Re: HARDWARE FOR J2EE apps

2000-10-13 Thread Michael Rimov

At 06:53 PM 10/12/2000 +0200, you wrote:
Hi Kevin,

Do you create the database in IBConsole ? Because the ibconsole has an bug 
when
creating databases.
If so, you probably have to get the newer version of IBConsole :
URL : http://www.interbase.com/open/downloads/ib_download.html
goto the miscellaneous section (at the bottumn of the page) and download
IBConsole - build 308. You must overwrite the ibconsole in your interbase
install directory. If you want the interbase-database-schema.xml (for 
orion cmp
managed persistence) mail me and i will sent it to you.

There's also a newer build somewhere on http://community.borland.com/.  Its 
where the author is posting all his updates.

Further more i think interbase 6 is an very good database but I would like to
know
if it handle a large amount of connections at the same time (like 500 or
something).
There isn't much info about that in the docs.

I haven't scaled to that level, but you might want to post to the inprise 
newsgroups on that, and I bet you'll get a response from someone.

HTH,
 -Mike








RE: HARDWARE FOR J2EE apps

2000-10-12 Thread Robert Krueger


I would reconsider implementing such low-level functionality. the deeper 
you dig into it, the more little things you will discover that have the 
potential to break your neck. It's probably not that hard to this stuff up 
and running. the problem is to get it really stable. we're currently 
planning on the following setup.

2 rather powerful dual processor machines, that run orion and interbase 6 
on linux with heartbeat for ip-address takeover. one is running as a hot 
standby and they are syncronized using the IBReplicator product that was 
already mentioned in this thread (http://www.synectics.co.za/). the 
performance of the syncronization that's described in the docs is enough 
for our application to believe it will basically be real-time (something 
like 200 updates per second over a 10Mbit link and pentium 200s).

of course this only takes care of availability and doesn't do any load 
balancing but for us the performance we can achieve by using good caching 
algorithms at the application level and fast dual processor machines is 
enough. for a really scalable solution I'd probably look at sybase (the 
commercial versions, the free version doesn't support clustering). of 
course, if the database is not the bottleneck one could use the standby 
solution for the database only and put a real cluster of orion servers in 
front.

robert

At 23:37 11.10.00 , you wrote:
Hi!

On Wed, 11 Oct 2000, Robert Krueger wrote:

[description of budget-friendly Orion setup - snip]

  sounds very nice but what about the database? how do you cluster that
  without spending an arm and a leg? our experience is, that it's not that
  hard to set up clustered web services with static pages and servlets but
  the really expensive part is, when you want that high availability for 
 your
  database. it doesn't buy you much if you have highly available ejbs when
  the database server goes down. many people use clustered apache/jserv on
  linux and cheap pc-hardware for high volume transactional websites but 
 have
  a large enterprise sun running oracle in the back. anyone out there 
 running
  a configuration with orion that includes a database with failover that
  doesn't blow up the budget too much (compared to other components)?

Ah. Deja Vu. :-) I'm involved with a startup company here in Sweden,
and considering Orion as app server we are also contemplating what
non-arm-and-a-leg-spending ways there are to enable a database for
failover and load-balancing functionality.

We are looking at either PostgreSQL, InterBase, SAP-DB or MySQL (yeah,
I know) as our database. We have not seen any currently freely available
solutions for failover for either of those (anybody?). So, we are thinking
about implementing something by ourselves. Nothing definite so far, but:

- The synchronizing of data will be done on the application-level, not
   by the database servers themselves. See below.
- We'll avoid numeric sequences for record keys to make this easier.
   We will implement some unique key-generation scheme based on whatever
   is needed to make keys unique but still not rely on strict monotone
   numeric sequences or similar ( md5(table_name+timestamp()), perhaps? ).
- We'll code a DB-abstraction layer that takes care of executing
   all update queries against all configured database servers and
   all read queries against one of them known to be alive and lightly
   loaded (or not recently accessed, or some other scheme).
- I guess that if we need database-specific stuff such as stored
   procedures or similar we need to use the same database software
   for all failover machines.
- If we stay away from database-specifics we could possibly allow
   failover between different database products. Would be cool.
   Using straight, standard SQL, could make this feasible.

These are very premature thoughts, we are getting closer to the
planning and design stage, but we haven't actually started yet.
Any thoughts? Ridiculously naive? Or possible to pull off?

  robert

/David

Reach me by
- Email: [EMAIL PROTECTED], WWW: http://www.dtek.chalmers.se/~d0dak/ -
--/David---


(-) Robert Krüger
(-) SIGNAL 7 Gesellschaft für Informationstechnologie mbH
(-) Brüder-Knauß-Str. 79 - 64285 Darmstadt,
(-) Tel: 06151 665401, Fax: 06151 665373
(-) [EMAIL PROTECTED], www.signal7.de





RE: HARDWARE FOR J2EE apps

2000-10-12 Thread Frank Eggink

I was wondering whether it would be possible to let the orion server(s) switch 
datasource as a sort of
failover procedure. I haven't digged myself into clustering and failover, but here's 
the idea.

Assuming you have made sure the server are running with failover and you have bought 
yourself a SCSI-raid
configuration with a Y connection (that is the disk are connected to two systems), the 
largest risk of failure
is the database server. If however orion could automatically switch  on some sort of 
connection timeout from
one server to another, you can solve this problem almost to easy.

This stil leaves you with a number of risks like:
- write failure of the db-software corrupting your database.
- program logic failures.
- site disaster (fire, flooding, tornado etc.).
- human error.

Depending on the sort system I guess, I could live with that.


Frank

On Wednesday, October 11, 2000 11:37 PM, David Kinnvall [SMTP:[EMAIL PROTECTED]] 
wrote:
 Hi!
 
 On Wed, 11 Oct 2000, Robert Krueger wrote:
 
 [description of budget-friendly Orion setup - snip]
 
  sounds very nice but what about the database? how do you cluster that 
  without spending an arm and a leg? our experience is, that it's not that 
  hard to set up clustered web services with static pages and servlets but 
  the really expensive part is, when you want that high availability for your 
  database. it doesn't buy you much if you have highly available ejbs when 
  the database server goes down. many people use clustered apache/jserv on 
  linux and cheap pc-hardware for high volume transactional websites but have 
  a large enterprise sun running oracle in the back. anyone out there running 
  a configuration with orion that includes a database with failover that 
  doesn't blow up the budget too much (compared to other components)?
 
 Ah. Deja Vu. :-) I'm involved with a startup company here in Sweden,
 and considering Orion as app server we are also contemplating what
 non-arm-and-a-leg-spending ways there are to enable a database for
 failover and load-balancing functionality.
 
 We are looking at either PostgreSQL, InterBase, SAP-DB or MySQL (yeah,
 I know) as our database. We have not seen any currently freely available
 solutions for failover for either of those (anybody?). So, we are thinking
 about implementing something by ourselves. Nothing definite so far, but:
 
 - The synchronizing of data will be done on the application-level, not
   by the database servers themselves. See below.
 - We'll avoid numeric sequences for record keys to make this easier.
   We will implement some unique key-generation scheme based on whatever
   is needed to make keys unique but still not rely on strict monotone
   numeric sequences or similar ( md5(table_name+timestamp()), perhaps? ).
 - We'll code a DB-abstraction layer that takes care of executing
   all update queries against all configured database servers and
   all read queries against one of them known to be alive and lightly
   loaded (or not recently accessed, or some other scheme).
 - I guess that if we need database-specific stuff such as stored
   procedures or similar we need to use the same database software
   for all failover machines.
 - If we stay away from database-specifics we could possibly allow
   failover between different database products. Would be cool.
   Using straight, standard SQL, could make this feasible.
 
 These are very premature thoughts, we are getting closer to the
 planning and design stage, but we haven't actually started yet.
 Any thoughts? Ridiculously naive? Or possible to pull off?
 
  robert
 
 /David
 
 Reach me by
 - Email: [EMAIL PROTECTED], WWW: http://www.dtek.chalmers.se/~d0dak/ -
 --/David---
 
 




Re: HARDWARE FOR J2EE apps

2000-10-12 Thread wim veninga

Hi Kevin,

Do you create the database in IBConsole ? Because the ibconsole has an bug when
creating databases.
If so, you probably have to get the newer version of IBConsole :
URL : http://www.interbase.com/open/downloads/ib_download.html
goto the miscellaneous section (at the bottumn of the page) and download
IBConsole - build 308. You must overwrite the ibconsole in your interbase
install directory. If you want the interbase-database-schema.xml (for orion cmp
managed persistence) mail me and i will sent it to you.

