LWN.net Logo

Linux Foundation End User Summit wrap-up

From:  Kim Terca <kim-AT-pageonepr.com>
To:  lwn-AT-lwn.net
Subject:  Linux Foundation End User Summit wrap-up
Date:  Thu, 16 Oct 2008 16:20:22 -0700
Message-ID:  <F5B3E820-2989-4319-A43A-8ED97D86F3AA@pageonepr.com>


On Monday and Tuesday this week the Linux Foundation held the first  
Linux Foundation End User Summit in New York. Companies who attended  
included Credit Suisse, CME, AIG, Merrill Lynch, Dreamworks, NYSE,  
Fidelity, UBS, NYPD, US NAVY, Metlife, Morgon Stanley, JPMorgan Chase,  
Aetna, NAVTEQ, Goldman Sachs, Citigroup, Mitsubishi UFJ Financial  
Group (MUFJ) and many more. There was concern ahead of time that  
financial services companies may not attend due to the recent  
financial crisis on Wall Street.  We were pleasantly surpirsed,  
however, to have a packed house.  Perhaps in these times companies are  
committed to making the most of their investments, especially open and  
lower cost investments.

Highlights of the Summit included:

A detailed unveiling of the next generation file system for Linux:  
BtrFS.  Chris Mason of Oracle detailed how BtrFS will solve many of  
the issues customers are facing with all platforms as the volume of  
their data continues to increase.  BtrFS is currently being worked on  
by a group of Linux Foundation members.  Led by Oracle, this group  
also includes developers from IBM, Intel, Novell, and other  
companies.  The group hopes to have a version in the kernel by the end  
of this year.  Customers' initial reactions were very favorable on  
ease-of-use in particular.  Ted Ts'o of the Linux Foundation also  
detailed the new ext4 file system already in the mainline kernel.

Several users are using DM multipathing in production and were very  
interested in the newly reworked multipathing code that is now  
upstream (request-based instead of bio-based) which should give better  
control and insight into error handling.  Native device mapper (DM)  
multipathing was compared with commercial alternatives.

The presentation part of the file system focused more on data  
integrity features in the IO stack and file systems with an eye  
towards reducing exposure to possible data loss.  Specific features  
that are new include the SCSI T10 DIF work (allows end to end data  
integrity checks), checksumming support for data and/or metadata in  
ext4, BtrFS and XFS, and a substantial improvement in fsck time for  
ext4.

The developers in attendance learned that one use-case that many large  
financial customers have is a large number of mid sized nodes in a  
data center (say 2000 or so) which have relatively little data on  
local hard disks and rely on network file systems (NFS or AFS today)  
to supply the critical data.  There was interest in the still in- 
development work on parallel NFS (part of NFS v4.1 and heavily  
promoted by large storage vendors like EMC and NetApp) and some  
interest in persistent NFS client caching (again, actively under  
development, not in upstream).

Community representatives encouraged these large consumers of Linux to  
advocate for open source drivers from their suppliers so that they can  
be properly supported.

Also, they were offered help evaluating upstream (non-distro) kernels  
to give them a chance to validate and suggest fixes or tweaks on the  
very newest technology.  One large bank agreed to begin testing these  
newer, non-distro kernels and give feedback to the community.  On the  
competitive side, no users were found that had actually deployed ZFS  
in a production environment.

In the best example of the intention and promise of the event, end  
users, developers, and vendors met for multiple hours on the topic of  
systems management in Linux.  System management leaders received  
valuable operational requirements for performance monitoring from  
customers.  As a result of these meetings, the end user council of the  
Linux Foundation will now create use-cases and other detail that will  
then be given to community developers and vendors.  This type of  
collaboration is invaluable in advancing the state of the art of the  
Linux platform.  The Linux Foundation will report on results from this  
collaboration next year.

In other systems and performance management sessions, it was found  
that users in attendance had experiences and best practices but not  
always the awareness of the wide variety of tool/community choices.   
As a result of this meeting, a blueprint or web based knowledge center  
will be created on the Linux Foundation site.  Both the vendor and  
user advisory councils will contribute to this knowledge center and  
encourage its use by a wide audience.

Gerrit Huizenga from IBM gave a well received session on Cloud  
Computing and how Linux is the OS of the cloud.  The Linux Foundation  
will be publishing a paper containing some of this information in the  
next month.


Kim Terca
Page One PR - San Francisco


(Log in to post comments)

Linux Foundation End User Summit wrap-up

Posted Oct 17, 2008 18:22 UTC (Fri) by kragil (guest, #34373) [Link]

Btrfs in the kernel this year?? Will there be another kernel this year?

Linux Foundation End User Summit wrap-up

Posted Oct 17, 2008 22:19 UTC (Fri) by rahulsundaram (subscriber, #21946) [Link]

Linux kernel releases are roughly about every 2 to 3 months. So yes, another kernel is still possible within this year if it is a quick cycle. Andrew Morton expressed interest in merging Btrfs as early as 2.6.28 or 2.6.29. Even if it isn't this year exactly, it is going to be be available pretty soon but with caveats similar to Ext4dev up until recently. Large scale production deployments are probably going to have to wait for much longer than that however.

Linux Foundation End User Summit wrap-up

Posted Oct 18, 2008 23:29 UTC (Sat) by sbergman27 (subscriber, #10767) [Link]

From http://searchenterpriselinux.techtarget.com/news/article/...

"""
The roadmap for adoption includes alpha tests on laptops this year followed by alpha tests on servers next year, preview releases on community distros in 2010, incorporation in production OSes in 2011 and the start of enterprise adoption in 2012. Even if btrfs is "feature complete" in 2009, it will require extensive work to debug and tune for optimized performance before it is enterprise-ready, he said. Another critical step ahead is the addition of a user space file system consistency checker for file recovery, Ts'o said, adding that the last 20% of the project requires 80% of the effort.
"""

So this not a "just around the corner" thing. (I've never believed in a 20/80 rule. It's more like a 10/90 rule.) I cannot help but feel that the Namesys hype machine, the false start that was Reiser4, and the fan-base that fell for both may have cost us some time getting our future filesystem planning in gear. However, btrfs's feature set seems solid. And Chris Mason is certainly worthy of our faith. Ext4 looks adequate to get us to 2012. So I would say the future of Linux filesystems looks bright, if delayed.

Linux Foundation End User Summit wrap-up

Posted Oct 19, 2008 4:16 UTC (Sun) by eru (subscriber, #2753) [Link]

I'm a bit confused about ext4 and btrfs, what is the difference and what are the advantages of them over each other, or over earlier file systems. If I may offer a suggestion to the editors: I think here a LWN article summarizing the situation would be helpful.

ext4 and btrfs

Posted Oct 19, 2008 21:43 UTC (Sun) by corbet (editor, #1) [Link]

ext4 is an extension of the earlier ext* filesystems. It retains a semilar organization, and it has some of the same advantages and disadvantages. One big gap in ext4, even after all the work, is that it does not do checksumming of data (or metadata).

btrfs does have that checksumming, along with a lot of other features that people have been asking for (snapshots, for example).

The long-term plan is that btrfs will be the "next-generation filesystem" that we'll all be using. But btrfs is new, under heavy development, and does not yet have a stable on-disk format. OTOH ext4 is ready (or nearly ready) now. So think of ext4 as the interim solution which carries us through the next few years until btrfs is ready.

ext4 and btrfs

Posted Oct 20, 2008 10:20 UTC (Mon) by eru (subscriber, #2753) [Link]

Thanks for the answer!

Copyright © 2008, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds