Some folks have asked about the sources of Nehalem’s performance improvements. Certainly a major contributor is its new on-board memory controller and the new QuickPath Interconnect (QPI).  Rather than having multiple sockets (each possibly containing multiple cores) of CPUs sharing a bus to get to memory in a Unified Memory Access (UMA) model, Intel with Nehalem has moved to a Non-Uniform Memory Access (NUMA) architecture. As the number of cores of compute power within a system increase, the more the need to have fast interconnects between the cores and their memory. Unfortunately at scales of greater than 4 or more cores, its unfeasible to have all components talking directly to all other components (uniformly). A single bus can be overwhelmed and become a bottleneck, and just cranking up CPU and bus speeds has failed to solve the problem because the amount that the crank can turn is limited.  Rather, components connect to other components, which then connect to other components. Each connection is very fast (especially when it is non-shared and there is no contention to have to mitigate), but a component take multiple hops across these fast connections  to reach some other components.  Some components are “closer” than others, so communication is faster. Thus the creation of NUMA architectures.

Read more…

In our industry, new CPU announcements are a dime a dozen. Most of them are simple speed bumps, or at the most an increase in the number of cores per socket.

The Nehalem announcement was different, in that Intel was solving the bus limits that have impacted overall system throughput.  The announcement was exciting but being able to buy systems containing those CPUs took a long time to reach fruition. Pre-release performance tests of these new Xeon 5500 CPUs revealed that the announcement was more than marketing. They show that Nehalem provides breakthrough increases in many aspects of system throughput. In many areas, the 5500 is twice as fast as its predecessor 5400. Unfortunately the 5500 is not a drop-in replacement for other Xeon CPUs. It requires a new chip set and socket.

Read more…

We’ve rescheduled the sixth NEOSUG meeting for March 11. Dave Miner will talk about the state of OpenSolaris and demo some new functionality, and I’ll talk about what’s new in the Solaris 10 update 6 (11/08) release. Hope to see you there. For full details and registration info have look at the NEOSUG discussion forum.

Interested in learning how Sun leveraged Open Solaris as the core technology of the Sun 7000 storage array? I will be presenting on the Sun 7000 at the January 28th meeting of the Open Solaris User Group (NEOSUG). The plan is to have a system there for demonstration and for everyone to check out in person.

Solaris Ambassador Jeff Victor will be presenting the latest features in Solaris 10 10/08. This list will include ‘ZFS Boot’ – the ability to install Solaris using only ZFS file systems.

Here is a link to the detailed agenda and registration information on the opensolaris.org website. I hope to see you there.

With the release of the Sun Storage 7000 line of storage appliances, Sun has included a new “Analytics” toolkit. These analytics are based on DTrace (http://en.wikipedia.org/wiki/DTrace), but essentially hide the DTrace complexity in a cloak of Ajax-based browser graphics. Through the GUI, a storage administrator can determine which clients are causing which files on the server to be “hot”, or resource use-intensive. Also the administrator can see the latency of each request to the blocks of that file, or how many request of each protocol are being processed, or how many cache hits a file had. In this blog I’ll explore the basics of Analytics.

The analytics component of the Sun Storage 7000 line can provide useful information to a storage administrator who is trying to manage and monitor the appliance and the files and blocks stored there. Just like DTrace, the analytics run in real time, and allow quick progression from hypothesis through data gathering to new hypothesis, data and conclusions. Unlike DTrace, the analytics component has a very complete and useful graphical interface and visualization engine.

Read more…

I’ll be teaching 4 half-day tutorials at the LISA conference in San Diego, CA. Hope to see you there (the week of Nov 8).

My October 2008 column has been published in ;login:. This month it’s about Solaris System Analysis – detailed steps to take to determine why a system is “slow” or “busted”. Some ;login: contents is freely available at ;login:, but my column this month is not one of them. I’ve posted the .pdf here for those without a USENIX membership (although I strongly recommend you get one if you are interested in all things Unix).

The wiki that started with my August 2008 column will be expanded (as soon as I get the time) to include this new content. It’s very lonely having a wiki of one, so please consider contributing your thoughts to what I’ve started. It would be a great advance in systems administration if there was a canonocal source of first-step debugging information, and hopefully you will help make this wiki that source: http://wiki.sage.org/bin/view/Main/AllThingsSun

My August 2008 column has been published in ;login:. This month it’s about Solaris System Analysis – a checklist approach to solving a system being “slow” or “busted”.   Some ;login: contents is freely available at ;login: August 2008, but my column this month is not one of them. I’ve posted the .pdf here for those without a USENIX membership (although I strongly recommend you get one if you are interested in all things Unix).

I hope this column will turn into a living wiki about (Solaris) system analysis. I’ve prepopulated a wiki with the contents of the column, so now it’s up to you to add your thoughts to the procedure. It would be a great advance in systems administration if there was a canonocal source of first-step debugging information, and hopefully you will help make this wiki that source: http://wiki.sage.org/bin/view/Main/AllThingsSun

Sorry for the delay in announce / posting this. My June 2008 column has been published in ;login:. This month it’s about the state of ZFS – features, functions, stability, useability, performance, production use, and so on.   Some ;login: contents is freely available at ;login: June 2008, but my column this month is not one of them. I’ve posted the .pdf here for those without a USENIX membership (although I strongly recommend you get one if you are interested in all things Unix).

Jim Mauro is our guest speaker, talking about DTrace and all things performance, at the next New England Open Solaris User Group (NEOSUG) meeting on Sept 10th in Burlington, MA.

Read all about it here: OpenSolaris NEOSUG.

Hope to see you there. Please register if coming so we can plan the refreshments…