Linux Clustering

SPK writes "A colleague and I recently discussed how New Riders's most highly regarded book -- Paul DuBois's MySQL -- corresponds to O'Reilly's worst dud: MySQL & mSQL. Charles Bookman's Linux Clustering does nothing to improve New Riders's reputation. The book is divided into eleven chapters, unevenly distributed among three sections: an overview of clustering for Linux, building clusters, and maintaining clusters. Four appendices provide brief information about online clustering resources, options for RedHat's 'Kickstart,' options for DHCP, and information on 'Condor ClassAd Machine Attributes.'" To find out why Krause was so displeased with this book, read on below for his review. Linux Clustering. Building and Maintaining Linux Clusters author Charles Bookman pages xv + 265 publisher New Riders rating 2/10 reviewer Steve Krause ISBN 1578702747 summary A guide to clustering software, networking, and journaling filesystems

Bookman emphasizes a central piece of wisdom that no system administrator should ignore: redundancy. In the case of high availability clusters, parts redundancy is the name of the game, but one should not forget the human component; no administrator should be caught with only a cell phone -- keep a pager just in case. However, in a post-modern turn that might seem brilliant if it were applied in a work of fiction rather than a technical book, the author seems to apply the concept of redundancy to the text itself.

That the book began not as a book but rather as a collection of talks or presentations, or some other smaller format, is evidenced by the repetition of information between chapters and sections. Such nearly poetic repetitions also occurs within sentences and paragraphs (e.g. "nightly backups each night" on page 25).

An editor never looked at Linux Clustering; the book had two "technical reviewers" but their contributions seemingly didn't include fixing mangled syntax and strained style. On page 14 in the second paragraph a large segment of a sentence from the previous page is pasted into another sentence, resulting in a nonsensical block of text. The number of hyphenation, syntax, word choice, and subject-verb agreement errors is atrocious and makes the book difficult to read.

Some of the misinformation in the text appears to be unintentional (but ignorance is no excuse for a UNIX systems administrator); some is due to the fact that the author deals only with old (2.2) kernels (though the book came out 18 months after the 2.4 kernel release), old versions of journaling filesystems, and old distributions; and yet other misinformation is the result of misplaced attempts at humor (such as stating that GNU stands for the Gateway Naming Utility; one can only hope that this was intended to be funny). Other jokes often misfire, but do point to the intended audience (consider, for example, the section heading "Space: The Final Frontier").

In the Introduction, the author indicates that the book should be read by "Linux enthusiasts and users who want to get a Linux cluster up and running with the least amount of fuss." The organization of the book will not, however, aid this enterprise, for there is little "how to" information provided, but rather a great deal of background information on compiling kernels, various types of journaling file systems, and RedHat's Kickstart (perhaps inappropriate considering that the book specifically states that basic information will not be covered). Another section or two deal with basic networking and security. Various types of clusters are discussed, as are a few of the types of clustering software (e.g. Condor and Mosix) available.

The book, however, is clearly intended for administrators of clustering systems; a special emphasis is high-availability and load-balancing clusters. Parallel computing and the types of applications end users would wish to run receive far too little discussion.

Almost all technical books regurgitate the contents of freely available FAQs and HOWTOs to some degree, yet the good ones summarize the relevant points, make dry documentation more accessible, and give the reader some new insights. Because Bookman's Linux Clustering suffers from heinous spelling, grammar, and style errors; deals primarily with outdated software; contributes little new to the discussion; and doesn't speak to non-admins, I can only recommend that those interested in Linux clustering stick to online FAQs and HOWTOs.

You can purchase the Linux Clustering: Building and Maintaining Linux Clusters from bn.com. Slashdot welcomes readers' book reviews -- to see your own review here, read the book review guidelines, then visit the submission page.

Comments

MySQL & mSQL

[Gortbusters.org]

Sometimes I still wake up screaming from reading that!

Re:MySQL & mSQL

[elmegil]

I didn't have to lay them side by side. As far as I can tell, most people here didn't. But apparently some people have a harder time recognizing the problems, and for them I recommend laying side by side to acheive enlightenment. One could argue that the inability to recognize the difference lies with the observer.... :-)

Re:MySQL & mSQL

[Large Green Mallard]

You know, I've always thought that book got a bad rap.. it was one of the first O'Reilly books I bought (after sed and awk).

Using it as a reference, and my knowledge of SQL, I did my first paid perl/php/mysql work while still in earnt University, and earnt close to $10,000 working from home, while studying. Having this work on my resume helped get me a $53,000/year job at age 21.

Ok, maybe other books could have helped too, but this was 1999, and like I said, the book did me well. I keep it at work for quick SQL reference.

No good books?

[bluethundr]

Why is it so tough to find a decent book on this topic? Even O'Reilly failed here.

Re:No good books?

[Gortbusters.org]

Perhaps they're not big sellers. I mean, you're not going to sell as many cluster books as you will PHP Cookbooks.

My friend worked at a lab in Princeton modeling the inside of a reactor. He worked with a 32 node linux cluster and did all the graphics modeling using a modified version of Unreal Tournament.

Re:No good books?

[EverStoned]

There just isn't any demand. Clustering is nice and all, but it's really just a linux gimmick. When you factor in power and staffage, it's much easier and cheaper just to get one really beastly computer than 20 old ones clustered.

Re:No good books?

[Wakkow]

Trust me.. The O'Reilly book isn't any good. I tried using it right when it came out for a summer internship project.

Perhaps it's because theses authors write books about clusters but never had to build one for any particular reason.

Re:No good books?

[HowlinMad]

Redundancyis the key our are missing here. If one cpu goes down ina cluster, the performance starts to suffer, but you are still up. If you one really beastly CPU goes down, you are down. That is the point o cluster. So while it may be "more" expensive, you are paying for the uptime.

Re:No good books?

[ComputerSlicer23]

Hmmm, that's why most serious clusters are built out of state of the are Dual Processor boards loaded with highend Xeon chips. Clusters aren't a "gimmick". There are some people who want to build one that have no use for it. How's that any different the car junkie who soups up his 1970's muscle car? It's just a gimmick. There's really no need to put a 500HP engine in it, no need to get it new paint, new tires, or a turbo, or a new dual exhaust system. It's just fun for them.

