essenially every benchmark ever disagrees with you.
IIS is incredibly fast. SQL server is constantly bouncing in and out of 1st place for tpm's, both absolute and cost adjusted. How is that slow ?.NET runtime is pretty fast for a garbagecollected JITted environment. Certainly faster than any known JVM for most things.
re: IIS security issues - you realize that the majority of IIS holes have to do with external components accessible via IIS, and now ship in a turned-off-by-default state ?
I'm not sure what you think is unproven about.NET. Two versions of it have shipped now (runtime 1.0 and 1.1) and there are some huge internet properties that are using it.
oh, you mean the meaningful replication that took down orbitz for a day or two ?:)
Re:Why are they running Windows then?
on
Can .NET Really Scale?
·
· Score: 2, Informative
if you have a reproducible environment where sql server 2000 is losing or corrupting data, please email me directly with details.
having said that, i think you are way off base. I used mySQL a lot and use SQL server extensively now. Comparing the two just isn't worth while. mySQL is a lot closer to berkeley DB than it is to SQL server, feature wise, scalability wise, management wise, and reliability wise.
SQL server has several licsensing options - per conection or per processor are the two default ones. Given that SQL server is perf competitive with Oracle, (it infact beats it on official benchmarks) and a HECK of a lot easier to deal with, SQL server is a STEAL compared to oracle pricing and support.
for most OLTP type work mySQL will NOT be faster than SQL server, especially as the volume of data grows, the complexity of queries grows, and the contention increases.
Put another way - when i see mySQL displacing SQL Server in the tpc benchmarks, i'll pay attention to you again. mySQL is very cool, but its a toy compared to DB2, Oracle, or SQL server. If all you need is a toy, by all means, use a toy -- you'll be happier. When you're ready to graduate to real systems, SQL server will be waiting for you. Maybe mySQL or Postgres will evolve faster than your needs do, and you'll never need a commercial quality database (or one of those two will turn into one)
It is unfortuneate that your experience has shown that Java+Jboss+{some db here, as you named like 5 of them}+Linux is "more reliable" than Windows +.NET. I suppose i can't change your mind and i definitely cant change your experiences, but i am highly skeptical that you're not using some rose colored lenses when you're examing how your app stack compares.. Windows + ASP.NET + SQL server works pretty well out of the box. I don't want to be too presumptuous here, but unless you've had work experience with one of a handful of simply huge internet properties, i can say that Windows + ASP.NET + SQL server is reliably running many sites bigger than you've ever dealt with. In other words, some very big places seem to think its just fine:)
i got the following email in response to this:
> I saw your comment here on/. and that you're using 12 spindles. Are these mirrored (RAID 1) to guard against data loss? Can you give me more specifics? We have a dual Xeon with a RAID 5 that we're looking to possibly upgrade. We have a second identical SQL box that is configured for passive failover. Do you have a failover machine? If so, how does it access these same 12 spindles?
here's my response:
Yes. We're using Raid 1 and Raid 10. Each disk is 36gb so realistically
we could probably get away with only raid 1s, as none of our individual files
is larger then 15gb or so. But we've got the following spindles setup
primary DB MDF file - 4 spindles, raid10
primary DB LDF file - 2 spindles, raid 1
tempdb MDF file - 2 spindles, raid 1
tempdb LDF file - 2 spindles, raid 1
backup staging area / scratch area - 2 spindles, raid 0
the reason for breaking spindles apart like this is to eliminate disk head
contention. that is the killer in disk performance - anytime the heads have
to seek you're talking milliseconds of latency. The access pattern of a
transaction log (LDF) is purely sequential appends. If a spindle were only
serving as space for a transaction log you could expect the head to do almost
no seeking - which means log writes (which have to be atomic) happen as fast
as possible.
The access to.MDF files (data pages, indexpages, etc) happen effectively
randomly. Thats nothing BUT seek traffic. Imagine the contention you'd have
for disk heads if you had random scattered reads competing with sequential
appends on the same spindle.
splitting your logs and data files is critical for this reason - disk io rate
is really helped along. Incidentally, caching raid controllers help basically
zero for this problem. You want the fastest rpm disks possible. For a SQL
machine we'd always choose smaller higher RPM drives than slower larger ones.
i also advocate spliting tempdb off onto its own set of spindles, as any query
thats awfully big sql server will have to use some tempdb space in its
calculations..presumably we're streaming pages off of the mdf and sticking
them in tempdb, tempdb had better not be contending for seek time with the
read source..
we write the SQL backup jobs to the raid0 and then immediately copy them else
where, where they are replicated to a few machines and also written out to
tape. I don't care if we lose the raid0 volume becase we'll just run the
backup jobs again - and we've probably already copied them off the box onto
our backup systems anyhow
an aside on backups - your backup is worth precisely nothing unless you've done
a blind restore, that is, take your.BAK file and restore it on some separate
machine. Then run "dbcc checkdb with physical" and do some sanity queries
to make sure that you're in good shape. More than one person has been
bitten by a backup plan that was happily backing up a db with a torn page.
When they finally figured out they had a torn page and went to restore, they
were SOL beacuse the backups were garbage. We have a.vbs script i cobbled
up that copies the most recent.bak file, writes out a sql script to restore
it, runs dbcc and a few other commands, and parses the output, searching for
known success messages. It then drops the restored copy and does it again,
this time applying all the transaction logs since the.bak was taken, and does
the same checks. Myself and the rest of the management team get the "yes
or no" mail about the backups vitality, as well as the job logs emailed to them
every morning.
Re: clustering
We don't use clustering because it complicates lots of things. Our take
is this - on this machine, if there is any kind of a failure, we're coming
into work to fix it and nobody else has a higher priority than getting the
service available. We have similar class hardware that we c
splinter cell for xbox outsold metroid prime in the same time period of about a month after they came out. brute force had more preorder sales than halo - indicating that there is more than just halo for the xbox, i'd say.
the Purple purse is done. sega has dropped sports title support for gamecube. luckily for them, nintendo's handheld francise is a license to print money. nintendo continues to miss the boat with their home machine offerings.
i look at it like this. if _everyone_ loves nintendo's franchises, and has the nostalgia factor making them less than objective about a GC purchase, then it seems all nintendo has to do to be wildly successful is give their fans what they want. I mean, if everyone wants a mario game and a zelda game and a metroid game, it seems like the purse and those three titles should be selling themselves, right ? those brands are as old as many gamers. so why isn't nintendo mopping the floor ?
what you infact see is a _nobody_ in the console world (microsoft) which has the complete distrust of lots of people (certainly slashdot, and a bunch of lawyers) edging out nintendo's CHEAPER system with their very first console release ever.
oh, the xbox has the highest attach rate of any current console. Given that, i don't see how you figure that game-to-system ratio is poor for xbox. Because its poorer for every other current-generation machine.
A guy down the hall from me was in charge of taking customer web apps written in V6 technologies (vb, asp, etc) and porting them in several ways to.net. They did extensive scalability testing on these apps. They measured requests/sec vs # of cpus, etc etc, to see how asp.net utilized multiproc machines.
What im saying here, is that you are not the first person to ever consider how.net might run for database driven web apps, of an arbitrary size. Infact, much of it is designed for _exactly_ that.
Re: SQL server 2000 SQL server 2000 has more performance then you know what to do with, even on non-ridiculous hardware. Give it processors with lots of L2 cache (xeons) and lots of ram, and read all the docs about keeping MDF and LDF files on separate volumes (as well as tempdb) and you'll find that life is thrilling.
Data point: On a quad HT P4 Xeon with 8GB of ram and 12 spindles (a significantly less than $50k box) we support 1800 simultaneous connections, doing OLTP work against a ~15GB database. The most commonly hit table in the system has about 10 million rows that get added and deleted in batches of between 20 and 10,000, and updated singly or in bulk. Other apps select from this table on a polling basis (i.e. decision monitors). We could make our db and app design much "better" w.r.t performance, but we don't need to - the money we save not having to do genius level feats of programming, app rewrites, and perf tuning more than pays for the occasional new hardware or upgrade.
Continuing, Run perf monitor on your SQL server machine. Look at the physical spindle(s) that hold your MDF. If you're reading from them, buy more ram until you're not:) Look at the I/O per sec rate to your tempdb disks and primary LDF disk(s). It is seriously to your advantage to go with an individual spindle for each role, because IO rate is what is so critically important to SQL server. Also, avoid RAID5 like the plague, as it decimates IO Rate.
You can tune SQL server without application changes until you're blue in the face, honestly. Use profiler to see what kind of queries you're doing. Put those queries in Query Analyzer and show the execution plan. QA breaks it down for you and shows execution time percentages of each sub-tree of the execution plan. If you've got something eating 80% of your time and its doing a table scan, do whatever you can to put some selectivity in that query (i.e. an index, or maybe a query change).
If you want to save yourself some headaches, setup management tasks to recalc indexes over the weekend (or nightly, if you see that much index fragmentation after a day).
if you read the mailing list id say roughly 50% of the messages on the subject say "make it work like outlook 2003" and the responses seem to be along the lines of "we'll consider that"
sure. you weren't going to bring cases primarily about people whining about "fairness" (what a ridiculous concept)
this is a key difference between democrats (who are essentially socialists) and republicans (who are essentially assholes)
me, id rather be an asshole. at least assholes don't go around making laws to reel in other assholes.
re: working on my dime; id happily pay the congress people the same salary to just stop working. i don't think i need any more laws. after we pay them to not pass new bad laws for a while, we can start shifting the focus to repealing existing bad laws. we should give bonuses to congresses that repeal the most number of laws.
ideally rather be libertarian, except they're a bunch of pot head do-nothings as near as i can tell:) (which is better than the greens, who are a bunch of pot head do-everything-wrongs.)
at this point im interested in the team thats going to pass the least restrictive laws that affect my daily life. thats currently the republicans. democrats really need to get over this whole self righteous "this is how i think it should be, lets pass a law to make it that way" attitude. ralph nader _especially_ needs to get over that attitude, as he has it worse then just about anyone.
the point is that people shouldn't have to deal with 80234 ways to get their machines patched. there should be ONE patch format. Also, the bar for what qualifies as a consumable patch is wildly different - some SQL server patches require you to stop sql server, copy files to places (!!!) and then run some.SQL files after you read the code and verify that you wont cook your data (!!!)
re: looking at linux - i (and many others) have:) i stopped running linux pretty early on, back when rpm still sucked completely. package managmeent back then was
tar xfvz slackware_package.tgz
i understand its much better now, and perhaps i'll give it another shot. solaris has a pretty livable package system, but the patch system is kind of ridiculous. even so, i'd say solaris has the upper hand compared to MS currently as far as ease of patch maintnance.
as to why we dont support a zero-footprint install, im not sure. There is some work in this area, and there's something called WinPE which is fantastic for install automation.. but i'd still say that many commercial unixes have us beat in install flexibility, openBSD probably being my favorite..
To compare RPM with MSI, consider the following: 1) can i determine which file(s) are used by which packages ? 2) can i verify that there are no incorrectly versioned files on my disk ? 3) can i apply packages/patches in a transacted nature ? 4) can i have per system and per-user package databases ? 5) can administrators of the directory service sign packages such that users with non-admin rights can install them in admin-rights-controlled places ? (i.e. my IT guy says its ok for me to have Office, so i get to install office even though im not an admin on the machine)
these are all things that MSI supports. im not saying they are difficult requirements to satisfy, and im also not saying that any of these reasons are why windows doesn't use MSI. For some technical reason, they just dont (im not an MSI or an OCM expert by any means)
I have a pretty strong linux, solaris, and irix background, and there are things i like and dislike about those operating systems installers and patch systems. Fwiw, i am not at all involved in installing or patching any microsoft product, so its irrelevant what experience i have with other platforms in terms of making MS stuff better. But, there are other people like me that know what "the competition" is doing, and feedback about our patching systems especially has made it high enough that you're seeing announcements like this one, which basically say "yeah, we could do better, and we plan on it"
That said, OS installation is something else that is being worked on for future windows products. Sometime leading up to the release of some future OS there will probably be some announcement about how it installs "better":)
First off, Windows isn't distributed as MSIs. So you cant (really) use MSI to patch windows. Windows uses another packaging/patching format called OCM.
MSI, for what its worth, was never really designed to do patches. It's a total hack. MSI has a mini database that works similalry to SQL but isn't (don't ask) that keeps track of objects in a few different tables. for instnace, you've got products, features, files, etc. A product has several features, and features use files. Multiple features/products can share files. Files can be versioned or unversioned.
Note that there isn't an intrinsic patch primitive in MSI. To "patch" an MSI, you update the feature/file defs in the MSI database. Then the MSI engine sees that the file system is out of sync with the database, and asks you to reinstall the feature. You supply a data source that has the files of the proper versioned in the new MSI database, and then the feature is "repaired", i.e. the filesystem is made to match the feature database.
MSI was designed to allow for repairing software that got broken or had files deleted, by knowing what was supposed to be on the disk and what is on the disk, and where there is feature overlap. Patching works by updating the data items in a table and then forcing a re-install.
If this all sounds ridiculous, it sort of is. When VS7 shipped the MSI was authored such that most of VS7 was a giant "feature". Based on what I've told you, you can see that to patch VS7 you'd need to "reinstall" it, which takes on the order of hours, not seconds. Initial attempts at patching VS7 were terrible. Therefore we created fake features that had file overlap with what vs7 files we wanted to patch and did a reinstall on those features only. I think we also do some work to clean up the feature definition in the MSI database itself for the VS7 feature(s) so that everything stays consistent.
Office has its own installer/patching system. Windows has its own installer/patch system. SQL doesn't really have a patch system to speak of.
You're right, its all very stupid and we're trying to fix it. Customers shouldn't have to put up with this crap. Thats why we're working on better patching systems.
Incidentally, i think the primary reason that OS and App patching are separate is that the OS packaging/patch system sort of works in a bootstrap environment, i.e. the full functionality you need to do good app patching isn't available at OS install time.
the executives of microsoft have donated more REAL, physical dollars to various causes around the world then you ever will.
Look at the Gates foundation sometime. You are a nobody in the world of philanthropy, comparatively, regardless of what pseudo-intellectual way you want to measure it.
When I do managed programming, i use VB.NET almost exclusively because its just easier to write than C#.
There's been one recent project that i did in C# because i was re-using some code that was already in C# and didn't feel indignant enough to wrap that code and call it from vb.net..
Also, Managed C++ lets you do anything you want, on essentially a per statement and per object basis. Power-wise, it crushes C#.
I think you're probably also confusing CLR and CLS, or even CLI (but thats pretty easy to do)
Let's say this - C# basically exists to be the "natural" language for the.NET platform, but there are gobs of reasons to NOT use c# in.NET programs. An obvious move is to use MC++ to expose existing c/c++ code to managed applications in a precislely designed and controlled fashion, or to leverage the execution speed of native C code with the benefits of managed conventions.
I already made my case for why i love VB.NET (you go off and make fun of me for not being a "real" programming, meanwhile i'll be beating deadlines and delivering useful functionality)
Also, people with more factual content than you or I have presented have already commented on the "language adaptability" of MSIL and the.NET VM vs the JVM and the consensus i've gathered is that.net's does have some non-fluff advantages.
Re:Lots of good info here...
on
I, Spammer
·
· Score: 1
And blowing people up might be taking things a little far;)
I don't think it would be. You can only ask somebody to stop doing something to you so many times before its clear they don't respect your wishes. At that point, what's left ?
What i have a problem with is vigilante acts, because for each cause, there's a vigilante for and against it. There are many amongst the slashdot crowd that probably think it's ok to bomb my place of employment. Obviously i disagree, the families of thousands of my co-workers disagree, etc. In that sense, im definitely glad that vigilante's don't run the physical world.
Mob rule is a bad thing, but I didn't take enough political science to understand how mob rule gains legitimacy and becomes a representational democratic government (although pedants amongst you will remind me this is a 'republic' we live in)
If the federal government decided that spam was illegal, and made it perfectly clear what the discovery and prosecution phase would be for those continuing to spam, and then started gunning down continued-spammers in their homes, i'd be ok with it.
Think of it as a DEA except for spammers.
Incidentally, i think the DEA has probably outlived its usefulness because it seems that there is a sizeable portion of our population that _wants_ drugs to be LEGAL, not just the peddlars. However, i cant think of ANYONE that says they want to get spammed. AFAIK, nobody gets drugs pushed on them (in a way thats not already criminal), but _everyone_ is victimized by spam. Ergo, a DEA type agency to take out spammers in meat-space would get 120% backing from me.
its safer then you think. House-supply gas out here is run at an extremely low pressure. THe local gas company trains the fire fighters to put out active gas line fires. It's pretty tame apparently - they take a 12" pipe running at way higher PSI than the residential lines run, then they cut a giant hole in it, light it on fire, and then have fire fighters go put it out.
The gas supply in your neighborhood is nothing, comparatively. Additionally, the gas concentration in the atmosphere required to be combustible rather than flammable is such that you would be choking to death before it would have a chance of exploding.
I'm not entirely disagreeing with you - i think environmentalists are generally alarmists fools, and im 100% in favor of nuke plants. I just wanted to point out that NG is a great residential energy application. I've personally done the additional piping work in my home to put in a gas dryer, tankless hot water heater, and gas stove. I can't tell you how nice infinitely long showers are. Yes, there are electric instant water heaters, but the flow -rate is usually substandard and more over the basic model typically takes 3 40amp 240v circuits - 60% the load capacity of a modern home electrical service.
If the task is "make something hot", gas usually does a much better job than electricity. Also, your wife will seriously thank you if you get a gas clothes dryer:)
IBM isn't doing MS any favors as it stands. IBM is Microsoft's biggest competitor.
IBM is pushing the ludicous Linux/z-series to anyone that will listen. They're using linux as a way to make a KILLING on consulting hours (because IBM knows damn well that nobody in the real world can do anything with linux). Once the consulting people (you did realize that IBM Global Services is the biggest revenue generator for all of IBM, right ?) are in its easy to upsell them on websphere, DB/2, or whatever.
Later, when its clear that linux isn't up to the task they move them onto RS6k, AS/400, or zOS stuff, where microsoft has no chance of regaining a foothold because of platform issues.
IBM is the only reason traditional businesses take linux seriously, and using linux locks businesses into a framework of IBM software stacks and consulting hours. Linux costs them nothing to "resell" but has huge consulting contracts attached to it. It's tailor made for how they want to do business.
if your "music" is comprised entirely of time shifting/compressing a recording of someone else's work, while you "add value" by speaking profanely over it......maybe you should consider not checking "musician" on your census..
eh, when i click on those ads, i get a new window, the size of my screen (but not taking up my screen, its just some other window), with a little blurb in the upper left corner in blue text that says
yeah, nobody would ever objectively choose a platform that anybody could install, can be set to patch itself, and just works with all the programs they already use. What a dumb idea that would be!
Here's a sample size of one: i used linux. Unless you were running kernel 98 (thats ZERO point 98), I used it before you did. I've written two (trivial) unix-only software packages, and i've got a one line bugfix in the openbsd IDE driver. I've sent bug reports to a few differnet open source projects, and I had Alan cox personally tell me that my VM starvation issue that was crashing my production linux webserver would be fixed in the next release of "buy more ram and leave me alone". I was writing multi-thread programs on unix back before linux had a respectable pthreads implemention. I was making my living on unix machines back when i was still telling people i'd quit if some employer asked me to use an NT machine. Suffice it to say, I've had more than enough exposure to linux, and unix in general.
Now, I choose Windows. Not because I don't know about anything else. Because to me it just wasn't worth fighting with *nix any more to make it do what i wanted to do.. why waste my time if Windows already does it ?
There are lots of companies that try lots of things and decide on windows. If thats hard for you to beleive you should try talking to some of those companies. Be sure to leave your blinders at home.
here's what m-w.com has to say: 1 : the introduction of something new 2 : a new idea, method, or device : NOVELTY
apparently to use, innovation must be a brand new product category that has never been thought of before.
well gosh, when microsoft comes out with "MS Velvet Toy Unicorn Stuffed with BirdShit and a Large Green Salamander Tatoo", you better be first in line to praise Microsoft for coming up with a brand new type of thing that nobody else has done before (or do you know of some prior company working on shit-stuffed toys with tatoos ?")
Alternatively, you could make the argument that innovation is taking an idea that was poorly executed on, analyzing its shortcomings, fixing them, and really delivering on the possibilities.
Let's take the Xbox Complaint: Developing for custom processors with shitty dev tools is hard; lack of ram makes games look like Shit (PS2) Solution: Use industry standard hardware, use volume to keep costs low, use industry standard programming tools and libraries; add a boatload of ram Result: XBox is one of the easiest platforms to develop for, one of the easiest platforms to port to, and visually and sonically crushes PS2. I'd call that innovation.
Let's take Excel. Complaint: Lotus 1-2-3 is slow, doesn't take advantage of any kind of graphical environment, and is difficult to use Solution: Make a spreadsheet that takes advantages of the new Windows graphical environment. Work solidly on making it perform fantastically, so that instead of letting macros run OVERNIGHT they can run significantly faster. Make writing formulas easier so that more people can use it. Do away withthe idea of "spread sheet programmer" Result: Lotus who ?
Consider Windows NT (which, first off, your comment about it being os2 is bullshit. MS was working on OS2 for IBM, not the other way around.) Complaint: OS/2 was hard to setup and install, had no applications, had hellacious system requirements, had poor hardware support. Getting users to adopt it was impossible. Home users couldn't enjoy the benefits of modern computing that a modern OS could provide. Solution: WNT introduced as a server only solution. Initial adoption rate extremely low, but core tenets of high reliability and integrated security architecture and SMP support are there. Prove the architecture, evolve the platform slowly, provide a Win16 and Win32 execution environment out of the box to run existing apps. As the platform matures and stabilizes, grow the installed base, add support for more commodity hardware, do more testing with applications starting at the high end and working towards the end-user. Continue to refactor and add functionality until you've got a viable desktop platform built off of enterprise quality internals. Result: Server class features/operating environment brought to the average home desktop. NT 4 Workstation, W2k Pro, and XP Personal are the first desktop operating systems built off the guts of a server-class system. Mac OS X is the only other contender. Arguably, Linux is not currently a desktop contender, and DEFINITELY was not when NT4W and W2k Pro were released.
If you want to go innovate some shit-filled stuffed animals and beat MS to the next big innovation, be my guest.
I'm sure you've seen the internals of the windows 2k3 code and the windows 1.0 code, thusly qualifying you to make that statement ?
I also assume you're qualified to comment on the internals of a unix kernel, linux specifically. Please comment on the major changes between UTSSE (7th edition) and Linux, as i see that minix and then UTSSE are linux's predecessors.
Slashdot Mirror
essenially every benchmark ever disagrees with you.
.NET runtime is pretty fast for a garbagecollected JITted environment. Certainly faster than any known JVM for most things.
.NET. Two versions of it have shipped now (runtime 1.0 and 1.1) and there are some huge internet properties that are using it.
IIS is incredibly fast. SQL server is constantly bouncing in and out of 1st place for tpm's, both absolute and cost adjusted. How is that slow ?
re: IIS security issues - you realize that the majority of IIS holes have to do with external components accessible via IIS, and now ship in a turned-off-by-default state ?
I'm not sure what you think is unproven about
oh, you mean the meaningful replication that took down orbitz for a day or two ? :)
if you have a reproducible environment where sql server 2000 is losing or corrupting data, please email me directly with details.
.NET. I suppose i can't change your mind and i definitely cant change your experiences, but i am highly skeptical that you're not using some rose colored lenses when you're examing how your app stack compares.. Windows + ASP.NET + SQL server works pretty well out of the box. I don't want to be too presumptuous here, but unless you've had work experience with one of a handful of simply huge internet properties, i can say that Windows + ASP.NET + SQL server is reliably running many sites bigger than you've ever dealt with. In other words, some very big places seem to think its just fine :)
having said that, i think you are way off base. I used mySQL a lot and use SQL server extensively now. Comparing the two just isn't worth while. mySQL is a lot closer to berkeley DB than it is to SQL server, feature wise, scalability wise, management wise, and reliability wise.
SQL server has several licsensing options - per conection or per processor are the two default ones. Given that SQL server is perf competitive with Oracle, (it infact beats it on official benchmarks) and a HECK of a lot easier to deal with, SQL server is a STEAL compared to oracle pricing and support.
for most OLTP type work mySQL will NOT be faster than SQL server, especially as the volume of data grows, the complexity of queries grows, and the contention increases.
Put another way - when i see mySQL displacing SQL Server in the tpc benchmarks, i'll pay attention to you again. mySQL is very cool, but its a toy compared to DB2, Oracle, or SQL server. If all you need is a toy, by all means, use a toy -- you'll be happier. When you're ready to graduate to real systems, SQL server will be waiting for you. Maybe mySQL or Postgres will evolve faster than your needs do, and you'll never need a commercial quality database (or one of those two will turn into one)
It is unfortuneate that your experience has shown that Java+Jboss+{some db here, as you named like 5 of them}+Linux is "more reliable" than Windows +
Yes. We're using Raid 1 and Raid 10. Each disk is 36gb so realistically we could probably get away with only raid 1s, as none of our individual files is larger then 15gb or so. But we've got the following spindles setup
the reason for breaking spindles apart like this is to eliminate disk head contention. that is the killer in disk performance - anytime the heads have to seek you're talking milliseconds of latency. The access pattern of a transaction log (LDF) is purely sequential appends. If a spindle were only serving as space for a transaction log you could expect the head to do almost no seeking - which means log writes (which have to be atomic) happen as fast as possible.
The access to .MDF files (data pages, indexpages, etc) happen effectively
randomly. Thats nothing BUT seek traffic. Imagine the contention you'd have
for disk heads if you had random scattered reads competing with sequential
appends on the same spindle.
splitting your logs and data files is critical for this reason - disk io rate is really helped along. Incidentally, caching raid controllers help basically zero for this problem. You want the fastest rpm disks possible. For a SQL machine we'd always choose smaller higher RPM drives than slower larger ones.
i also advocate spliting tempdb off onto its own set of spindles, as any query thats awfully big sql server will have to use some tempdb space in its calculations..presumably we're streaming pages off of the mdf and sticking them in tempdb, tempdb had better not be contending for seek time with the read source..
we write the SQL backup jobs to the raid0 and then immediately copy them else where, where they are replicated to a few machines and also written out to tape. I don't care if we lose the raid0 volume becase we'll just run the backup jobs again - and we've probably already copied them off the box onto our backup systems anyhow
an aside on backups - your backup is worth precisely nothing unless you've done a blind restore, that is, take your.BAK file and restore it on some separate machine. Then run "dbcc checkdb with physical" and do some sanity queries to make sure that you're in good shape. More than one person has been bitten by a backup plan that was happily backing up a db with a torn page. When they finally figured out they had a torn page and went to restore, they were SOL beacuse the backups were garbage. We have a .vbs script i cobbled
up that copies the most recent .bak file, writes out a sql script to restore
it, runs dbcc and a few other commands, and parses the output, searching for
known success messages. It then drops the restored copy and does it again,
this time applying all the transaction logs since the .bak was taken, and does
the same checks. Myself and the rest of the management team get the "yes
or no" mail about the backups vitality, as well as the job logs emailed to them
every morning.
Re: clustering We don't use clustering because it complicates lots of things. Our take is this - on this machine, if there is any kind of a failure, we're coming into work to fix it and nobody else has a higher priority than getting the service available. We have similar class hardware that we c
splinter cell for xbox outsold metroid prime in the same time period of about a month after they came out. brute force had more preorder sales than halo - indicating that there is more than just halo for the xbox, i'd say.
the Purple purse is done. sega has dropped sports title support for gamecube. luckily for them, nintendo's handheld francise is a license to print money. nintendo continues to miss the boat with their home machine offerings.
i look at it like this. if _everyone_ loves nintendo's franchises, and has the nostalgia factor making them less than objective about a GC purchase, then it seems all nintendo has to do to be wildly successful is give their fans what they want. I mean, if everyone wants a mario game and a zelda game and a metroid game, it seems like the purse and those three titles should be selling themselves, right ? those brands are as old as many gamers. so why isn't nintendo mopping the floor ?
what you infact see is a _nobody_ in the console world (microsoft) which has the complete distrust of lots of people (certainly slashdot, and a bunch of lawyers) edging out nintendo's CHEAPER system with their very first console release ever.
oh, the xbox has the highest attach rate of any current console. Given that, i don't see how you figure that game-to-system ratio is poor for xbox. Because its poorer for every other current-generation machine.
