Notes from SCALE10x — More I/O for the Cloud, Please

January 22, 2012 - by alexsalkever

A crew of Joyeurs went to Los Angeles to attend SCALE 10x - the 2012 Southern California Linux Expo over the weekend.

We went for a number of reasons. First, we were invited. Two of our engineers, Brendan Gregg and Robert Mustacchi, were on the speaker schedule. Second, we have been eagerly working with other members of the illumos community to grow the base of developers building applications and tools for this descendant of OpenSolaris (you probably already know by now that Joyent maintains and uses extensively SmartOS, an illumos distribution). A few people asked what we were doing at a Linux conference but after we explained that illumos / SmartOS are completely open sourced, they were happy to learn more. And we got some great feedback from people who are interested in developing on illumos that will help guide our community building.

Brendan delivered a well-received talk - "Performance Analysis: New Tools and Concepts from The Cloud". This is a topic we were asked about a lot in the booth - namely, how can I gain real transparency into what's happening in my cloud architecture and, in particular, in my CPU, file system, I/O and other core functions that remain largely inaccessible in other clouds such as Amazon's EC2 and Rackspace. Here's a link to the slides and a description from the program:

Cloud Computing introduces new challenges for performance analysis, for both customers and operators of the cloud. Apart from monitoring a scaling environment, issues within a system can be complicated when tenants are competing for the same resources, and are invisible to each other. Other factors include rapidly changing production code and wildly unpredictable traffic surges. For performance analysis in the Joyent public cloud, we use a variety of tools including Dynamic Tracing, which allows us to create custom tools and metrics and to explore new concepts. I'll discuss a collection of these tools and the metrics that they measure. While these are DTrace-based, the focus of the talk is on which metrics are proving useful for analyzing real cloud issues.

This is an excellent primer on what's so cool about DTrace and, by extension, Joyent Cloud Analytics. What do you think? Send us feedback as we'll be posting more on this topic. And thanks for reading.