However, there are good uses for modification of vehicles, like say air bags. I don't call air bags, gimmicks, just because I think that guys who put dual exhaust systems on a 20 year old car seem like they are wasting money to me.

However, in terms of redundancy, your far, far better off with 10 P3 500's, then with one P4 5Ghz machine. One of the PIII's is having problems, shut if off, run the diagnostics. The P4 has problems, you shut if off, you are in deep shit.

If I had my choice, I'd rather have a cluster of 5-10 well built, redundant machines then one machine 10 times as fast for any problem that can easily be distributed (think websites, DNS, mail servers). No, I don't want to use 10, 3 year old Dell workstations to serve up my enterprise website, but I wouldn't have any objections to 10 Dell Servers that were bleeding edge 3 years ago assuming it uses parts that are still commonly available.

Kirby

Re:No good books?

[Jad LaFields]

My friend worked at a lab in Princeton modeling the inside of a reactor. He worked with a 32 node linux cluster and did all the graphics modeling using a modified version of Unreal Tournament.

Sweet. So did the reactor use the shock rifle alt-fire to start up or something? ;)

But more on topic, I think you're right about the size of the market. About how many big clusters are there out there anyway?

Re:No good books?

[malfunct]

It depends on why you are using the clustering, I use NT clusters at work all the time to increase reliability of the system. It doesn't help performance one bit in my situation (and isn't meant to) but if one node of the cluster goes down all the services start up on the other node and there is next to no downtime.

Re:No good books?

[afidel]

What kind of beastly machine goes down because of a single bad cpu? Only Intel based machines and none of those are all that beastly anyways. Big Iron does not go down to a single cpu failure. Where clusters rule is when you can chunk datasets into the ram space of a cheap rackmount box (2-4GB) and you have low enough interprocess communications that using comodity ethernet won't totally pooch your performance. If you meet those two requirements then the incredibly low cost per MIPS of x86 hardware will make it basically a no-brainer to go with a cluster.

Re:No good books?

[Usquebaugh]

Bollocks,

granularity, I'm tired of explaining this on /. but here goes.

There are two main typesd of cluster, High Performance and High Availability.

HPC tries to increase the power of the cluster by spreading jobs out over the whole cluster. HPC breaks the work down into blocks and farms this block out to the nodes. In the worst case a single node failure could cause the whole cluster to fail.

HAC tries to increase the uptime of the cluster by running the same job on more than one node. If a node fails then the job on the mirror node takes over. It's worth noting that no Linux cluster has 100% HA.

If a single node is going to fail 5% of the time what is the up time of a 100 node cluster?

I work in the commercial not the scietific world. HPCs bore me. HACs could be a god send.

Imagine if you will a cluster that automatically deals with node addition/subtraction. I have 1,000 users connected to this cluster using Xterms. I need more power, add more nodes. If any nodes fail the user never loses anyuptime as their work is switched to a mirrored node or nodes.

Centralised computing rocks.

Re:No good books?

[drgroove]

Clustering is nice and all, but it's really just a linux gimmick. When you factor in power and staffage, it's much easier and cheaper just to get one really beastly computer than 20 old ones clustered.

I'm sure Brin and Page would love to hear more about your unique insights into the cost-effectiveness of Linux clusters.

http://www.internetwk.com/lead/lead060100.htm
http://www.redhat.com/about/presscenter/2000/press _google.html
[each link opens in new window]

Re:No good books?

[Matrix272]

I've been looking for a good how-to clustering book, but haven't found anything yet. If anyone knows of one, can you post a link?

If anyone would like to send me on the righteous path towards clustering enlightenment, here are the details of what I have, and what I'd like to do... I haven't looked into clustering too much, so I don't really even know if this is possible.

I have 18 Linux workstations, and 1 server. I'd like all the workstations to work together to run all the processes on them, and also have a RAID 5-type architecture for the hard drives. Each workstation is a complete, stand-alone computer, with monitor, keyboard, mouse, etc. I need to have each workstation able to be logged into locally, and the cluster has to be remotely accessible (duh). I've read brief articles on Coda and Mosix, but I really don't understand all the particulars. The computers are in a lab, and run high-end software that generally takes several days to complete. Anybody have any advice for me?

Re:No good books?

[sogoodsofarsowhat]

Um sounds like to me the above poster is who edited the book... No if someone would be kind as to translate the above posting back into english or even enGRISH!!!! Try the preview button, and try reading your comments out loud...it might just save us all a lot of "WTF is he trying to say"...

Re:No good books?

[Anonymous Coward]

sounds like to me you need to get laid. Seriously, do the world a favor, buy a hooker, and then kill yourself. Don't forget to make a scene so you get your 15 minutes of fame on the local news.

Re:No good books?

[Gortbusters.org]

Heh, after work they would play the actual game on it. And did I mention it had an entire wall as its screen with some ridiculous resolution? OOH BABY!

Re:No good books?

[Ed Avis]

If you search on Freshmeat you can find various 'distributions' for Mosix that mean you need only stick in a CD to have the machine boot up and add itself to your cluster. At least, that's the claim.

Re:No good books?

[cyclist1200]

What happens when your one beastly computer goes down?

Re:No good books?

[BiggerIsBetter]

You mean like an LTSP server which is actually an OpenMosix cluster?

I don't think you'll get 100%, but possibly a "modified" version that generates paired processes and compares function returns are equal on each call... Granted, that's not switching but is actually redundancy, and therefore more expensive.

Re:No good books?

[Usquebaugh]

I didn't think LTSP used MOSIX? I thought it was just a stock Linux kernel.

The problem with your idea for HA is who watches the watchers? The process comparing the results must have a backup, so who checks that the checking process and it's back up are in sync? Quite quickly becomes a nightmare.

FYI, Tandem are the only company I know of with a fault tolerant intel server. They had to design their own mother boards as I recall. Lockstep of the processors was the way they went.

Re:No good books?

[BiggerIsBetter]

Out of the box, it doesn't, but it's no biggie to integrate the two of them.

My thoughts on the watcher-watching involve a P2P scheme. Each kernel knows about the processes it's running, and if it knows about the other kernel which is running the parallel process, the two can compare results. If it's good (eg, they concur), they send the result back to the process owner. If not, maybe try again, or migrate one or both processes to different nodes and mark the current one(s) are bad.