A guy down the hall from me was in charge of taking customer web apps written in V6 technologies (vb, asp, etc) and porting them in several ways to .net. They did extensive scalability testing on these apps. They measured requests/sec vs # of cpus, etc etc, to see how asp.net utilized multiproc machines.
.net might run for database driven web apps, of an arbitrary size. Infact, much of it is designed for _exactly_ that.
:) Look at the I/O per sec rate to your tempdb disks and primary LDF disk(s). It is seriously to your advantage to go with an individual spindle for each role, because IO rate is what is so critically important to SQL server. Also, avoid RAID5 like the plague, as it decimates IO Rate.
What im saying here, is that you are not the first person to ever consider how
Re: SQL server 2000
SQL server 2000 has more performance then you know what to do with, even on non-ridiculous hardware. Give it processors with lots of L2 cache (xeons) and lots of ram, and read all the docs about keeping MDF and LDF files on separate volumes (as well as tempdb) and you'll find that life is thrilling.
Data point: On a quad HT P4 Xeon with 8GB of ram and 12 spindles (a significantly less than $50k box) we support 1800 simultaneous connections, doing OLTP work against a ~15GB database. The most commonly hit table in the system has about 10 million rows that get added and deleted in batches of between 20 and 10,000, and updated singly or in bulk. Other apps select from this table on a polling basis (i.e. decision monitors). We could make our db and app design much "better" w.r.t performance, but we don't need to - the money we save not having to do genius level feats of programming, app rewrites, and perf tuning more than pays for the occasional new hardware or upgrade.
Continuing, Run perf monitor on your SQL server machine. Look at the physical spindle(s) that hold your MDF. If you're reading from them, buy more ram until you're not
You can tune SQL server without application changes until you're blue in the face, honestly. Use profiler to see what kind of queries you're doing. Put those queries in Query Analyzer and show the execution plan. QA breaks it down for you and shows execution time percentages of each sub-tree of the execution plan. If you've got something eating 80% of your time and its doing a table scan, do whatever you can to put some selectivity in that query (i.e. an index, or maybe a query change).
If you want to save yourself some headaches, setup management tasks to recalc indexes over the weekend (or nightly, if you see that much index fragmentation after a day).
if you read the mailing list id say roughly 50% of the messages on the subject say "make it work like outlook 2003" and the responses seem to be along the lines of "we'll consider that"
innovative indeed.
1. are you a tool ? HT results in a 30% speed increase for typical workloads. turning it off has a HUGE effect on a P4 HT proc's throughput.
Incidentally, what you propose is roughly what SPEC is. the details of the submissions are clearly disclosed
sure. you weren't going to bring cases primarily about people whining about "fairness" (what a ridiculous concept)
:) (which is better than the greens, who are a bunch of pot head do-everything-wrongs.)
this is a key difference between democrats (who are essentially socialists) and republicans (who are essentially assholes)
me, id rather be an asshole. at least assholes don't go around making laws to reel in other assholes.
re: working on my dime; id happily pay the congress people the same salary to just stop working. i don't think i need any more laws. after we pay them to not pass new bad laws for a while, we can start shifting the focus to repealing existing bad laws. we should give bonuses to congresses that repeal the most number of laws.
ideally rather be libertarian, except they're a bunch of pot head do-nothings as near as i can tell
at this point im interested in the team thats going to pass the least restrictive laws that affect my daily life. thats currently the republicans. democrats really need to get over this whole self righteous "this is how i think it should be, lets pass a law to make it that way" attitude. ralph nader _especially_ needs to get over that attitude, as he has it worse then just about anyone.
the goal of this isn't to affect patch quality.
.SQL files after you read the code and verify that you wont cook your data (!!!)
the point is that people shouldn't have to deal with 80234 ways to get their machines patched. there should be ONE patch format. Also, the bar for what qualifies as a consumable patch is wildly different - some SQL server patches require you to stop sql server, copy files to places (!!!) and then run some
The goal is to make all that better.
re: looking at linux - i (and many others) have :) i stopped running linux pretty early on, back when rpm still sucked completely. package managmeent back then was
:)
tar xfvz slackware_package.tgz
i understand its much better now, and perhaps i'll give it another shot. solaris has a pretty livable package system, but the patch system is kind of ridiculous. even so, i'd say solaris has the upper hand compared to MS currently as far as ease of patch maintnance.
as to why we dont support a zero-footprint install, im not sure. There is some work in this area, and there's something called WinPE which is fantastic for install automation.. but i'd still say that many commercial unixes have us beat in install flexibility, openBSD probably being my favorite..
To compare RPM with MSI, consider the following:
1) can i determine which file(s) are used by which packages ?
2) can i verify that there are no incorrectly versioned files on my disk ?
3) can i apply packages/patches in a transacted nature ?
4) can i have per system and per-user package databases ?
5) can administrators of the directory service sign packages such that users with non-admin rights can install them in admin-rights-controlled places ? (i.e. my IT guy says its ok for me to have Office, so i get to install office even though im not an admin on the machine)
these are all things that MSI supports. im not saying they are difficult requirements to satisfy, and im also not saying that any of these reasons are why windows doesn't use MSI. For some technical reason, they just dont (im not an MSI or an OCM expert by any means)
I have a pretty strong linux, solaris, and irix background, and there are things i like and dislike about those operating systems installers and patch systems. Fwiw, i am not at all involved in installing or patching any microsoft product, so its irrelevant what experience i have with other platforms in terms of making MS stuff better. But, there are other people like me that know what "the competition" is doing, and feedback about our patching systems especially has made it high enough that you're seeing announcements like this one, which basically say "yeah, we could do better, and we plan on it"
That said, OS installation is something else that is being worked on for future windows products. Sometime leading up to the release of some future OS there will probably be some announcement about how it installs "better"
yeah, how did that work out for AOL last time ? :)
what did they buy netscape for ? how much is AOL/TW/Netscape worth now ? What settlement did they get ?
Good question.
First off, Windows isn't distributed as MSIs. So you cant (really) use MSI to patch windows. Windows uses another packaging/patching format called OCM.
MSI, for what its worth, was never really designed to do patches. It's a total hack. MSI has a mini database that works similalry to SQL but isn't (don't ask) that keeps track of objects in a few different tables. for instnace, you've got products, features, files, etc. A product has several features, and features use files. Multiple features/products can share files. Files can be versioned or unversioned.
Note that there isn't an intrinsic patch primitive in MSI. To "patch" an MSI, you update the feature/file defs in the MSI database. Then the MSI engine sees that the file system is out of sync with the database, and asks you to reinstall the feature. You supply a data source that has the files of the proper versioned in the new MSI database, and then the feature is "repaired", i.e. the filesystem is made to match the feature database.
MSI was designed to allow for repairing software that got broken or had files deleted, by knowing what was supposed to be on the disk and what is on the disk, and where there is feature overlap. Patching works by updating the data items in a table and then forcing a re-install.
If this all sounds ridiculous, it sort of is. When VS7 shipped the MSI was authored such that most of VS7 was a giant "feature". Based on what I've told you, you can see that to patch VS7 you'd need to "reinstall" it, which takes on the order of hours, not seconds. Initial attempts at patching VS7 were terrible. Therefore we created fake features that had file overlap with what vs7 files we wanted to patch and did a reinstall on those features only. I think we also do some work to clean up the feature definition in the MSI database itself for the VS7 feature(s) so that everything stays consistent.
Office has its own installer/patching system. Windows has its own installer/patch system. SQL doesn't really have a patch system to speak of.
You're right, its all very stupid and we're trying to fix it. Customers shouldn't have to put up with this crap. Thats why we're working on better patching systems.
Incidentally, i think the primary reason that OS and App patching are separate is that the OS packaging/patch system sort of works in a bootstrap environment, i.e. the full functionality you need to do good app patching isn't available at OS install time.
if you want to do it with forced induction, you probalby just want a MAP sensor on the intake manifold.
Generally, MAP > MAF > AFM (flapper)
they do.
the executives of microsoft have donated more REAL, physical dollars to various causes around the world then you ever will.
Look at the Gates foundation sometime. You are a nobody in the world of philanthropy, comparatively, regardless of what pseudo-intellectual way you want to measure it.
thats nonsense.
.NET platform, but there are gobs of reasons to NOT use c# in .NET programs. An obvious move is to use MC++ to expose existing c/c++ code to managed applications in a precislely designed and controlled fashion, or to leverage the execution speed of native C code with the benefits of managed conventions.
.NET VM vs the JVM and the consensus i've gathered is that .net's does have some non-fluff advantages.
When I do managed programming, i use VB.NET almost exclusively because its just easier to write than C#.
There's been one recent project that i did in C# because i was re-using some code that was already in C# and didn't feel indignant enough to wrap that code and call it from vb.net..
Also, Managed C++ lets you do anything you want, on essentially a per statement and per object basis. Power-wise, it crushes C#.
I think you're probably also confusing CLR and CLS, or even CLI (but thats pretty easy to do)
Let's say this - C# basically exists to be the "natural" language for the
I already made my case for why i love VB.NET (you go off and make fun of me for not being a "real" programming, meanwhile i'll be beating deadlines and delivering useful functionality)
Also, people with more factual content than you or I have presented have already commented on the "language adaptability" of MSIL and the
And blowing people up might be taking things a little far ;)
I don't think it would be. You can only ask somebody to stop doing something to you so many times before its clear they don't respect your wishes. At that point, what's left ?
What i have a problem with is vigilante acts, because for each cause, there's a vigilante for and against it. There are many amongst the slashdot crowd that probably think it's ok to bomb my place of employment. Obviously i disagree, the families of thousands of my co-workers disagree, etc. In that sense, im definitely glad that vigilante's don't run the physical world.
Mob rule is a bad thing, but I didn't take enough political science to understand how mob rule gains legitimacy and becomes a representational democratic government (although pedants amongst you will remind me this is a 'republic' we live in)
If the federal government decided that spam was illegal, and made it perfectly clear what the discovery and prosecution phase would be for those continuing to spam, and then started gunning down continued-spammers in their homes, i'd be ok with it.
Think of it as a DEA except for spammers.
Incidentally, i think the DEA has probably outlived its usefulness because it seems that there is a sizeable portion of our population that _wants_ drugs to be LEGAL, not just the peddlars. However, i cant think of ANYONE that says they want to get spammed. AFAIK, nobody gets drugs pushed on them (in a way thats not already criminal), but _everyone_ is victimized by spam. Ergo, a DEA type agency to take out spammers in meat-space would get 120% backing from me.
Violence is the last resort of the patient.
its safer then you think. House-supply gas out here is run at an extremely low pressure. THe local gas company trains the fire fighters to put out active gas line fires. It's pretty tame apparently - they take a 12" pipe running at way higher PSI than the residential lines run, then they cut a giant hole in it, light it on fire, and then have fire fighters go put it out.
:)
The gas supply in your neighborhood is nothing, comparatively. Additionally, the gas concentration in the atmosphere required to be combustible rather than flammable is such that you would be choking to death before it would have a chance of exploding.
I'm not entirely disagreeing with you - i think environmentalists are generally alarmists fools, and im 100% in favor of nuke plants. I just wanted to point out that NG is a great residential energy application. I've personally done the additional piping work in my home to put in a gas dryer, tankless hot water heater, and gas stove. I can't tell you how nice infinitely long showers are. Yes, there are electric instant water heaters, but the flow -rate is usually substandard and more over the basic model typically takes 3 40amp 240v circuits - 60% the load capacity of a modern home electrical service.
