A new article reviews three J2EE app. servers: BEA's WebLogic 5.1 and 6.0, iPlanet E-Commerce Solutions' iPlanet Application Server 6.0, and Sybase's EAS. While all three servers dissapointed the reviewers, they concluded that BEA WebLogic's problems were the most manageable.
Read Java Application Servers Simplify Web Development
- Comparing BEA Weblogic, iPlanet, and Sybase by jason legris on August 09 2001 16:24 EDT
- Comparing BEA Weblogic, iPlanet, and Sybase by William Louth on August 09 2001 19:01 EDT
- Comparing BEA Weblogic, iPlanet, and Sybase by Cary Bloom on August 09 2001 08:31 EDT
Comparing BEA Weblogic, iPlanet, and Sybase by Cary Bloom on August 09 2001 08:32 EDT
- Comparing BEA Weblogic, iPlanet, and Sybase by Saji Mathew on August 09 2001 08:36 EDT
Comparing BEA Weblogic, iPlanet, and Sybase by Dimitri Rakitine on August 09 2001 09:35 EDT
- Comparing BEA Weblogic, iPlanet, and Sybase by Samrat Sengupta on August 17 2001 01:16 EDT
- Comparing BEA Weblogic, iPlanet, and Sybase by Pri Vasu on August 10 2001 08:28 EDT
Every client I've been with has faced the same issue: websphere or weblogic. one day maybe we'll know... until then, reviewers keep missing the boat.
Yes, you are correct reviewers keep missing the boat. Most of the issues they raised about the 3 appservers are the strengths of Borland's AppServer. PowerJ crappy then try JBuilder 5. WebLogic console flaky then try the Borland AppServer Console. There is even a command line tool for most of the functionality presented in the Borland AppServer console.
And not forgeting that Borlands AppServer is much faster than the rest irrespective of the cedrics 'for loops'.
yeah, and george bush is the "elected" president of the USA!
"And not forgeting that Borlands AppServer is much faster than the rest"
yeah ... and george bush is the "elected" president of the US!
You meant Goerge Art ?
Before a heated debate erupts I would like to state that my comments regarding performance come from MY own testing of different appservers with J2EE applications I have built. I have not tested all appservers.
As I have stated in the book J2EE In Practice:
"In early benchmark tests specifically tailored to the problem domain, Borland AppServer appeared to outperform other offerings."
So you should performance benchmark inline with the characteristics of your architecture/application.
Another point raised is that performance is not the only criteria. I agree totally and I have stated in the book the many other items that AT&T Unisource considered before chosing Borland.
My own view is that over time that the performance differences will become smaller though they will exist and that there is another battlefield (technical that is) waiting for J2EE vendors to march onto.
Could I just leave with one little gripe I have at the moment. How can a person say that they are a J2EE/AppServer Technology Consultant AND advise a company on the choice of application server when they have only ever used one. What appserver will that consultant recommend?
Could I just leave with one little gripe I have at the >>moment. How can a person say that
>>they are a J2EE/AppServer Technology Consultant AND advise >>a company on the choice of application server when they >>have only ever used one. What appserver will that
The answer is that they can't in all rights recommend an app server.
Regarding William's question:
I think TheServerSide and/or the Middleware company
and/or J2EE profesional consultants including me :-)
should test BEA, IBM, Borland, IPlanet, EAS
appservers before we can talk about them in fair
"Do not walk behind me, for I may not lead. Do
not walk ahead of me, for I may not follow. Do not
walk beside me, either. Just leave me the hell alone."
zen master -
I would like to add just one more story to this thread. This story shows that sometimes the decision is not made by management but other interested parties.
I was last year headhunted by a banking software company to be the architect of their latest and greatest offering. It was going to be a rewrite of a system at that time deployed in many banks in Northern Europe that just could not scale; everything was written in pl/sql.
The money offered was great with a relatively large number of shares on the table. This was the flagship product of the company and the next release was going to move them into the big time (so I was told).
Well after meeting the teams involved and hearing about their ambitious undertaking I was offered the top job. But before I took the job I needed to resolve something that had troubled me during the interview process. This company, which had no experience with appservers, had mentioned that they where already considering which appserver to work with. I offered my opinions on the market leaders and informed them of my preferred choice considering the undertaking and the experience levels of the team with J2EE (It seemed a nonbrainer to me even an consultant hired from a major consultancy company agreed with me). Well I phoned the MD and had one last conversation and tested my real powers. During the phone call I was told that their biggest customer had told this software company just that day which application server to choose. The MD said he had no option since this bank was going to foot most of the products development costs (the would be the major beta user).
OK, you might think that hey maybe this bank knows what it is talking about. Well you would be wrong because this same bank had purchased the current banking product which was already a disastor in terms of performance and scalability. In fact when they bought this product they marched in to this companies head offices with a team of 15 IT people and instead of having concerns about scalability and performance, which would have been so apparent to most engineers, they asked could the buttons be moved on some screens and the colors changed on some frames.