I think hardware is the only way to get real non-redundant process switching. Otherwise it's a matter of software dealing with problems "after the fact". Maybe something could be done with processes checkpointing across the network, or even something mad like STR in real-time, but man, the traffic generated would get huge and latency may be a problem. But if that's works as a concept, there's no reason you couldn't run a separate fault tolerance channel over iSCSI or Firewire or whatever.

Just ideas...

Paul DuBois, MySQL

[MattRog]

Not to mention he knows *nothing* of relational theory:
http://www.pgro.uk7.net/qu092902.htm

Re:Paul DuBois, MySQL

[ppanon]

Not to mention he knows *nothing* of relational theory:

It seems that a large majority of MySQL users also know little or nothing about relational theory. The MySQL core developers fought a long time against including support for foreign key constraints. Thus not knowing anything about relational theory may not be a drawback in writing a book about MySQL; it matches the target audience, even if it fails to educate.

Re:Paul DuBois, MySQL

[someguy42]

I feel a flame-war coming on. This comment is functionally equivalent to saying "Real men use VI" or "Real men use EMACS." It's only intended to start a "my (insert topic here) is bigger than yours" conversation. Real men know better than to assume that their personal preference is, by definition, going to be better than the next guy's preference.

Re:Paul DuBois, MySQL

[dasmegabyte]

Whoa, my personal preference is Sybase. MySQL, as used by most of the open source projects I've seen, is not a database. it's a data repository. Shit like "SELECT *"s with no joins and one-to-many relationships performed by adding extra columns.

This is text file stuff -- comma delimited stuff. And it doesn't help you learn SQL any more than writing a clever perl script helps you learn software engineering.

Re:Paul DuBois, MySQL

[Anonymous Coward]

Then, what's wrong with this statement???

"A PRIMARY KEY and a UNIQUE index are very similar. In fact, a PRIMARY KEY is just a UNIQUE index with the name PRIMARY. This means that a table may contain only one PRIMARY KEY because you can't have have two indexes with the same name."

At least "A PRIMARY KEY and a UNIQUE index are very similar.", makes sense to me.

Regards

Re:Paul DuBois, MySQL

[Anonymous Coward]

Whats wrong with that statement is that the whole point of relational databases is using the primary keys to link data together. The idea is never EVER to duplicate data in your database.

Re:Paul DuBois, MySQL

[Anonymous Coward]

At the risk of being trolled,

The quote in question:

A PRIMARY KEY and a UNIQUE index are very similar. In fact, a PRIMARY KEY is just a UNIQUE index with the name PRIMARY. This means that a table may contain only one PRIMARY KEY because you can't have have two indexes with the same name.

is referring to MySQL's specific implementation of primary keys. It has nothing to do with relational theory.

Re:Paul DuBois, MySQL

[ppanon]

This comment is functionally equivalent to saying "Real men use VI" or "Real men use EMACS."

No, it's not. It's equivalent to saying "Don't expect a book called RAD with Notepad to actually have anything significant to say about Rapid Application Development". You might be pleasantly surprised, but it's unlikely.

O'Reilly's worst dud was also about Linux clusters

[Frater 219]

Here's O'Reilly's worst dud.

How bad was it? It came with a CD-ROM that was supposed to automate the process of setting up a Beowulf cluster. None of the software on the CD-ROM worked. Running the install script printed out a message telling you to go to a Web site and download the newest beta version of the software. No such software was available ... ever. O'Reilly shortly withdrew the book ... and, reportedly at least, fired the editor who approved it for publication.

Want more details? Here you go. Waiting for this book, and then discovering slowly just how awful it was, set back a clustering project at my workplace by several months, by the way.

Re:Would you prefer RIAA math?

[Artemis P. Fonswick]

...the book is divided into eleven chapters, unevenly distributed among three sections...

Wow.

[jmt9581]

I was going to respond to that comment with a clever, humorous comment, but I think that I need to go negotiate a book deal with New Riders.

:P

What did you say?

[Anonymous Coward]

New Riders's most highly regarded book -- Paul DuBois's MySQL -- corresponds to O'Reilly's worst dud: MySQL & mSQL. Charles Bookman's Linux Clustering does nothing to improve New Riders's reputation.

Wait, my universal translator is busted here. Is this a new rider's book or an o'reilly book? How does "highly regarded book" do "nothing to improve new rider's reputation" ? Maybe you should have discussed it with fewer beers?

Re:What did you say?

[Anonymous Coward]

Interesting how his main criticisms related to the book's unreadability.

Ironic, eh?

Re:What did you say?

[Queelix]

Exactly. I can't believe this guy is complaining about a murky writing style while at the same time f-ing up a simple introductory paragraph.

Bah!

Re:What did you say?

[Anonymous Coward]

I think this is what he ment:

New Riders's book "MySQL" by Paul DuBois was similar in content to O'Reilly's "MySQL & mSQL". The implication is that quality was also similar -- the quality of New Riders's BEST book corresponds to the quality of O'Reilly's WORST.

The New Rider book "Linux Clustering" by Charles Bookman is the one that "does nothing to improve New Riders's reputation".

His statement probably could have been written better, but it is quite intelligible.

O'Reilly's Worst Failure

[s.d.]

In my mind, this is simple -- I have never read a worse O'Reilly book than Building Linux Clusters. There is a reason that ORA pulled this book out of print after only 6 months, and haven't even bothered to try to fix it and reprint a new edition. It was basically a commercial for the company the author ran, it read as if it hadn't been edited (spelling and grammar mistakes everywhere, included pictures were of the wrong thing that the text referred to), and the code included was so buggy it wouldn't work at all without a lot of fixing.

This was the first book on Linux Clustering I read, and I was hugely disappointed

Re:O'Reilly's Worst Failure

[Usquebaugh]

Yep,
all the above is true, bought the book, read the book, tossed the book.

Re:O'Reilly's worst dud was also about Linux clust

[Xpilot]

I bought that book... any idea how to get a refund?

O'Reily's Msql and Mysql

[mcc]

Uh.. what was wrong with MySQL and msql? I mean, it wasn't the most incredibly intensely mind-opening technical book i've ever read, but it wasn't useless, either. Far as i could tell the first couple chapters introduced you to SQL pretty well (which isn't exactly difficult, but they didn't really flub it), which you would never read more than once, and the rest was just various bits of random somewhat-disorganized reference material, sample sql, and sample database code in a few languages. It wasn't really any more useful than it would have been to have a printed and bound book that just contained the mysql manual, the dbi perldoc, and the manpage for the c database library.. and now that i'm used to mysql i just use the online manual.. and i will probably never dig my copy of the book out of the bottom of my closet never again.. but i don't really think i'm -sorry- i bought it.