Further more i think interbase 6 is an very good database but I would like to
know
if it handle a large amount of connections at the same time (like 500 or
something).
There isn't much info about that in the docs.

Greetings Wim Veninga.



"Duffey, Kevin" wrote:

 Hey Scott,

 Interbase 6 is pretty nice in my opinion. Its SQL 92 compliant (if that
 means anything), its completely free and is now open-source, has a type IV
 jdbc driver, and it has a VERY small foot-print. It runs on most platforms I
 believe, and thus far its pretty darn fast. However, I have not been able to
 successfully create a database in it. I keep getting some stupid error when
 I try to create a database and I can't find any information on it. I'll have
 to post to the Interbase site probably..or maybe there is a newer version. I
 wish there was some more detailed info on how to completely set up
 Interbase. I can however connect via my java app and get connections out. I
 just can't seem to create tables and what not. Anyways..I worked with
 Interbase 4 and it was pretty fast, so I am sure 6 is better in many ways
 (that is not always true..but I hope it is).

  -Original Message-
  From: Scott Stirling [mailto:[EMAIL PROTECTED]]
  Sent: Wednesday, October 11, 2000 3:25 PM
  To: Orion-Interest
  Subject: RE: HARDWARE FOR J2EE apps
 
 
  Why would you use mySQL over Postgresl?  They're both free,
  but Postgresql has a
  JDBC driver that's XA-compliant.  Also, mySQL is known to
  blow away your whole
  database if it has a bad crash, whereas Postgresql is better
  at persisting data
  through a bad crash.
 
  How does Interbase 6 compare to Postgresql?  Is it free?
 
  Scott Stirling
  West Newton, MA
 
   -Original Message-
   From: [EMAIL PROTECTED]
   [mailto:[EMAIL PROTECTED]]On Behalf Of
  Duffey, Kevin
   Sent: Wednesday, October 11, 2000 4:01 PM
   To: Orion-Interest
   Subject: RE: HARDWARE FOR J2EE apps
  
  
   Hi,
  
sounds very nice but what about the database? how do you
  cluster that
without spending an arm and a leg? our experience is, that
it's not that
hard to set up clustered web services with static pages and
servlets but
the really expensive part is, when you want that high
availability for your
database. it doesn't buy you much if you have highly
available ejbs when
the database server goes down. many people use clustered
apache/jserv on
linux and cheap pc-hardware for high volume transactional
websites but have
a large enterprise sun running oracle in the back. anyone out
there running
a configuration with orion that includes a database with
failover that
doesn't blow up the budget too much (compared to other
  components)?
  
   Well, to start off with there is mySQL and the one I like
  the most is
   Interbase 6..a free powerful RDBMS. However, as far as
  clustering them..I
   don't quite know the best way. I would think a
  load-balancer or a switch or
   something, would be required. Nobody ever said it was
  cheap! ;) I used to
   think $25,000 could easily set up a website from front to
  back, software,
   hardware, etc..but not even close these days. If you are
  just starting out,
   I would use Orion for front-end and ejb-logic tiers in a clustered
   environment (for developing/testing), with a single server
  running mySQL or
   Interbase for the database. Once you get some funding and
  move beyond the
   concept phase, you should put a sizeable chunk to invest in
  co-locating your
   site and doing the full load-balancing setup. I would
  estimate a cheap setup
   with 2 front-end web/servlet servers failed over, 2 ejb
  servers, and a
   database cluster will still run around $100,000, to co-lo
  it..which is
   probably the best thing to do to make sure its up 24/7.
  
  
 
 





RE: HARDWARE FOR J2EE apps

2000-10-12 Thread Steve Mabbutt

Wim,

We've used IB extensively for the past few years - it's an excellent
low-maintenance choice (DBA-free) for small to medium applications, but
doesn't scale well to large numbers of users or heavy loads.  Depending on
the OS platform, there is a hard limit of 250 simultaneous connections - but
we often run into concurrency problems well below that number, even on large
Sun servers.

