

What's The Best Linux Distribution For Clustering? 57
syn1 asks: "There has been a proliferation of Linux distros over the last couple years. Many are specialized for specific tasks or needs. In terms of Beowulf Clusters, there are a growing number of distros specialized for these clusters. Although the old favorite among specialized Beowulf distros is Extreme Linux, other distros such as Syclid Linux and Scali Linux are catching up in terms of user share. Additionally, more people are using conventional distros (Red Hat, Debian, Mandrake, SuSE, etc..) and adding Beowulf support. I am just wondering what fellow Slashdotters think about these various distros when it comes to Beowulf Clusters and which ones they think are best."
Beowulf Beowulf Beowulf... (Score:1)
Debian (Score:1)
Depends... (Score:1)
SUWAIN: Slashdot User Without An Interesting Name
Obviously! (Score:2)
You should have read the all-encompassing Linux-HOWTO!
Or better yet, the more specific, completely non-generic Beowulf-HOWTO!
Everyone knows that.
-------
CAIMLAS
Re:Beowulf Beowulf Beowulf... (Score:1)
Works very well. The restrictions on mmap
call though can be a problem, as it prevents
process migration from the home node.
This is indeed... (Score:2)
(note: looking at how obvious the submitter tries to start a distro-war, it doesn't take to much cleverness to lure the editors)
Wanna bet there will not be any useful discussion in this thread?
I really wish Slashdot would start moderation on articles too, then this would be dismissed as Flamebait fast enough.
the best Beowulf distro? Scyld, of course. (Score:3)
It's a second generation Beowulf, with some
very interesting features (see below). You can
download it for free or purchase it for cheap
(see link at http://www.scyld.com/ )
http://www.scyld.com/clustering_overview.html
[...]
Scyld Beowulf installation is easy. It's like loading Linux onto a single PC.
The Scyld Beowulf software provides the capability to start, observe, and control
processes on cluster nodes from the cluster's front-end computer.
Scyld Beowulf's cluster process control, BProc, decreases time to start processes
remotely. With process migration times of ten milliseconds, BProc provides an
order of magnitude improvement over other job spawning methods. Additionally
BProc provides insight into job and cluster performance.
Scyld Beowulf features Large File Summit (LFS) support via Scyld's Linux kernel
updates and GNU C library which support 64 bit file access on the ext2 filesystem.
Scyld Beowulf also includes utilities modified to take advantage this. (Basic text
utilities, scp, ftp client and server).
Scyld Beowulf includes GUI-based cluster node configuration, control and status
tools.
Scyld Beowulf ships with a customized version of the popular MPICH message
passing library. This version is modified to take advantage of the unique process
creation and management facilities provided by BProc which makes running MPI
applications easier than before.
Scyld Beowulf includes MPI-enabled linear algebra libraries and Beowulf
application examples.
Clustering is way custom (Score:4)
1) Process clustering - This beowulf, it is designed to rip every last shred of CPU time out of boxen. It is a VERY custom, machine dependant thing. A good B-cluster will be so hand tweaked as to be almost unrecognizable as what ever distro.
2) Server clustering - this is failover stuff, and distros can do this much better. Most people call it something like High Availability. But you are still likely to teak it up.
This is not a very good question, because clusters tend to be so custom. Its like asking: "Whats the best frame to base a kit car on?" There
-- Crutcher --
#include <disclaimer.h>
High Availability Clustering. (Score:2)
I've done this myself, and without starting a flame war, I've found that the easiest setup was achieved using RedHat. Their piranha tools make things easier and since the servers came with RedHat, I didn't have to waste too much time, nor did I have to drop a couple thousand dollars for their cluster distro, it all comes in the general distribution. During research for this project I read quite a bit about the TurboLinux distribution. The internals aren't much more than lvs, but the price tag scares you away (not that you couldn't do it with a stock TL and LVS, but to use their special distro it costs ... just like RedHat's. You're not really paying for the software, but rather the tech support). Whatever you decide, keep in mind a few things ..
1. Any distro can do it.
2. When you get the cluster up, do what you can to keep the distro/OS in the cluster the same. You'll save yourself a good bit of headaches in administration and make using the weighted algorithims a reality (ex: NT won't respond to the uptime, or ruptime polling requests, so you're stuck with the static weight that you assigned read the HOWTO for more).