And of course, it's been a long time since i first read the book, but i don't remember it being unpleasant. Why all the disdain?

Re:O'Reily's Msql and Mysql

[srw]

I agree. In fact, I often still use it as a reference. Granted, I'm not a university-trained RDBMS expert. Perhaps that wasn't the target audience. As far as I'm concerned, they could have left out mSql entirely (shudder... my first big DB project ever ran on mSql. I eventually converted it to Mysql.) but I guess it makes sense to include it as Mysql grew out of it.

Re:O'Reily's Msql and Mysql

[elmegil]

Y'all seem to have missed the new edition, "Managing and Using MySQL". Buy a copy, lay the two side-by-side, and tell us which one is more useful. My bet is you'll become enlightened as to how poorly organized and unfocussed the first book is.

Re:O'Reily's Msql and Mysql

[Schwartzboy]

I have to agree with the parent here. MySQL and MSQL is not, by any stretch of the imagination, the same caliber as any other O'Reilly book that I own or have read. If I had paid full price for my copy, I would probably have been disappointed. As someone who doesn't work professionally with MySQL and will probably never need the level of impressive knowledge and coolness that comes from most of the "in a Nutshell" books, however, it's not a particularly offensive read. Yes, it's probably less useful than the aforementioned online manual, but I wouldn't hold it up as the database world's answer to Valley of the Dolls or anything.

Re:O'Reily's Msql and Mysql

[irix]

I own maybe 30 O'Reilly Titles and "MySQL and mSQL" is easily the worst one of the bunch. Here's my take as to why.

I came from an Oracle background (i.e. already understood SQL, relational databases, etc.) and I was interested in 2 things; how to administer a MySQL database and how to do simple access from PHP/Perl.

Now go and pick that book and try and find that information. The description of the MySQL security model is muddled and confusing. No good details on how to do backup and restore. The examples for using PHP/Perl are horrible. The book has several chapters of filler.

A year or two later I bought the New Riders title mentioned in the writeup. It is a massive improvement over "MySQL and mSQL" - read them side by side and you'll see.

One thing that book taught me - just because it is a publisher you trust, don't assume the book will be good. Read it or read a review first!

Re:O'Reily's Msql and Mysql

[nagora]

I own maybe 30 O'Reilly Titles and "MySQL and mSQL" is easily the worst one of the bunch.

Ah, so you don't have "Linux Network Administration", then? Otherwise known as "10% of the information of TCP/IP Network Administration for 90% of the price".

TWW

Re:O'Reily's Msql and Mysql

[smack.addict]

The book was not really aimed at you. It was aimed at open source programmers and hackers getting their first taste of database development.

I think the book was great for that audience.

No book is for everyone. In the end, you can only judge it by its success communicating to its intended audience.

Fortunately, with Managing and Using MySQL, we did tackle the needs of people with dba backgrounds and dba needs.

Re:O'Reily's Msql and Mysql

[irix]

The book was not really aimed at you. It was aimed at open source programmers and hackers getting their first taste of database development.

Ok, I can buy that. Like I said, my bad for not reading the book first or reading a review. I just wanted a book on MySQL and I automatically grabbed the O'Reilly title.

I'm not a DBA, just a programmer that is familiar with databases. I guess for a more "advanced" user MySQL and mSQL is just not a good book might be a fairer statement.

Re:Oh please!

[MagikSlinger]

One of the major complints about this book is the lack of editting. Thes is just like those people that complain about mispelings on the front page of /.

His complaint was the lack of editing made it almost unreadable. I.e., re-reading a sentence several times to figure out what it said.

Good grammar and spelling aren't "windowdressing". They are essential for easy reading.

Re:Oh please!

[flok]

Weird sentence constructions can make it very hard to focus on the information.

So...what books DO you all recommend?

[SailFly]

For somebody who wants to learn about Linux clusters. I've played with Mosix and was impressed. What are good books and sources to learn about Linux Clusters?

Re:So...what books DO you all recommend?

[Anonymous Coward]

Scrolling to the end of the rather bitter review...

I can only recommend that those interested in Linux clustering stick to online FAQs and HOWTOs.

Re:So...what books DO you all recommend?

[jpcampbell]

I think the best thing to do would be download some of the open source tools that are out there and build a small one.

Some of the more popular that are used in the bio-sciences are.

http://www.platform.com/ (demo only, but most powerful)
http://www.openpbs.com/ (open source)
http://gridengine.sunsource.net/ (open source)

All these run on linux.

jpc

Re:So...what books DO you all recommend?

[terremoto]

For somebody who wants to learn about Linux clusters. I've played with Mosix and was impressed. What are good books and sources to learn about Linux Clusters?

I built a trial cluster of 5 PIIs back in 2001 using the 1.0 release of OSCAR from the Open Cluster Group. Their documentation (and software) was pretty good back then. I followed up the trial with a for-real-use 20-node cluster of Athlons (since expanded to 32 nodes) using OSCAR 1.1, which is still going strong. I haven't looked at the latest OSCAR packages but they're now at 2.2.1 and the docs may be better yet.

Other stinkers from O'Reilly

[talexb]

I have to add that 'Apache: The Definitive Guide (Second Edition)' was pretty horrible as well. Like the MySQL book, it was heavy on re-hashing available information and light on useful information like a dash of theory or a hint of how the authors used it to solve a particular problem.

And I hate it when O'Reilly comes out with a bad book, because generally their books are great.

Easy way to get a cluster up and running?

[kperrier]

Why buy one prebuilt, of course....

Re:Easy way to get a cluster up and running?

[DeathPenguin]

(Shameless plug)

Head over to LinuxNetworx now for a LinuxBIOS-ready Evolocity II cluster!

Note: I'm not an employee of LinuxNetworx, but they still kick ass.

Re:Easy way to get a cluster up and running?

[jpcampbell]

(Another Shameless plug)

Head over to RLX for a rack of 70 dual Xeon 3.0 GHz server blades all set up with the necessary clustering software and completely managed from a single web browser.

jpc

OK, so who's got a GOOD book on this topic?