E-mail me if you need more info,
Steve

 -Original Message-
 From: [EMAIL PROTECTED]
 [mailto:[EMAIL PROTECTED]]On Behalf Of wim veninga
 Sent: 12 October 2000 18:53
 To: Orion-Interest
 Subject: Re: HARDWARE FOR J2EE apps


 Hi Kevin,

 Do you create the database in IBConsole ? Because the ibconsole
 has an bug when
 creating databases.
 If so, you probably have to get the newer version of IBConsole :
 URL : http://www.interbase.com/open/downloads/ib_download.html
 goto the miscellaneous section (at the bottumn of the page) and download
 IBConsole - build 308. You must overwrite the ibconsole in your interbase
 install directory. If you want the interbase-database-schema.xml
 (for orion cmp
 managed persistence) mail me and i will sent it to you.

 Further more i think interbase 6 is an very good database but I
 would like to
 know
 if it handle a large amount of connections at the same time (like 500 or
 something).
 There isn't much info about that in the docs.

 Greetings Wim Veninga.



 "Duffey, Kevin" wrote:

  Hey Scott,
 
  Interbase 6 is pretty nice in my opinion. Its SQL 92 compliant (if that
  means anything), its completely free and is now open-source,
 has a type IV
  jdbc driver, and it has a VERY small foot-print. It runs on
 most platforms I
  believe, and thus far its pretty darn fast. However, I have not
 been able to
  successfully create a database in it. I keep getting some
 stupid error when
  I try to create a database and I can't find any information on
 it. I'll have
  to post to the Interbase site probably..or maybe there is a
 newer version. I
  wish there was some more detailed info on how to completely set up
  Interbase. I can however connect via my java app and get
 connections out. I
  just can't seem to create tables and what not. Anyways..I worked with
  Interbase 4 and it was pretty fast, so I am sure 6 is better in
 many ways
  (that is not always true..but I hope it is).
 
   -Original Message-
   From: Scott Stirling [mailto:[EMAIL PROTECTED]]
   Sent: Wednesday, October 11, 2000 3:25 PM
   To: Orion-Interest
   Subject: RE: HARDWARE FOR J2EE apps
  
  
   Why would you use mySQL over Postgresl?  They're both free,
   but Postgresql has a
   JDBC driver that's XA-compliant.  Also, mySQL is known to
   blow away your whole
   database if it has a bad crash, whereas Postgresql is better
   at persisting data
   through a bad crash.
  
   How does Interbase 6 compare to Postgresql?  Is it free?
  
   Scott Stirling
   West Newton, MA
  
-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED]]On Behalf Of
   Duffey, Kevin
Sent: Wednesday, October 11, 2000 4:01 PM
To: Orion-Interest
    Subject: RE: HARDWARE FOR J2EE apps
   
   
Hi,
   
 sounds very nice but what about the database? how do you
   cluster that
 without spending an arm and a leg? our experience is, that
 it's not that
 hard to set up clustered web services with static pages and
 servlets but
 the really expensive part is, when you want that high
 availability for your
 database. it doesn't buy you much if you have highly
 available ejbs when
 the database server goes down. many people use clustered
 apache/jserv on
 linux and cheap pc-hardware for high volume transactional
 websites but have
 a large enterprise sun running oracle in the back. anyone out
 there running
 a configuration with orion that includes a database with
 failover that
 doesn't blow up the budget too much (compared to other
   components)?
   
Well, to start off with there is mySQL and the one I like
   the most is
Interbase 6..a free powerful RDBMS. However, as far as
   clustering them..I
don't quite know the best way. I would think a
   load-balancer or a switch or
something, would be required. Nobody ever said it was
   cheap! ;) I used to
think $25,000 could easily set up a website from front to
   back, software,
hardware, etc..but not even close these days. If you are
   just starting out,
I would use Orion for front-end and ejb-logic tiers in a clustered
environment (for developing/testing), with a single server
   running mySQL or
Interbase for the database. Once you get some funding and
   move beyond the
concept phase, you should put a sizeable chunk to invest in
   co-locating your
site and doing the full load-balancing setup. I would
   estimate a cheap setup
with 2 front-end web/servlet servers failed over, 2 ejb
   servers, and a
databa

RE: HARDWARE FOR J2EE apps

2000-10-12 Thread Lorin Kobashigawa

As far as DB redundancy, 

I have heard that it is possible to have two boxes using a shared RAID that
holds the data.  The first box is the primary database, and if that one is
down, the second one takes over.  The RAID itself taking care of the disk
level redundancy, and the two boxes handling the System level redundency.  

I have never personally tried this, but it sounds intriguing. 