3. If you are using lvs, use direct routing. It's fast.
Clustering ain't just Beowulf (Score:2)
What is all this Beowulf crap? For highly-available systems, clustering usually means server fail-over. It means an active-standby configuration with a shared disk. If the active server dies, the standby mounts the disk, starts up the app, and carries on.
For examples of shrink-wrapped versions, see Sun Cluster [sun.com], Veritas Cluster Server [andhttp], and a Linux based one, Turbo Linux Cluster Server [turbolinux.com].
A lot of services have to be active-standby; only one server can be doing the job at a time. Any database falls into this category, including SQL-based, LDAP, and mail stores. This is where the above products would get used. For services that can be active-active, like web servers, DNS, mail relays, some form of load balancing is better and cheaper.
There are distributed databases on the horizon, but few of them are ready for primetime. These would feel more like a Beowulf cluster.
I'm not trying to tell you that calling Beowulf a cluster is wrong, but limiting clustering to just Beowulf is.
Latin... (Score:1)
Caveat Emptor = Buyer Beware
Better Questions? ACE, others (Score:2)
Clustering, high-performance computing in general encompasses a huge number of problems and solutions. There are literally gobs of different routes one could take. Beowulf and Benchmarks, while easy to remeber and look at, are not the solution to everything. Perhaps you need the vector performance of a Cray or maybe the cache-coherent shared-memory system of a Data General AViiON or Silicon Graphics Origin. It all depends on your needs. Do the research before assuming you need one exact solution.
FWIW, you may want to look at SGI's Advanced Clustering Environment [sgi.com] for an all-inclusive, free, open-source solution. It's available for both SGI MIPS IRIX and IA-32/Intel Linux and works quite well with SGI's great Performance Copilot analysis software. They also know a thing or two about high performance computing. If you need more power you can build a warehouse of Linux boxes or a buy a 512-processor Origin 3000 [sgi.com] (w/ 1TB RAM and 714 GByte/sec bandwidth)... or a cluster of those!
My $0.02
Re:Eerste Post (Score:1)
Re:Beowulf Beowulf Beowulf... (Score:2)
--
Extreme Linux is a little out of date (Score:4)
For you I would like to recommend some reading:
Building Linux Clusters by David HM Spector [oreilly.com] published by O'Reilly, (hmmm site seems to be down, come back later, or check Google cached version [google.com])
This book comes with a CD together with clustering software. It also comes with step-by-step instructions. I believe, however, that there are some errata, which means that some hacking will need to be done to get your cluster online.
It also goes through some aspects of choosing hardware etc...
A more in-depth resource, without step-by-step instructions, but with in-depth discussions on granularity of Beowulf systems and whether they are actually good for the tasks you have in hand is:
How to Build a Beowulf, A guide to the implementation and application of PC Clusters [mit.edu] by the MIT Press
Also check the The Beowulf Project Site [beowulf.org] and the The Beowulf Underground Site [beowulf-underground.org]
Have fun!
---
Re:Clustering is way custom (Score:2)
I never really seen number 2 as a cluster solution. But wasn't that the first NT "cluster"? I think it was because MS called their failover system a cluster at that time that people started calling a failover system for a cluster. I might be wrong, but it's just my impression. Oh well never mind.
--------
Moderation on story submissions? (Score:3)
I really wish Slashdot would start moderation on articles too
You really wish you were looking at Kuro5hin [kuro5hin.org]. All logged-in users are always moderators at all times, and all logged-in users can vote +1 or -1 (remind you of [e2] [everything2.com]?) on story submissions in the public queue.
Re:Depends... (Score:1)
When you set one up on your favorite distro of linux, you are accountable.
For me a bigger issue is the reusability of code written with one type of cluster in mind to another type of cluster. Anyone have any experience with this type of thing? NonyMouse the Coward
"Best"? (Score:1)
I've used 4 distributions in the last 2 years. I only use Linux, at the moment I however rank them by which is least sucky.
My intention is to have my N machines properly clustered, so I read this thread with excitement. However, if the conclusions are XxxXxx or XxXX, then I'll give up right away.
FatPhil
(Xxxxxx is least sucky presently)
TurboLinux Cluster Server (Score:1)
Red Hat provides a package called High Availability Server [redhat.com] that includes load balancing, fault tolerance, and improved scalability for IP-based applications.