[Elias Israel]

I hate to turn this into an Ask Slashdot, but truth is I could really use a good book on Linux clustering, especially if it covers:

1. Clustering (not just replicating) MySQL databases.
2. Network attached storage.
3. Load balancing and failover.
4. Probably six other things I'm not thinking of right now.

Anyone got any suggestions?

Re:OK, so who's got a GOOD book on this topic?

[ashpool7]

I wrote up a paper for my employer a while ago about most of those topics. The sad truth is that a comprehensive guide is not available, and most of the solutions are proprietary. However, there are a few bright lights.

Eddie: Load Balancing Software
http://eddie.sourceforge.net/

Linux Virtual Server Project: Clustering Tools
http://www.linuxvirtualserver.org/

OpenAFS: Efficient Distributed Storage
http://www.openafs.org/

Load-balancing and failover are tough nuts. You can do some stupid things like Round Robin DNS or Rotary NAT, but to be actual balancing, you need a balancer box. You can either make your own (using proprietary software or the stuff above) or buy a piece of hardware to do the job for you. I've heard Cisco makes some good ones.

NAS units usually operate using CIFS, AFP, or NFS, all of which are pretty lame options for a modern cluster. SANs are pretty cool, but you need some big-ass hardware to support them. Personally, I'm working on an OpenAFS cluster, which is pretty easy if you look into the capabilites of the software. Coda is another option of which I'm not using because it doesn't play as well with Windows.

As for clustering MySQL: If you read the Slashdot interview log they had a couple days ago, you'd see that the setup here is a master writer that replicates to a couple of reader databases. This is about as effective as it gets with MySQL. If you need higher power, I've read that commercial versions of Postgres support clustering/synchronization. More powerful than that and you're into Oracle territory.

Re:OK, so who's got a GOOD book on this topic?

[Anonymous Coward]

Why does everyone always mention Oracle, and only Oracle, when they discuss high end RDBMS packages? Oracle replication hardly resembles "real-time", unlike Sybase and DB2, for one thing. Moreover their sales people, and support folks, tend to be incredibly difficult to work with when you need answers, let alone answers in a hurry.

We have a 6 server load balanced, redundant cluster of Sybase ASE servers here, and they work beautifully. We've got a single pair of Oracle 8i servers trying to be redundant, and the "replication" we use that actually works isn't even built into oracle...its tar and scp. How pathetic is that?

Re:OK, so who's got a GOOD book on this topic?

[NetworkImpossible]

Right now, most of the information is scattered across a galaxy of FAQs and HOW-TOs. To get the best out of these, you need an overview of clustering, IMHO. The really old and wise clustering greybeards got this knowledge from working on VAX clusters.

I'm not that old or that grey; I got mine in the forerunner of this course (http://www.clam.com/education/q1554.html) at CLAM. It was expensive (and I see by the link it still is) but it is excellent. No relationship with CLAM (Curly, Larry, And Moe?) except as a satisfied student. Also took their basic AIX.

Re:OK, so who's got a GOOD book on this topic?

[Anonymous Coward]

>1. Clustering (not just replicating) MySQL databases.
PolyServe Matrix Server (not just clustering but
consolidating without replication! it also can
consolidate/run IBM DB2 and Oracle 9i RAC)
>2. Network attached storage.
PolyServe Matrix Server
>3. Load balancing and failover.
PolyServe Matrix Server
>4. Probably six other things I'm not thinking of right now.
PolyServe Matrix Server (if it's HA/load-sharing/
fail-over)

Seriously, check it out. It's commercial but very
nice.
GPL stuff: it's nice but you'd better have
a geek or two in-house. It's not easy to maintain
and re-configure.

You don't need any books; Google is enough.
If you're interested in HA/fail-over, get eval
versions of PolyServe Matrix HA, SteelEye
LifeKeeper and maybe one-two other commercial
software and try them first (because the docs are
good and the software is better quality).
After you're familiar with it, you can try some
GPL software.
Load-balancing: I don't understand why people
like to use such software; if you do it right,
you want QoS and convenient management (you buy
a switch). If you want it cheap, you use HA
software with fail-over and round-robin DNS
load balancing...

UnderAchiever

Re:Oh please!

[teaserX]

I realize that your post was meant to be a joke, but I have to point out that spelling errors and grammar mistakes do nothing to instill confidence in the reader. It seems unlikley that any of the information in the book was verified, nor the sources checked, if no none could bother to run the text past a spell checking program.

Re:O'Reilly's worst dud was also about Linux clust

[espo812]

Sell it on e-bay in a few years. The rare/novelty factors should make it sell for better than what you bought it at - there's your refund plus interest.

Your sig

[Anonymous Coward]

"...there's love and darkness and my sidearm" -- Moby

Said the short, bald, vegan pacifist.

Re:Oh please!

[the_rev_matt]

I'm going to go out on a limb and say he was trying to be funny. He succeeded, in my opinion.

I didn't find the O'reilley

[BiteMeFanboy]

MySQL book to be THAT bad. I didn't pay for it, so of course I wouldn't have felt cheated. We had a red one around for a while but I forget who put that one out. It was a little better.

Unevenly distributed?

[sczimme]

From the review:

The book is divided into eleven chapters, unevenly distributed among three sections:

That's good news: I would hate to read a fractional chapter.

Re:Unevenly distributed?

[PetWolverine]

You'd think for a book about clustering, they'd be able to at least distribute the chapters evenly.

mSQL

[sporty]

I know mSQL had a release in Jan past.. but does anyone know the merits of using mSQL vs MySQL? We all have pounded the MySQL vs Postgres vs Uber-commercial (oracle, ms-sql, sybase) to death.

Re:mSQL

[ratboy666]

mSQL is small and fast. Doesn't do anything except manage tables. Very simple. I find it useful for quicky projects that need a database.

And if something works with mSQL, it is pretty certain that it will run with anything else.

Anyway, its off-topic, but I hope it helps.
If you need socket bindings for mSQL, email me at
fred_weigel at h o t m a i l dot c o m.

Ratboy

Re:mSQL

[MattRog]

I wasn't aware mSQL was still under development. In any rate, I couldn't see using mSQL when there are better solutions that you are probably already using.

Re:O'Reilly's worst dud was also about Linux clust

[Anonymous Coward]

Their xml + java book (with the picture of some kid toy on front) is the worst ORA book I own (i got it for $5, and overpaid).