-Lkb

At 04:26 PM 10/11/00 -0700, Duffey, Kevin wrote:
Is a veritos cluster expensive? I know we are using that here with Oracle
8i, but I am not sure if that is outrageous in price, and if it works with
all db's, or if its specific. Also, does Oracle, IBM DB2, and so on support
clustering by themselves..or does it require 3rd party software to make this
work?


 -Original Message-
 From: David Kinnvall [mailto:[EMAIL PROTECTED]]
 Sent: Wednesday, October 11, 2000 2:37 PM
 To: Orion-Interest
 Subject: RE: HARDWARE FOR J2EE apps
 
 
 Hi!
 
 On Wed, 11 Oct 2000, Robert Krueger wrote:
 
 [description of budget-friendly Orion setup - snip]
 
  sounds very nice but what about the database? how do you 
 cluster that 
  without spending an arm and a leg? our experience is, that 
 it's not that 
  hard to set up clustered web services with static pages and 
 servlets but 
  the really expensive part is, when you want that high 
 availability for your 
  database. it doesn't buy you much if you have highly 
 available ejbs when 
  the database server goes down. many people use clustered 
 apache/jserv on 
  linux and cheap pc-hardware for high volume transactional 
 websites but have 
  a large enterprise sun running oracle in the back. anyone 
 out there running 
  a configuration with orion that includes a database with 
 failover that 
  doesn't blow up the budget too much (compared to other components)?
 
 Ah. Deja Vu. :-) I'm involved with a startup company here in Sweden,
 and considering Orion as app server we are also contemplating what
 non-arm-and-a-leg-spending ways there are to enable a database for
 failover and load-balancing functionality.
 
 We are looking at either PostgreSQL, InterBase, SAP-DB or MySQL (yeah,
 I know) as our database. We have not seen any currently 
 freely available
 solutions for failover for either of those (anybody?). So, we 
 are thinking
 about implementing something by ourselves. Nothing definite 
 so far, but:
 
 - The synchronizing of data will be done on the application-level, not
   by the database servers themselves. See below.
 - We'll avoid numeric sequences for record keys to make this easier.
   We will implement some unique key-generation scheme based 
 on whatever
   is needed to make keys unique but still not rely on strict monotone
   numeric sequences or similar ( md5(table_name+timestamp()), 
 perhaps? ).
 - We'll code a DB-abstraction layer that takes care of executing
   all update queries against all configured database servers and
   all read queries against one of them known to be alive and lightly
   loaded (or not recently accessed, or some other scheme).
 - I guess that if we need database-specific stuff such as stored
   procedures or similar we need to use the same database software
   for all failover machines.
 - If we stay away from database-specifics we could possibly allow
   failover between different database products. Would be cool.
   Using straight, standard SQL, could make this feasible.
 
 These are very premature thoughts, we are getting closer to the
 planning and design stage, but we haven't actually started yet.
 Any thoughts? Ridiculously naive? Or possible to pull off?
 
  robert
 
 /David
 
 Reach me 
 by
 - Email: [EMAIL PROTECTED], WWW: 
 http://www.dtek.chalmers.se/~d0dak/ -
 --/David--
 -
 




/** 
 * @author: Lorin Kobashigawa-Bates [EMAIL PROTECTED]
 * @title:  CodeMonkey / COO - Robot6 Inc. 
 * @phone:  415.345.8872
 * @addr:   1177 Polk St. San Francisco, CA 94109
 */




RE: HARDWARE FOR J2EE apps

2000-10-11 Thread Duffey, Kevin

Well, for development, in terms of how fast you want to compile your code, I
would go with the fastest PC you can afford. However, I have a PIII650 with
256MB RAM, and 9GB HD that is more than fast enough for development
purposes. Unless you plan on playing games, you can probably get a $700
workstation made with oem/generic parts that works fine for development (not
including monitor). I would spend the most money on a big monitor, or a nice
19" flat-screen. Go for flat if you can..its easier on the eyes and takes
little room, but at 3x+ the cost of a good 19" monitor, I don't think its
quite worth it yet.

As for deployment, well..that depends too. You can go as little as a P133Mhz
(if you could find one) with linux, running a 100% java app server (like
orion)..although it will probably be pretty slow. Because Orion is 100%
Java, you can run it on any JDK 1.2+ platform, probably with little problems
on all of them. I think the best deal, being that Orion is clusterable, is
to use cheap Athlon or Intel based servers. You can spend $10K on a dual
900Mhz Dell server, or you can build a nice 1.1Ghz Athlon farm (say 3
servers per island, 2 islands) for as little as $1200 each. Since they are
servers, opt for 4MB or 8MB video card, Win2K OS, JDK 1.3, and Orion. That
is all we have on our server. We did put Diskkeeper, and have Terminal
Services installed (with win2K Server), which allows us full administration
of the server from anywhere. Figure $700 or so for Win2K, $1200 for the
Athlon server (although I would wait a few more months to use the Athlon SMP
servers coming out soon..where you can put dual athlons and they are more
inline with Intel servers), $1500 for Orion, and a few hundred for other
software (give or take). Each deployable server would cost under $5000 and
you would have pretty darn good performance. Ofcourse, you can go with a
Dell server and WebLogic..that would run you about $250K for the same setup
that will cost you $30K with Orion (being that WebLogic is only 10x more
expensive).

I would also say it depends on your application. If that new Mac 8" cube had
a dual processor in it, I would choose to run that! At $1700 for a computer
that does 3GFLOPS of calculations, it beats the Intel and Athlon hands down
in calculations..which is good for intensive math use. Thus, if you were
building say an imaging farm that had to turn html into faxes (pdf, rtf,
etc), I would opt for faster math capable cpus. Ofcourse..if I had my
choice, I would really choose the Alpha cpu..that is what the Athlon is
based on too.

But, SUN offers a pretty good setup too, for a lot more money (usually).

Honestly, if you want a setup that can probably handle several 1000 users at
one time I would set up an Orion farm using Athlon or Intel 1Ghz+ servers,
do two islands, each with 3 servers for fail-over and load-balancing (so you
can handle more people). Do the same type of setup for the EJB
servers..perhaps using dual cpu servers. Being that Orion costs nothing
extra for dual/quad cpus, its ok to go that route. You can even buy a dual
board with only 1 cpu and add a second cpu later (but don't wait more than 1
year..you wont be able to get that cpu if you wait that long..it will be
obsolete).

Hope that provides some insight.


   I would be very gratefull if anyone with experience tell me what
 it's the appropiate hardware to develope ejbs, servlets and 
 jsps, and what
 it's the appropiate in order to run an app with a great load 
 of concurrent
 users.
 
 I wanna know your opinion about what to choose 
 Unix(solaris,HP,..) or Win
 (intel) for both development and real app.
 
 TIA.
 
 -
 David Sierra Fern ndez
 --
 
  -- Sierr@ --
 




RE: HARDWARE FOR J2EE apps

2000-10-11 Thread Robert Krueger


-snip
As for deployment, well..that depends too. You can go as little as a P133Mhz
(if you could find one) with linux, running a 100% java app server (like
orion)..although it will probably be pretty slow. Because Orion is 100%
Java, you can run it on any JDK 1.2+ platform, probably with little problems
on all of them. I think the best deal, being that Orion is clusterable, is
to use cheap Athlon or Intel based servers. You can spend $10K on a dual
900Mhz Dell server, or you can build a nice 1.1Ghz Athlon farm (say 3
servers per island, 2 islands) for as little as $1200 each. Since they are
-snip

sounds very nice but what about the database? how do you cluster that 
without spending an arm and a leg? our experience is, that it's not that 
hard to set up clustered web services with static pages and servlets but 
the really expensive part is, when you want that high availability for your 
database. it doesn't buy you much if you have highly available ejbs when 
the database server goes down. many people use clustered apache/jserv on 
linux and cheap pc-hardware for high volume transactional websites but have 
a large enterprise sun running oracle in the back. anyone out there running 
a configuration with orion that includes a database with failover that 
doesn't blow up the budget too much (compared to other components)?

robert
(-) Robert Krüger
(-) SIGNAL 7 Gesellschaft für Informationstechnologie mbH
(-) Brüder-Knauß-Str. 79 - 64285 Darmstadt,
(-) Tel: 06151 665401, Fax: 06151 665373
(-) [EMAIL PROTECTED], www.signal7.de





RE: HARDWARE FOR J2EE apps

2000-10-11 Thread Duffey, Kevin

Hi,

 sounds very nice but what about the database? how do you cluster that 
 without spending an arm and a leg? our experience is, that 
 it's not that 
 hard to set up clustered web services with static pages and 
 servlets but 
 the really expensive part is, when you want that high 
 availability for your 
 database. it doesn't buy you much if you have highly 
 available ejbs when 
 the database server goes down. many people use clustered 
 apache/jserv on 
 linux and cheap pc-hardware for high volume transactional 
 websites but have 
 a large enterprise sun running oracle in the back. anyone out 
 there running 
 a configuration with orion that includes a database with 
 failover that 
 doesn't blow up the budget too much (compared to other components)?

Well, to start off with there is mySQL and the one I like the most is
Interbase 6..a free powerful RDBMS. However, as far as clustering them..I
don't quite know the best way. I would think a load-balancer or a switch or
something, would be required. Nobody ever said it was cheap! ;) I used to
think $25,000 could easily set up a website from front to back, software,
hardware, etc..but not even close these days. If you are just starting out,
I would use Orion for front-end and ejb-logic tiers in a clustered
environment (for developing/testing), with a single server running mySQL or
Interbase for the database. Once you get some funding and move beyond the
concept phase, you should put a sizeable chunk to invest in co-locating your
site and doing the full load-balancing setup. I would estimate a cheap setup
with 2 front-end web/servlet servers failed over, 2 ejb servers, and a
database cluster will still run around $100,000, to co-lo it..which is
probably the best thing to do to make sure its up 24/7.





RE: HARDWARE FOR J2EE apps

2000-10-11 Thread David Kinnvall

Hi!

On Wed, 11 Oct 2000, Robert Krueger wrote:

[description of budget-friendly Orion setup - snip]

 sounds very nice but what about the database? how do you cluster that 
 without spending an arm and a leg? our experience is, that it's not that 
 hard to set up clustered web services with static pages and servlets but 
 the really expensive part is, when you want that high availability for your 
 database. it doesn't buy you much if you have highly available ejbs when 
 the database server goes down. many people use clustered apache/jserv on 
 linux and cheap pc-hardware for high volume transactional websites but have 
 a large enterprise sun running oracle in the back. anyone out there running 
 a configuration with orion that includes a database with failover that 
 doesn't blow up the budget too much (compared to other components)?

Ah. Deja Vu. :-) I'm involved with a startup company here in Sweden,
and considering Orion as app server we are also contemplating what
non-arm-and-a-leg-spending ways there are to enable a database for
failover and load-balancing functionality.

We are looking at either PostgreSQL, InterBase, SAP-DB or MySQL (yeah,
I know) as our database. We have not seen any currently freely available
solutions for failover for either of those (anybody?). So, we are thinking
about implementing something by ourselves. Nothing definite so far, but:

- The synchronizing of data will be done on the application-level, not
  by the database servers themselves. See below.
- We'll avoid numeric sequences for record keys to make this easier.
  We will implement some unique key-generation scheme based on whatever
  is needed to make keys unique but still not rely on strict monotone
  numeric sequences or similar ( md5(table_name+timestamp()), perhaps? ).
- We'll code a DB-abstraction layer that takes care of executing
  all update queries against all configured database servers and
  all read queries against one of them known to be alive and lightly
  loaded (or not recently accessed, or some other scheme).
- I guess that if we need database-specific stuff such as stored
  procedures or similar we need to use the same database software
  for all failover machines.
- If we stay away from database-specifics we could possibly allow
  failover between different database products. Would be cool.
  Using straight, standard SQL, could make this feasible.

These are very premature thoughts, we are getting closer to the
planning and design stage, but we haven't actually started yet.
Any thoughts? Ridiculously naive? Or possible to pull off?

 robert

/David

Reach me by
- Email: [EMAIL PROTECTED], WWW: http://www.dtek.chalmers.se/~d0dak/ -
--/David---





RE: HARDWARE FOR J2EE apps

2000-10-11 Thread Scott Stirling

Why would you use mySQL over Postgresl?  They're both free, but Postgresql has a
JDBC driver that's XA-compliant.  Also, mySQL is known to blow away your whole
database if it has a bad crash, whereas Postgresql is better at persisting data
through a bad crash.

How does Interbase 6 compare to Postgresql?  Is it free?