--Loge
Re:Sparc clustering (Score:1)
Re:One word (Score:1)
MOSIX DIPC (Score:2)
No one seems to be able to answer the question... (Score:2)
Personally, I find that while Red Hat is not my favorite of the linux distros, Red Hat offers Red Hat Professional Services, and this is a very nice thing for management, if the cluster in question is going to be a in a production environment at a company or business somewhere. If it's for your home use, do what you like, but most PHBs tend to take extreme comfort in the fact that if something linux related breaks, they can call Red Hat if the cluster admin on-site can't fix it, and Red Hat will either try to help on the phone, or you can pay for RH Prof. Services to come out to your site and take a look.
Re:One word (Score:1)
Re:Latin... (Score:1)
Blacklab Linux (Score:1)
I can't really think of any others, unless you're a crazy mac user. But don't listen to me; I'm just a crazy mac user.
--
Lagos
Re:Clustering is way custom (Score:1)
I think that someone just made it up to confuse people, I try to refer to them as a "HA Cluster" or a "Computing Cluter"
*Not a Sermon, Just a Thought
*/
HA vs. Parallel Processing (Score:1)
Heartbeat.. (Score:1)
(as the SuSE liker nevertheless ends up developing for RH..)
But missioncriticallinux.com's Convolvo says "any deestro".
But for actual stability like trying to get the job done? The last two VA Linux boxes I bought had RedHat on them already, and hardware cost is a pretty big factor. Or did you want to start repartitioning that 50GB RAID array? Is there such a big difference between deestros after you shut everything down? How about which HA distros not which Linux distro?
Someone's going to say BSD or die, etc etc. I'd much rather see people with actual experience responding and backing up what they say, and hear people with experience using the HA tools.
Better yet screw the distro idea, someone just post a list of tools they like and ideas about compiling, resource management, and security.
Re:Extreme Linux is a little out of date (Score:2)
For you I would like to recommend some reading:
Building Linux Clusters by David HM Spector published by O'Reilly, (hmmm site seems to be down, come back later, or check Google cached version)
That book is not very good. I wouldn't recommend it to anyone building a Beowulf Cluster. The examples are broken and there is a LOT of errata (a whole new ch2 from what I've heard?).
Go get the Scyld Beowulf2 Beta... really, it's the way HPLC (High Performance Linux Clusters) are going... it's easy to admin, easy to setup, easy to understand. It's a big step in useability for Linux clusters.
.laz
oh, and how much will you pay for /. ID 87 ;)
--
My car is orange, my sig is not.
Any distro can do it, but... (Score:2)
So, in *practice* your best choices would be, in my experience, RedHat for Beowulf-type clustering (process distribution) and TurboLinux for high-availability clustering (fail-overs)...
Re:Extreme Linux is a little out of date (Score:1)
How much anyone would pay would depend on your Karma!!
Here we go again!
---
If it helps... (Score:2)
Re:Moderation on story submissions? (Score:1)
While we're on the subject, Why does it need to be limited to the -1 to +5 range? How about a -100 to +100 range? Just imagine the bragging rights to get +100 Karma!
For the record, I'm in favor of NOT letting all logged on users being able to moderate. I think it's wise to make newbies wait a while to get a feel for the website first.
I'd be interested in finding out from an experienced Website Admin. just how much extra webserver load (if any) would result from letting all logged in experienced users moderate.
Re:Clustering ain't just Beowulf (Score:1)
A very good place to start looking at various stuff available for linux clustering is www.linux-ha.org [linux-ha.org].
Also worth mentioning if you think about the high availability (active/standby) configuration: if there's more than one service to be provided, you can get quite nice performance boosts by distributing active / standby roles on the machines in your cluster - having a database server for an ISP with oracle active on one node and postgres / mysql on the 2nd node gives you both great performance and high availability.
It means an active-standby configuration with a shared disk
Not necessarily; Personaly I like the solution of having seperate, local raid0 (or raid5) disk arrays in each of the nodes and keeping them synchronized over the network.
For a practical implementation of disk synchronisation at the blockdevice level have a look at drbd [tuwien.ac.at].
If you do want to go with shared media you'd best consider two seperate raid 0 or raid5 devices, each connected to both boxes (seperate scsi bus for each device). The two devices are then configured as raid 0 (mirror); if you throw in some scsi seperatores you should be set - the aim is to avoid the problems arising from a single device rendering the whole scsi bus unusable if it fails in a nasty way.
