Stack versus best of breed
Best of breed, or the ‘stack’? This is a long-fought argument between companies that specialise in an area, and those that have for one reason or another broadened their horizons.

When I used to meet the CEO of BEA Systems, Alfred Chuang, he used to argue vociferously in defence of his company that was – predominantly – focused on the application server space. ‘We don’t do hardware, we don’t do applications, we do infrastructure software,’ he would say.

After the firm he ran was sold to Oracle for $8.6bn in 2008, he presumably had to change his tune during his short time there (as an aside he has since left to head up Magnet Systems, developing social applications for business).

In any event, Oracle’s acquisition spree – which really kicked off in earnest under the watch of president Charles Phillips in 2003 – has been designed to turn it into a full ‘stack’ vendor. Already king of the database market, it rapidly acquired some key enterprise applications assets, not least PeopleSoft, Retek, Siebel, Hyperion and BEA Systems.

It rounded out the ‘stack’ by acquiring Sun Microsystems for $5.6bn in April 2009, giving it not just servers and storage (lest we forget that Sun had bought StorageTek for $4.1bn in 2005) but also considerable strength in identity management, the iconic Java language and the open source MySQL database which Sun had also bought.

Oracle now says its numerous assets are, "Designed and tested as an integrated, end-to-end stack".

But for smaller companies such as Infor – our cover stars in the next issue of CBR, so watch this space – there is the opportunity to use middleware to integrate their and third parties’ technologies into something that still looks like a ‘stack’.

But which is going to offer better ROI in the medium to long term? I’d be delighted to hear from readers which approach your organisation tends to use and how successful it has been, whether you prefer to post anonymously or not. So which is better? Stack or, er, not stack?