I am always asked by teams, on projects for which I have been brought in to try and save, "Will, is this the worst project you have been on"
I always answer, "No, there's always next week"
I really need to get back to my product. Bye.
Last year, I did a 2 week gig for BrandX application server in a cook off at a large company. We were given some J2EE code and asked to implement it. They also threw in some gotchas which were not really a test of the app server itself, but of our own ability. Stuff like being able to call a COM service in an MTS server etc. We did have some trouble with some of the code and it turned out that they had developed and deployed "flawlessly" on BAS. Not being overly partisan towards any particular product (so long as it is Java and not the heathenistic MS stuff from hell ), I asked them why they did not just go with BAS. It turned out, they were a heavy CORBA shop (Visigenic) and they had decided/been mandated to move the companies strategic direction towards J2EE. So... it seemd to me that BAS was the PERFECT choice. However, Borland "failed the financials" which I understand is a very common non-technical non-product merrit based litmus test which many companies use when performing product selection. I suspect that this may be a specific or more general reason why Borland has not fared as well as other companies in this arena. BAS seems to be the Rodney Dangerfield of the app server market, despite its great strengths.
As a closing note, I have seen many comments along the lines of " my EJB server runs great ". It is tough to take that without a grain of salt, because in many instances, I see people very happy with performance, but they are only doing 900 transactions a day. Good for them, I say, but how many of our peers are REALLY challenged with requirements significantly challenge an application server ? I think it makes sense to design and build for growth, but I suspect that many of these <insert your ejb server here> projects could have just as well been done using a nice servlet engine.
Hi guys does any one, know if there are any article around which addresses the comparision between HP Total-e-server and the BEA Weblogic???
What are "cedrics 'for loops'" ? Is it in the Patterns section?
Is it in the Patterns section?
It definitely should, I'm sure we can find a lot of code out there that uses it.
I guess this is in reference to http://www.theserverside.com/home/thread.jsp?thread_id=8319
May I ask you a stupid question? Have you actually done a performance comparison between Borland's and BEA's offerings.
I have done for a large American finance house before my association with Borland. The findings are consistent with many other findings from European customers. Borlands offering is many times faster than WebLogic when you start to use EJB's instead of just simply having some jsp/servlets running in a webcontainer.
I should point out that the finance house did not go with Borland even though the whole development team wanted it and it had been proven to be technically superior ( a bit like the US presidency and many other aspects of the American way). That was then but its much different now.
Another point Borland did have a performance whitepaper up on the web detailing who much faster Borland was than WebLogic but because of legal issues they had to remove it. It had nothing to do with the fact that the report was inaccurate. Look, I think it best that you simply perform some benchmarking yourself.
I am looking forward to ECPerf results from WebLogic and Borland.
I have to agree with your point that it is better to do performance benchmarking yourself, on your application or one that makes similar demands on the server as your application. Depending on the type of actions your application does most frequently; the perceived speed of application servers may vary. In addition, we must not forget that speed while important is not the only factor in choosing an application server.
INDEPENDANT general benchmarks with all known products (don't forget BAS, this time) and recent results
I guess http://ecperf.theserverside.com
initiative should be a good reference.
That's could be nice in order to choose a limited number of AS , you could bench by your own with specific functions.
(and not testing AS that make a JNDI lookup in 1s!)
IMHO, choosing a AS is a long and costy task, (raw performance is not the only criteria ... political point of view is very important too)
Politics is indeed a key player in AS decisions, unfortunately. We JUST had a situation where EA Server would have been the perfect solution, but management went with BEA because they were the more "known" company (and more costly). Managers hate to be seen as "making a mistake" and hardly tend to be innovative in their directions or choices - especially in very large companies. Its kind of like companies that bring in an Arthur Anderson type company instead of a good independent consulting house (okay - I admit - I run a good independent consulting house :-) because its the safe decision. And then they can say to their investors "Hey - we brought in a BIG_NAME_COMPANY or APP_SERVER - isn't that great?" and everyone (except perhaps the development staff and certain IT management staff) is happy. The way of the world, unfortunately. I think there will always be more than one way to solve problems - BEA isn't always going to be the solution, believe me. Its a fine product if your company can pony up the cash and it handles the requirements. But in the example above, one component of the problem was to take a powerbuilder app and transform it into a web-based app. EA Server is perfect for that. Now we are using BEA, which is good for some other things but not this project.
I guess my point is, even though some company, be it Borland, BEA, iPlanet, HP, Sybase, Oracle, ATG or whomever, can score highly on performance tests, even independent ones, that a tremendous number of other factors (should) go into making the final app server decision that goes (well) beyond market share or number of developers or that sort of thing.