You'll still want to have some aditional hardware for your cluster: having a good method for I/O fencing (guaranteeing that both nodes trying to write to a device at the same time scrambling the data) is a realy good idea; the easiest way to achive this is to provide a method for one node to controll the others power suply; in case a node decides it has to take over functionality because the previously active node is no longer responding it can power down or at least power cycle the other node to make sure it's REALLY down and not just hung for a few seconds.
Designing and building clusters can be fun
Re:This is indeed... (Score:1)
Actually, you just posted the first distro-war related thread. Sad for you, you're making your prediction come true!
Re:the best Beowulf distro? Scyld, of course. (Score:3)
oh wait.
Re:Extreme Linux is a little out of date (Score:1)
>
> Building Linux Clusters by David HM > > Spector published by O'Reilly, (hmmm site seems to be
> down, come back later, or check Google cached version)
Check the readers comments section: 15 are extremely negative, only 1 is positive.
Re:Blender can't cluster! Check the offical docs! (Score:2)
Fortunately for SkyWriter, MOSIX isnt "clustering" since it turns a cluster of machines essentially into a single very-large SMP machine. All the program needs to know is how to thread or fork itself to use multiple processors. At least thats the theory. Never used it myself
BSD Clusters? (Score:1)
------------------------------------------
bes
Re:Clustering is way custom (Score:1)
Speaking of High Availability clusters, , check out this site [linux-ha.com]
I'm also quite keen on clustering, so when I'm back at my PC I'll rummage through my bookmarks and post some more links...
Off the top of my head, I also remember Cplant... [sandia.gov]
Then, there is Plan 9 [fywss.com]... Do check out their "Related Links" section!
Trian
High Availability Clustering (Score:1)
The Cplant! (Score:1)
Re:What about... (Score:1)
OR a Beowulf cluster of
As a beo admin/coder (Score:1)
For a real beo, go for Debian.
Those so-called "beowulf specific" distros just won't cut it.
Thanks,
Re:BSD Clusters? (Score:1)
Re:Clustering is way custom (Score:1)
Re:Clustering is way custom (Score:1)
Oh well, they are doomed anyway
--------
You need ease of use, availability and scalability (Score:1)
But even in High Performance solutions availability and scalability are things that are not to be forgotten. (Unless you don't mind your High Performance cluster to crash every week or so due to harddisk failures and overheating Pentium chips.)
To come back to the submission question, to my opinion a Distribution for a Beowulf cluster should have:
- a means to automate installation completely. If you miss this it would take you a lot of time to install a new machine each time one of the machines in your (500 node) cluster crashes.
- an easy way to update your nodes (for the same reason)
- a clear and understandable filesystem layout, that also protects your nodes from the clustered processes (you wouldn't want your
- hardware support for the devices of your choice (which include hardware raid mirrored disks, gigabit ethernet cards, fast io devices)
- a good 'out of the box' security policy
Now I don't know what distro has these features, but I might have to know somewhere soon in the future as I try to find alternatives to the really expensive O2000 10proc r10000 HPC cluster and the evenly expensive SUN E3500 / A3500 HA cluster we use at my job.
Besides that I think the point noted above are also true for a distribution that just has to provide a platform for a serious bussiness server.
For as far as I'm concerned there are some key functions I mis in scalability for the linux solution (or they just exist and I haven't looked good enough):
- a filesystem that is journaled, life growable, has exellent performance on >1 TerraByte sizes and can be attached to two or more machines enabling failover (like veritas vxfs) (could coda help?)
- an architecture that has the capability of real number crunching (like the O2000/O3000) while maintaining reliability and low prices (maybe alpha's are a solution here, or I just have to cool down and settle for 1000Mhz Pentium IV machines)
All experienced users moderate on E2 (Score:2)
I'd be interested in finding out from an experienced Website Admin. just how much extra webserver load (if any) would result from letting all logged in experienced users moderate.
All experienced users can moderate on [Everything 2] [everything2.com]. Each user who has [at least 50 XP] [everything2.com] (like Karma but you also get one for each write-up) is given 10 to 100 or more points per day with which to vote +1 or -1 on a particular write-up and cannot see other users' write-ups' scores until after voting on them.
Re:Latin... (Score:1)
Small Distros (Score:1)
Re:Blacklab Linux (Score:1)
Now that Apple's G4s ship with onboard 10/100/1000 gigabit ethernet, I wonder if gbit switches will come down in price a bit.
offtopic, please ignore :) (Score:1)
Linux CLustering Question (Score:1)
Re:"Best"? (Score:1)