently online, we provide a virtual SCSI command tracing framework. Our online histograms encompass essential disk I/O performance metrics including I/O block size, latency, spatial locality, I/O interarrival period and active queue depth. We demonstrate our technique on workloads of Filebench, DBT-2 and large file copy running in virtual machines and provide an analysis of the differences between ZFS and UFS filesystems on Solaris. We show that our implementation introduces negligible overheads in CPU, memory and latency and yet is able to capture essential workload characteristics.

It seems a lot of people agreed with my previous post on the security of virtual switches. These include the originator of the information that prompted my blog post. Chris Wolf himself posted comments recognizing his misunderstanding. I think Chris did a great job of quickly following up after my blog post and getting in touch with us to resolve the misunderstanding. Kudos to him. Read his comments for yourself.

Next week, I'll be attending VMworld 2007, the virtualization community's annual conference. Actually I won't be the only one given that more than 10,000 people are attending this year! Furthermore, famous people are keynoting:

• Diane Greene, VMware's CEO
• Mendel Rosenblum, VMware's Chief Scientist
• John T. Chambers, Cisco Systems, Inc.
• Patrick Gelsinger, Intel Corporation
• Hector de J. Ruiz, AMD

This year, I'll be giving two talks: "Fast and Easy Disk Workload Characterization on VMware ESX Server" and "ESX Storage Performance - A Scalability Study".

VMmark 1.0 is now live.

This release is the culmination of lot of effort by the VMmark team. Big congrats to them to get this tool available to the virtualization community.

Having worked in the virtualization performance arena for about 5 years, I can attest to the need for standard benchmark. There are so many variables due to the layers of software running on layers and then resource sharing on top that it is very difficult to make sense of data presented by customers, partners, the press and the community at large. VMmark attempts to address this by creating a standard benchmark.