Scott Stirling
West Newton, MA

 -Original Message-
 From: [EMAIL PROTECTED]
 [mailto:[EMAIL PROTECTED]]On Behalf Of Duffey, Kevin
 Sent: Wednesday, October 11, 2000 4:01 PM
 To: Orion-Interest
 Subject: RE: HARDWARE FOR J2EE apps


 Hi,

  sounds very nice but what about the database? how do you cluster that
  without spending an arm and a leg? our experience is, that
  it's not that
  hard to set up clustered web services with static pages and
  servlets but
  the really expensive part is, when you want that high
  availability for your
  database. it doesn't buy you much if you have highly
  available ejbs when
  the database server goes down. many people use clustered
  apache/jserv on
  linux and cheap pc-hardware for high volume transactional
  websites but have
  a large enterprise sun running oracle in the back. anyone out
  there running
  a configuration with orion that includes a database with
  failover that
  doesn't blow up the budget too much (compared to other components)?

 Well, to start off with there is mySQL and the one I like the most is
 Interbase 6..a free powerful RDBMS. However, as far as clustering them..I
 don't quite know the best way. I would think a load-balancer or a switch or
 something, would be required. Nobody ever said it was cheap! ;) I used to
 think $25,000 could easily set up a website from front to back, software,
 hardware, etc..but not even close these days. If you are just starting out,
 I would use Orion for front-end and ejb-logic tiers in a clustered
 environment (for developing/testing), with a single server running mySQL or
 Interbase for the database. Once you get some funding and move beyond the
 concept phase, you should put a sizeable chunk to invest in co-locating your
 site and doing the full load-balancing setup. I would estimate a cheap setup
 with 2 front-end web/servlet servers failed over, 2 ejb servers, and a
 database cluster will still run around $100,000, to co-lo it..which is
 probably the best thing to do to make sure its up 24/7.








RE: HARDWARE FOR J2EE apps

2000-10-11 Thread Cory Adams

How much time do you have?  I say that becuase you may burn a lot of time
trying to solve those database issues that are solved in some of the other
commercial products.

For replication you could use snapshots wither by doing exports/imports or
data copying on a periodic basis so that you could at least recover fairly
quickly  

I have not used DB2 but people have mentioned that it is fairly inexpensive
when compared to Oracle.  

I would have thought that the SAP product would have had failover and
perhaps even Postgress???

Cory

At 11:37 PM 10/11/00 +0200, David Kinnvall wrote:
Hi!

On Wed, 11 Oct 2000, Robert Krueger wrote:

[description of budget-friendly Orion setup - snip]

 sounds very nice but what about the database? how do you cluster that 
 without spending an arm and a leg? our experience is, that it's not that 
 hard to set up clustered web services with static pages and servlets but 
 the really expensive part is, when you want that high availability for
your 
 database. it doesn't buy you much if you have highly available ejbs when 
 the database server goes down. many people use clustered apache/jserv on 
 linux and cheap pc-hardware for high volume transactional websites but
have 
 a large enterprise sun running oracle in the back. anyone out there
running 
 a configuration with orion that includes a database with failover that 
 doesn't blow up the budget too much (compared to other components)?

Ah. Deja Vu. :-) I'm involved with a startup company here in Sweden,
and considering Orion as app server we are also contemplating what
non-arm-and-a-leg-spending ways there are to enable a database for
failover and load-balancing functionality.

We are looking at either PostgreSQL, InterBase, SAP-DB or MySQL (yeah,
I know) as our database. We have not seen any currently freely available
solutions for failover for either of those (anybody?). So, we are thinking
about implementing something by ourselves. Nothing definite so far, but:

- The synchronizing of data will be done on the application-level, not
  by the database servers themselves. See below.
- We'll avoid numeric sequences for record keys to make this easier.
  We will implement some unique key-generation scheme based on whatever
  is needed to make keys unique but still not rely on strict monotone
  numeric sequences or similar ( md5(table_name+timestamp()), perhaps? ).
- We'll code a DB-abstraction layer that takes care of executing
  all update queries against all configured database servers and
  all read queries against one of them known to be alive and lightly
  loaded (or not recently accessed, or some other scheme).
- I guess that if we need database-specific stuff such as stored
  procedures or similar we need to use the same database software
  for all failover machines.
- If we stay away from database-specifics we could possibly allow
  failover between different database products. Would be cool.
  Using straight, standard SQL, could make this feasible.

These are very premature thoughts, we are getting closer to the
planning and design stage, but we haven't actually started yet.
Any thoughts? Ridiculously naive? Or possible to pull off?

 robert

/David

Reach me by
- Email: [EMAIL PROTECTED], WWW: http://www.dtek.chalmers.se/~d0dak/ -
--/David---