Second worst is the php pocket reference. It was ridiculously out of date, no index or corssreference, pretty much just a poorly sorted listing of functions with one-sentence descriptions that didn't tell you much that couldn't be guessed from the funtion name. At least it got money to the php developers, i guess. The online documentation is better

I agree

[Anonymous Coward]

that book is very poorly written. Beowulf Cluster Computing with Linux is a much better book.

MySQL & mSQL: Worst. Book. Ever.

[Traderdot]

Here's another vote for MySQL & mSQL as O'Reilly's worst book ever. That book singlehandedly ruined their previously stellar reputation for me.

Here are the books I've found most helpful on MySQL (and using MySQL with other things):

MySQL- Paul DuBois
MySQL and Perl for the Web- Paul DuBois
PHP & MySQL- Welling & Thompson

Re:Would you prefer RIAA math?

[WIAKywbfatw]

I think it's pretty clear that the original parent poster was pointing out the redundancy in the story text. Why was there a need to point out that the 11 chapters were unevenly distributed? It's not as if they could be evenly distributed is it? That's what the guy was saying.

Re:O'Reilly's worst dud was also about Linux clust

[DNS-and-BIND]

Nix. Microsoft Bob is still going for $15, less than its original list price.

New Riders

[Anonymous Coward]

There *ARE* some good New Riders books - it's just that they tend to deal with digital art:

"Digital Texturing and Painting"
"Digital Lighting & Rendering"

New Rider's focus is more on the artist / animator / illustrator side of things - and at that they excel (the above two are those I'm most familar with, and they are excellent).

I'm sure they'll gradually improve their hardcore technical books, but it's stupid to dismiss "all" their books as being bad. Just like O'Reilly has a reputation in some circles for being overly dry and out of date - *some* people find their books useful.

Re:New Riders

[bjohnson]

No sh*t. I own both, well thumbed. VERY good refs. Birns book on lighting and rendering is plain awesome. They also publish George Maestri's classic works on character animation. (in fact, based on the two books above I was very suprised to see that their computer books sucked so much.) Reading Birn's book let me do this which was at least an order of magnitude better than anything i'd done before...

Re:New Riders

[Un pobre guey]

I have "Digital Texturing and Painting," and can also vouch for it. It is an excellent book. The first few chapters are useful for any visual artist, regardless of whether she will be doing CG or not.

Re:New Riders

[MattGWU]

Don't forget "Network Intrusion Detection" and "Intrusion Signatures and Analysis". I'm a fan of both.

Re:New Riders

[elodan]

Not all New Riders' tech books suck either. I bought "Web Application Development with PHP 4.0" when I was just starting out with PHP: I was expecting a fairly beginnerish book, but it's actually an excellent text, covering many aspects of web app development, from security theory to actual code.

That one at least is highly recommended.

Re: mysql replication and high availability howto

[ubiquitin]

O'Reilly's Linux Hacks has one of the best explanations I've seen for setting up mysql replication. Load balancing and failover area are topics in their own right, but the Linux High Availability HOWTO is a good place to start. In general, the ibiblio site has been a helpful source.

Is this true?

[doc_traig]

Actually, it would be nice to slow down the pounceposters who stab us with one-liners as soon as an article hits the front page in order to grab that funny karma goodness...

Gah!

[autechre]

In attempting to point out that the original poster didn't get it, the replies have proven that they don't. He was saying that the word "uneven" is unnecessary, because 11 chapters can't possibly be divided into 3 sections evenly. Omit needless words! That's why it was funny. Sheesh.

Do you "get it" now, or do we need to lock you in a room with RMS?

Re:Gah!

[Artemis P. Fonswick]

In attempting to point out that the original poster didn't get it, the replies have proven that they don't.

No, I believe the replies have proven that the original poster was nit-picking just for the sake of nit-picking.

I read "unevenly distributed among three sections" to mean there's probably not an equal amount of coverage for each section. The statement was not necessarily redundant...it depends on how you read it.

I mean, I appreciate a clever jackass comment as much as the next geek...but you can't be throwing them out with reckless abandon.

Re:O'Reilly's worst dud was also about Linux clust

[Frater 219]

I bought that book... any idea how to get a refund?

Waited a while, didn't you? That book was withdrawn from sale two years ago. People who wrote to O'Reilly customer service at the time did get free books (an up-to-date copy of Linux in a Nutshell in my case, twice as thick and more useful for clubbing lusers^W^Winforming coworkers with) ... but I don't know if they'd do anything for it now.

Don't assume...

[xonker]

the book had two "technical reviewers" but their contributions seemingly didn't include fixing mangled syntax and strained style.

When you see a book in print, don't assume that the suggestions of the technical editors/reviewers have been heeded. The author basically has final say over the content of the book -- meaning that a tech reviewer/editor can be completely ignored no matter how much they complain about the content of the book or how much it doesn't address what it should.

And, the tech reviewers/editors are explicitly asked not to try to fix grammar and so forth -- that's supposed to be the job of a different editor.

Also... I'm surprised to see a review of this book popping up now, it was published about a year ago.

Re:Don't assume...

[PCM2]

The author basically has final say over the content of the book -- meaning that a tech reviewer/editor can be completely ignored no matter how much they complain about the content of the book or how much it doesn't address what it should.

If there's a single techincal book publisher out there that operates in this way, I'd be surprised to see them in business for long. That's like me giving you a business loan so you can staff your company with hamsters.

Re:Don't assume...

[xonker]

There are several, actually. I've worked with a few of them. New Riders is much better than a lot of them, but the fact remains that the author is usually free to ignore advice from the tech editors/reviewers -- and they often do.

Tech publishing is pretty slipshod with some of the major publishers. Their project editors and acquisitions editors aren't experts in the field, so they have no way of knowing whether the author knows what they're talking about or not. The authors sometimes recommend their own tech editors because, again, the editors working directly for the company don't know the field well enough to know who's an expert and who's not.

Much of the work is done by freelancers who may be completely overloaded... it's a mess with very little quality control. O'Reilly is a wonderful exception to this. Tech publishers that are part of the huge press conglomerates really don't place quality control very high on the list.

As to staffing my business with hamsters... well, that'd depend on what my business model was. ;)

Re:O'Reilly's worst dud was also about Linux clust

[watzinaneihm]

Funny that the reviews of the book look like this
So much for trusting reviews.........