I think the phrase should be 'commercial considerations' rather than politics. It's always great to use the technology you think is best but sometimes 'the management' do have to take into account commercial considerations, e.g. the funding customer insits on dictating the app server (so you have a choice their way or no way) or you have serious doubts about the vendor's long-term commitment to the app server business.
I know J2EE certification should make for portability but as anyone who has actually tried moving an app from one product to another will tell you it isn't simple. The ease of a project also still depends upon the level of support from the vendor and if the vendor is cash-strapped, interested in other areas, can't retain staff etc. you'll suffer in your project.
But I'd still like to use the best and fastest, if only we could work out which it was!
IMHO, EAServer is the best Appln Server. We are using EAServer in Production on two platforms (Solaris and NT). It's amazing and working fine.
Basically, i am a PowerBuilder developer. As a developer, i need openness and freedom to use any language to write my component. It allows the developer to write components in Java, PB, C, C++ & VB. I wrote my components using PB and aanother java expert wrote some components using Java. I didn't undergo the learning curve to learn java to code the components. Another fantastic feature is, u can make inter-component calls. i.e., call EJB component from PB Component and vice-versa. It is a full J2EE compatible Appln Server. It is very powerful in Clustering and Load-Balancing. In fact, our Production systems are running in Clustering with Load-Balancing. What else u people want ? Don't u think, these are the nice features to be considered to choose EAServer as an Appln Server. Will WebLogic or WebSphere allow a developer to write component other than Java. Bcos of market share, don't just talk about WebLogic and WebSphere ...
why in the the world should politics be involved? Do these corporate numbskulls not realize the beauty of J2EE is that it's a common API??? You're no longer playing the vendor-lock-in game, yet too many companies are still thinking in that paradigm :( I've seen projects flush tens of millions of $$'s down the drain because they were forced to use certain technology/vendors by ignorant "superiors"...
the majority of IT projects fail: I'd like to know what percentage of those projects involved heavy handed corporate politics, the wrong people making the wrong decisions, etc..
preaching to the choir :)
The unfortunate truth that I have been faced with is that at some upper levels within large organizations, decisions can and are made "on the golf course". Sad, but true.
Unless we are living in some sort of bizarre paradox, decision-making (of any type) is usually conducted by those who have "manager" in their titles – after all, they have a “legal” responsibility in many circumstances. This is the way it is – so get over it (or become a manager). The fact that techies bleat on about management not making the correct decisions - its a technical decision not a business one - is naïve in the extreme. Decisions must be made by taking into account many types of factors and constraints; technical, architectural, external influences, strategic requirements, financial, risk/reward, etc, etc. This is the case with the entire IT environment, not just J2EE app servers. If management are not making the correct decisions – remember, it IS their role to make decisions – from the point of view of technical people, then it is more likely that the technologists do not have the ability to make their views heard in a reasonable and structured manner.
In many respects, disconnect between technologists and management tend to infer a lack of architectural governance in the organization. That is, a structure that supports effective relationships and processes to direct and control the enterprise in order to achieve the enterprise’s goals by adding value while balancing risk versus return over IT and its processes. Governance is all about making the right decisions about technology in a way in which all constraints and requirements are balanced (and tradeoffs are made on a firm foundation of knowledge). It’s about creating a decision-making framework for IT that is bigger than both individuals and technologies.
So, don’t complain about an inability to make change. Be constructive, understand why the current governance (or lack of) approach in your organization does not work and do something about it.
i am working with power builder nvo component accessed by JSP web client.Though i am new to this technology so facing so many difficulties
As for example to get string array from PB nvo object method i am unable to do in PB.
So please help in this regard as you are working in thi stechnology
We were evaluating 3 app servers for our company.
BEA, Websphere and Borland App.Server.
I must admit that BAS is 4-10 times faster than BEA and Websphere, much more easier to maintain, and at least but not last, much cheaper.
Thanks God, our management is not that stupid to listen sales people but not their thech. stuff.
We are using now BAS, and experience with it is fantastic. Much better than any other app server I have dealt with before. (Those include BEA, Websphere and SilverStream)
I think that support is one of the most important things to choose appserver vendor. Taking into account Borland's support policy for JBuilder I'm not going to buy their appserver.
BAS customers, what can you say about Borland support for appserver?
I see Borland released ECPerf results - lets see how they stack up.
Reading the article seems EAServer - the actual server and management tools - did great - the article bashed PowerJ (which is admittedly poor) and then decided to give the nod to BEA anyway... go figure!
As for the election - read the papers Bush WINS in a recount! Sorry you lost but get over it!
Actually there is problem with weblogic.You cant deploy more than 300 beans in weblogic.For that you have to make a lot of changes at OS level.Does websphere,Borland application servers behaves in the same way??
I think we may never know if it is websphere, weblogic, iplanet or whatever. All whitepapers, performance valuations, comparisons are biased and sometimes politically motivated and also totally depends on the knowledge of people in that particular tool etc. Unless technical people make decisions on technical related stuff without political/management influence this trend will continue forever.