If the task is "make something hot", gas usually does a much better job than electricity.
Also, your wife will seriously thank you if you get a gas clothes dryer
IBM isn't doing MS any favors as it stands. IBM is Microsoft's biggest competitor.
IBM is pushing the ludicous Linux/z-series to anyone that will listen. They're using linux as a way to make a KILLING on consulting hours (because IBM knows damn well that nobody in the real world can do anything with linux). Once the consulting people (you did realize that IBM Global Services is the biggest revenue generator for all of IBM, right ?) are in its easy to upsell them on websphere, DB/2, or whatever.
Later, when its clear that linux isn't up to the task they move them onto RS6k, AS/400, or zOS stuff, where microsoft has no chance of regaining a foothold because of platform issues.
IBM is the only reason traditional businesses take linux seriously, and using linux locks businesses into a framework of IBM software stacks and consulting hours. Linux costs them nothing to "resell" but has huge consulting contracts attached to it. It's tailor made for how they want to do business.
if your "music" is comprised entirely of time shifting/compressing a recording of someone else's work, while you "add value" by speaking profanely over it... ...maybe you should consider not checking "musician" on your census..
Everyone on the "we finally get it" bus is waving goodbye to you.
eh, when i click on those ads, i get a new window, the size of my screen (but not taking up my screen, its just some other window), with a little blurb in the upper left corner in blue text that says
[flash]
I'm using IE6SP1. Thanks proxomitron!
yeah, nobody would ever objectively choose a platform that anybody could install, can be set to patch itself, and just works with all the programs they already use. What a dumb idea that would be!
Here's a sample size of one: i used linux. Unless you were running kernel 98 (thats ZERO point 98), I used it before you did. I've written two (trivial) unix-only software packages, and i've got a one line bugfix in the openbsd IDE driver. I've sent bug reports to a few differnet open source projects, and I had Alan cox personally tell me that my VM starvation issue that was crashing my production linux webserver would be fixed in the next release of "buy more ram and leave me alone". I was writing multi-thread programs on unix back before linux had a respectable pthreads implemention. I was making my living on unix machines back when i was still telling people i'd quit if some employer asked me to use an NT machine. Suffice it to say, I've had more than enough exposure to linux, and unix in general.
Now, I choose Windows. Not because I don't know about anything else. Because to me it just wasn't worth fighting with *nix any more to make it do what i wanted to do.. why waste my time if Windows already does it ?
There are lots of companies that try lots of things and decide on windows. If thats hard for you to beleive you should try talking to some of those companies. Be sure to leave your blinders at home.
you must have an interesting idea of innovation.
here's what m-w.com has to say:
1 : the introduction of something new
2 : a new idea, method, or device : NOVELTY
apparently to use, innovation must be a brand new product category that has never been thought of before.
well gosh, when microsoft comes out with "MS Velvet Toy Unicorn Stuffed with BirdShit and a Large Green Salamander Tatoo", you better be first in line to praise Microsoft for coming up with a brand new type of thing that nobody else has done before (or do you know of some prior company working on shit-stuffed toys with tatoos ?")
Alternatively, you could make the argument that innovation is taking an idea that was poorly executed on, analyzing its shortcomings, fixing them, and really delivering on the possibilities.
Let's take the Xbox
Complaint: Developing for custom processors with shitty dev tools is hard; lack of ram makes games look like Shit (PS2)
Solution: Use industry standard hardware, use volume to keep costs low, use industry standard programming tools and libraries; add a boatload of ram
Result: XBox is one of the easiest platforms to develop for, one of the easiest platforms to port to, and visually and sonically crushes PS2. I'd call that innovation.
Let's take Excel.
Complaint: Lotus 1-2-3 is slow, doesn't take advantage of any kind of graphical environment, and is difficult to use
Solution: Make a spreadsheet that takes advantages of the new Windows graphical environment. Work solidly on making it perform fantastically, so that instead of letting macros run OVERNIGHT they can run significantly faster. Make writing formulas easier so that more people can use it. Do away withthe idea of "spread sheet programmer"
Result: Lotus who ?
Consider Windows NT
(which, first off, your comment about it being os2 is bullshit. MS was working on OS2 for IBM, not the other way around.)
Complaint: OS/2 was hard to setup and install, had no applications, had hellacious system requirements, had poor hardware support. Getting users to adopt it was impossible. Home users couldn't enjoy the benefits of modern computing that a modern OS could provide.
Solution: WNT introduced as a server only solution. Initial adoption rate extremely low, but core tenets of high reliability and integrated security architecture and SMP support are there. Prove the architecture, evolve the platform slowly, provide a Win16 and Win32 execution environment out of the box to run existing apps. As the platform matures and stabilizes, grow the installed base, add support for more commodity hardware, do more testing with applications starting at the high end and working towards the end-user. Continue to refactor and add functionality until you've got a viable desktop platform built off of enterprise quality internals.
Result: Server class features/operating environment brought to the average home desktop. NT 4 Workstation, W2k Pro, and XP Personal are the first desktop operating systems built off the guts of a server-class system. Mac OS X is the only other contender. Arguably, Linux is not currently a desktop contender, and DEFINITELY was not when NT4W and W2k Pro were released.
If you want to go innovate some shit-filled stuffed animals and beat MS to the next big innovation, be my guest.
I'm sure you've seen the internals of the windows 2k3 code and the windows 1.0 code, thusly qualifying you to make that statement ?
I also assume you're qualified to comment on the internals of a unix kernel, linux specifically. Please comment on the major changes between UTSSE (7th edition) and Linux, as i see that minix and then UTSSE are linux's predecessors.
I'll be waiting for your report.