You picked low-hanging fruit (Jayson Blair?)

[Anonymous Coward]

You didn't have to know anything about the book to post what you did. The summary on the front-page was all you needed. +4 Funny??

Re:Would you prefer RIAA math?

[WIAKywbfatw]

Let me be the first to thank a work colleague for jumping to my defence. Let me also be the first one to call him a complete idiot for not checking just who exactly was logged in at the PC he posted from.

Oh well, I guess he's going to be buying the first round of drinks on Friday evening then.

Re:MySQL & mSQL: Worst. Book. Ever.

[elmegil]

The newer edition, Managing and Using MySQL goes a long way to correcting the sins of the first book. I haven't read these others to compare directly, but I find the new edition a lot more clear and useful than its precursor.

huh?

[z_gringo]

From the review: (but ignorance is no excuse for a UNIX systems administrator)

That's right. Also,

Ignorance is no excuse for the law.

MySQL & mSQL

[Chromodromic]

Er, I'll still have it, I still use it. I didn't think it was a terrible book, although I've always known it wasn't brilliant. Still, I've ended up using it actually quite a bit. The new book is better, no question. But if you have to lay the two side-by-side, well, that just means the first one isn't *that* bad. It would be nice if there was an authoritative resource for computer book reviews, some place the really tried to give the low-down and had a dependable, objective reputation ...

In other news: water determined to be wet

[Anonymous Coward]

It seems that a large majority of MySQL users also know little or nothing about relational theory.

Of course. If they did, they wouldn't be using MySQL.

why are negative reviews necessary?

[rjnagle]

Having the occasional less-than-glowing review on slashdot can be refreshing change of pace. But I feel that they are unnecessary. Letting these books lie in fallow obscurity is perhaps a better strategy especially when there are so many other books that need publicity. One problem with reviews of technical books is that they depend on the technical proficiency of the reviewer. What may seem bland and uninteresting to this reviewer may seem interesting to someone totally unfamiliar with the topic or someone who just wants it as a dead-tree reference guide.

BTW, I don't see anything wrong with regurgitating man pages and how to's in book form, especially if explicitly states. Mark Sobell's practical guide to rh linux 8.0 is a perfect example. This book includes man pages/how to's while including better examples.

Finally, if the reviewer didn't like this book, why didn't he recommend a title or two that did cover this topic well?

(PS, I do occasional book reviews for slashdot myself.)

Imagine...

[Daverd]

Imagine a Beowulf clust-- oh wait.

Interesting timing

[Snowdog668]

I saw the title for this review and was happy. Just this morning I was talking to my director about clustering, something I've wanted to toy with for some time so I was hoping for a good review. As has been pointed out, good books on clustering are hard, if not impossible to find. Oh well, back to the howto's for me.

On the funny side, my director started the conversation with "have you ever heard of a Beowulf Cluster"? He didn't understand why I couldn't stop laughing. I obviously spend waaaaaay to much time on /. . Anyway, he apparently read something in Scientific American recently about clustering and wanted to apply it to an auditing project that we're working on. Not a cluster but some of the underlying theory gave him an idea of dealing with all of the varibles we have.

Re:O'Reilly's worst dud was also about Linux clust

[SquadBoy]

No the worst book was the Debian book. http://safari.oreilly.com/?XmlId=1-56592-705-2 I also hated the cluster book but trust me the Debian was worse.

"O'Reilly's worst dud: MySQL & mSQL"

[presroi]

The topic reads: "O'Reilly's worst dud: MySQL & mSQL"

MySQL&mSQL was my first O'Reilly book, back in my old days in school. I spend many nights reading it and many classes trying out the things I read.

I still like it although it has become completely outdated now (at least my edition).

Maybe I should have a look at the /. archive. What was so bad about this book?

Re:"O'Reilly's worst dud: MySQL & mSQL"

[smack.addict]

Just for fair disclaimer, I am one of the authors and I naturally do not think it is one of "O'Reilly's worst duds".

In actuality, the book got a very good /. review. In fact, reviews on it tended to be fairly schizo. People either loved the book or they did not like it at all. I think this reflects partly the fact that we tried to deal with MySQL and mSQL together (a mistake, though it does reflect that at the time the market is was in transition from mSQL to MySQL) and that the book was aimed at MySQL programmers. If you were a MySQL programmer, you loved it. If you were a MySQL admin, you did not love it so much :)

Our current offering (Managing and Using MySQL) is, of course, dedicated solely to MySQL and addresses both programmers and admins.

Re:"O'Reilly's worst dud: MySQL & mSQL"

[Large Green Mallard]

I thought it was a great book for the time, see this message I just posted for how well I've done because of that book.

You're right that tackling both mSQL and MySQL was a bad idea in retrospect, but again, like you said, at the time it reflected market conditions.

Thanks!

Fault tolerant Linux clusters?

[Anonymous Coward]

Are there really any truly fault tolerant clusters available for Linux and MySQL? I mean like to have a set of two, three or more identical machines networked together such that the databases on all are kept in perfect sync with one another and a client workstation app connecting to that database which runs on the group, not necessarily exactly to one server machine, will automatically and seemlessly and invisibly to the end user switch over to another if one server machine quits. For example let's say I have a taxicab dispatch center in a city with a dozen workstations for the dispatchers who take calls from customers and enter them into the system to dispatch the next available nearest taxi. Suppose that Crazy Larry, the drunk psycho mechanic wanders in from the back shop into the front office's computer room and sledgehammers the main database server in a fit of rage. I need all the dispatchers workstations to automatically switch over to the backup server(s) without missing a lick, or the dispatchers even knowing anything happened to the system.

Is there any clustering technology available for Linux right now that will accomplish this? I know that if the servers were IBM RS/6000 machines arranged in a AIX/HACMP cluster and running Oracle 9i, this scanario is very do-able (provided that we eliminate SCO from the picture).

Re:Fault tolerant Linux clusters?

[Tsugumi]

Free? No... but veritas cluster server works fairly well.

Obligatory disclaimer: I have no connection to veritas other thanusing their stuff.

Mod parent UP

[Anonymous Coward]

Insightful.

