Enterprise software stacks

[bikemike]

You have to develop applications - primarily Windows desktop, and WinCE, but increasingly web browser and soon mobile targets (Android, Apple and Windows)

You have databases that in some cases are over 5GB.

You need fast network performance

You need fast data analysis and charting

You need Rapid Development, all the hooks for great code coverage, code optimising, code versioning and code review, continuous integration.

You need some degree of DB neutrality - includes Oracle, and MS MSQL plus one of the good Freebies.

You have Delphi and C# experience, existing investment in MS VS products - which is getting expensive....you use Scrum, SVN....


What are your options? What would you do, if it were your call?

[Grasshopperus]

That's a helluva question. You're talking about creating suites of different types of software right? Not just one piece of software?


In terms of a presentation-layer, HTML5 is going to have some cool features allowing for offline apps, that might cover your desktop apps as well as staying cutting edge for the mobile market. You could keep your in-house VS.NET/C# in-house skills which is probably the biggest consideration.
Of course, HTML5 hasn't yet been ratified and HTML has a history of being implemented pretty randomly on different OS and browser combos but if you're looking at distributing full apps for winCE/android/iOS you'll have to learn new skills anyway.

If I was to choose I'd go JBoss/JSP/Java for application development as the JRE is implemented everywhere and allows great 'enterprisey' software design. It's what C# copied so it's not a big learning curve. Develop using Hibernate and you'll be relatively DB agnostic if you want to switch at a later date, although, MySQL is a bloody good DB and you should give it an evaluation.

Use Jira/SVN/Bamboo for user stories, versioning and build/deployment respectively.

[bikemike]

Yes, helluva question, and helluva decision. I'm pushing to consider 'options' alternate to The MS Way, so we make a conscious decision rather than assuming.... It's just that compiling the tool set seems more 'obvious' with our current setup. Selecting from other flavours to consider viable alternatives is 'hard' just because of unfamiliarity...

Yes different products in a suite, some interoperability, and some code-commonality.

I have used Eclipse previously, with the whole iSeries WebSphere thing. (My team used JBoss, and Hibernate - eventually used Elastic Path eCommerce). I get the shakes when I think of EJBs...

I have used MySQL before also, with my own website, but what about BIG databases and performance; I'll go see some biassed metrics on the web somewhere....lol

I've been looking at some of the ALM suites, and Atlassian's stuff is appealing. They seem to rate well in terms of keeping it relevant, agile and reasonable pricing. What are the weaknesses in Jira/Bamboo? (We use FinalBuilder).

[Grasshopperus]

I have used MySQL before also, with my own website, but what about BIG databases and performance; I'll go see some biassed metrics on the web somewhere....lol

I've been looking at some of the ALM suites, and Atlassian's stuff is appealing. They seem to rate well in terms of keeping it relevant, agile and reasonable pricing. What are the weaknesses in Jira/Bamboo? (We use FinalBuilder).

I probably should've mentioned that I'm biased towards OSS

Yeah, Eclipse as an IDE is awesome, heaps of great plugins. I use CDT and it's real nice.

MySQL is absolutely ready for prime time, ask; Digg, FaceBook and Wikipedia to name a few.

The weaknesses of Jira are that it's quite complex (powerful though) and an absolute PITA to upgrade, we just went from 4.0 to 4.2 and it took about 5 hours of ops time to get all the doco, upgrade tomcat and micro-manage all the plugins that we use. Before going to Jira we had used: Mantis, Trac and then Rally.

Bamboo's been a dream we've got it integrated with Groovy for unit tests and an in-house puppet-like tool to manage releases.

[bikemike]

Ta. See if I can keep the debate and decision rational. Let you know how quickly it falls to MS :/

While you are here then - how's Jira etc compared with Rally? I read the Forrester report, both have pluses, but If you've used both... :-)

[Grasshopperus]

Ta. See if I can keep the debate and decision rational. Let you know how quickly it falls to MS :/

While you are here then - how's Jira etc compared with Rally? I read the Forrester report, both have pluses, but If you've used both... :-)

Yeah well MS is pretty well entrenched.

The problem we had with Rally was that it is hosted remotely and it was slow. Tying our dev team's productivity directly to shitty NZ internet wasn't a winning formula. Jira has more plugins available and as you set it up inside your organisation you can control or create new ones to fit your task.

Jira does cost more but it's more popular meaning contractors we bring in are more likely to be productive sooner. Also, the higher quality tracking and quality of the software that we've been producing since switching over has paid for it many, many times over.

[imdying]

existing investment in MS VS products - which is getting expensive

Shouldn't be a problem, customer pays for it...

[bikemike]

Price is always an issue; so long as we choose the most appropriate product and deliver what the organisation and it's customers want, then the issue goes away.

We have some devs who don't see other options (not directly related but to the extent of not even knowing what Android was until three weeks ago..), and some who find MS stuff too layered, too complex, too slow (e.g. EF) and just out of reach in terms of the best bits of their offerings being on their best and exclusively priced Editions.

What do you use?

[imdying]

All of the things you mention, funnily enough doing all of the things you mention... given that you're in Christchurch I wouldn't be surprised if you worked here... but there's no Mikes on a BMW here

Given you're doing enterprise work, it just seems odd that the price of tools are a problem when the job is probably quarter of a million bucks plus.

Most of the mobile work is on Motorola and Intermec devices, which means it's usually a mix of C++ and .NET CF stuff, running on CE and it's derivatives, which makes SQL mobile the easiest choice (well, once you've written the libraries to speed it up a tad haha), but at the other end, it doesn't matter as much.