This blog post describes yet again how you can use best-of-breed components to do the job.

In particular, it describes integration of Atomikos as well as all of the other cool libraries that together outperform the application server in simplicity, cost and performance… Good job!

The recent announcement of Apple's cloud illustrates yet again that a new era is about to start: that of the cloud - and with that, the massive adoption of service oriented architectures (SOA).

This will have an impact so huge that it is hard to underestimate. Some of the obvious changes that will happen are most likely the following:

• Infrastructure (operations) will become agile - at the expense of IT jobs in operations. Applications will be deployed with unprecedented ease and speed. This is because the cloud infrastructure will transparently allocate the resources that now are allocated by human administrators.
• SOA will become mainstream. This is because even the most common infrastructure services will be virtualized, running in the same or a different VM from the application, and scaling up and down as needed.

The interesting part - at least to us - is that transactions will become more distributed than ever. We are ready for this change - are you?

Data consistency in the cloud is a totally different ball game - as perfectly illustrated by this newsletter issue from ZapThink. Essentially, you have to cope with some form of partition tolerance and stale data values in caches. This is usually summarized as the CAP issue (for more details, see the ZapThink post).

Luckily, the solution to the problem can be as simple as outlined in our solution to the CAP problem.

Life is good

The cloud is an interesting phenomenon, to say the least. Just last week, I witnessed a demo by Adrian Colyer at the What's Next (Paris) event that left me very impressed because I could easily understand the problems that it solves:

• Slow infrastructure setup is replaced by instant availability of resources (databases, hardware)
• Scalability is trivially easy: on command is all it takes
• Services are promoted as the means of reuse - effectively enabling SOA strategies

I know from experience that these are all pain points in today's IT infrastructure for just about every big organization. So for mainstream operations, I think the cloud is going to be a killer.

While this is all fine, I still wonder how this will work out for one of our core markets: finance. Sure enough, initial offerings are showing up, but so far nothing really interesting for our retail banking audience. Add to this the fact that most banks are typically very conservative (they have to be) and the cloud for the banking world seems a bit more of a future situation than for the rest of the industry.

But certainly that time will come too, and for those days Atomikos has the exact kind of technology for cloud-enabling transaction guarantees that banks depend on so much. That is because we offer the third generation of TP monitoring technology: light-weight, embeddable, perfectly scalable and SOA capable too.

The future is bright;-)

Check out this interesting presentation on InfoQ by Cesare Pautasso.

We at Atomikos are working with Cesare to research REST and Transactions and how to integrate this with BPM for REST as well; check out this slideshare presentation (also by Cesare).