Re:O'Reilly's worst dud was also about Linux clust

[Frater 219]

I thought it was pretty obvious that those reviewers never actually attempted to do anything with the book, or particularly the CD-ROM; they just browsed the former and never let the latter touch their Windows systems. From a cursory (read: sloppy and shallow) first impression it is not nearly as bad as it is on a closer look, and a trade-rag pseudojournalist with enough cocaine in his system would probably find it quite impressive.

Re:O'Reilly's worst dud was also about Linux clust

[Cyno]

That's why I don't buy many books anymore. I can get most of the relevant information for any current topic/projects from the internet. I think the most innovative thing created in the last few years was tabbed browsing, I love Galeon.

WOW, what a cool sig...

[RandyF]

Do you "get it" now, or do we need to lock you in a room with RMS?

Can I barrow it?

Re:O'Reilly's worst dud was also about Linux clust

[themightythor]

Waiting for this book, and then discovering slowly just how awful it was, set back a clustering project at my workplace by several months, by the way.

So, let me get this straight: you tried to implement a technology that you knew nothing about from a book that you'd never read and then were surprised that everything didn't come up roses?

Re:FUNNY MODERATION EXPOSED!

[Paradise Pete]

Worst of all, these changes were implemented without any acknowledgement or notice.

Sheesh, Buddy, it's in the FAQ.

O'Reilly's MySQL book isn't so bad

[Un pobre guey]

Their PostgreSQL book is incomparably worse. The first edition is essentially useless.

uneven sections?

[Anonymous Coward]

"The book is divided into eleven chapters, unevenly distributed among three sections"

so why is this a bad thing?

OK, so they fired the editor, but not the author?

[Anonymous Coward]

O'Reilly shortly withdrew the book ... and, reportedly at least, fired the editor who approved it for publication.

Yet, funny enough, the author still writes a Linux column for O'Reilly. (His most recent entry elicited an endless stream of angry reader responses, calling it "crap," etc.) The whole thing just makes me wonder who is BS'ing who. He almost sounds like one of those guys that magically gets his way out of anything. Can't you just hear this guy promising his great program to automate the cluster installation, during his pitch? "That alone will make the book worth while!" Then when the time came to deliver, he prob. had nothing but excuses. "I told the editor the date was too early!" Most of these book authors are guys who want to use their books as a loss-leader to promote their consultancies. They make no money on these books; they use them to get profile thru speaking engagements, etc, all of which leads to lucrative high-paying consulting. Whether the book is good or bad doesn't matter half as much as getting out there in a timely manner and building brand recognition for themselves.

I think negative reviews can be helpful...

[scootr1]

At the least, they can keep me from spending my hard-earned cash on poor products.

However, I do agree that it should be a requirement that if you leave a negative review, you should also show an alternative title.

In other news...

[chaos-five]

Q) What do you call the computer lab of a group of open source monks?

---

A) A linux cloister.

chaos-five

Re:Oh please!

[jxliv7]

Please, oh! To concentrate on weird sentence structures hard make it? On facts concetrate not structure. Enlightened you will then be.

Speed Up Your Site

[pigwin32]

Andrew King's book Speed Up Your Site is the most enlightening book I've read in a year; well organised, well structured, readable, teeming with useful information, and it's published by New Riders. Eric Meyer on CSS, also New Riders, comes a close second. Rule of thumb, there's dog food in any stable.

Linux Magazine June Issue is Clustering

[Anonymous Coward]

the june issue is entirely clustering.

Re:Linux Magazine June Issue is Clustering

[Snowdog668]

Thanks for the info. Just might have to pick that issue up.

pots and kettles

[potsmaster]

it's pretty funny that a reviewer who trashes a book for its grammar employs faulty grammar himself...

Re:MySQL & mSQL: Worst. Book. Ever.

[JebusIsLord]

I didn't know the older edition was bad... I have the newer one though, and I concur - it is pretty good.

There is no good linux cluster book...

[dargaud]

When setting up my 24 processor cluster, I did read a lot of book reviews, but no one was satisfied by the 11 books I found. This is probably because clustering is a very dynamic medium, where patches are experimental, software is used only by a few groups, and once stability is reached, no one wants to touch anything anymore !!!

So I read online, whatever I found that was up to date and settled on the satisfying OpenMosix and... it works ! :-)

A solution for simple hight availability services

[frankie_guasch]

If you only want to have some servers giving redundant services, like web or mail, this is quite straightforward to configure and use.

Pen, a load balancer for "simple" tcp based protocols such as http or smtp. It allows several servers to appear as one to the outside and automatically detects servers that are down and distributes clients among the available servers. This gives high availability and scalable performance.

Ok, it's bad

[Amomynos Coward]

Ok, this book is bad but image a Beowulf cluster of those!

Thanks for the lesson

[zptdooda]

What I learned:
1. don't mention Beowulf again, even in irony
2. don't feed the trolls
3. don't mention moderation
4. don't comment on anticipated future negative comments
5. don't annoy
6. Last and most important: don't detract from the story or conversation

It was unintentional and I wasn't looking for replies, but I can see I made quite a number of mistakes in one post. It was a first and a last post of its nature from me. Not to be repeated.

O'Reilly Quality

[fm6]

The quality of O'Reilly's titles varies more than that of any publisher in the business. I blame this on a laid-back attitude combined with a hero-worship of the Larger-Than-Life Geeks they like to features as authors.

They must have pretty good editors, because they do avoid the mistakes most computer publishers make, like sloppy revisions that leaves Windows 95 material in a book on Windows NT. And they seem to have put a lot of work into their publication workflow. But it's pretty obvious that nobody at O'Reilly is in a position to tell the author to watch the stream-of-consciousness prose or question the relevence of his material.

Tech Book Trash

[PerfectWorld]

This is becoming very common in the tech book industry. All the tech book publishers are trying to foist all kinds of reall bad crap on us. Northcut's "Network Intrusion Detection" and the Solaris 9 exam cram book I started reading (and quickly dropped ... don't remember the author, some Prof.) both suffer from the same sort of issues.

Terrible writing style, bad use of language (not that mine is great, but I am not writing books), spelling mistakes, blatantly wrong information, "Technical Reviewers" who must be sleeping at the wheel ... I would be embarrassed(sp) to be one of those technical reviewers.

Not that this is a tech book industry issue. My wife rented "Atlantis 2" for my daughter ... wow, what an unbelievable piece of crap! But heck, the people producing this stuff aren't dumb ... they rest on their laurels while we keep buying their trash!

Mark