ACID, compensations and NoSQL cross-posting

A cross-posting relevant to my earlier entry.

Banks and ACID transactions

For the millionth time, I hate it when people exaggerate. (Yes, that's meant as humour!) But seriously, when you have a single point on a graph, you can make any line or curve you want fit it! In order to make true judgements, we need all of the facts. And as a scientist, I'm always on the lookout for experiments, experience, facts etc. that shoot holes in existing theories, since that's how science advances.

Now why do I say all of the above? Because we have a prime example of people jumping to the wrong conclusion based on incomplete facts. Once again it's around ACID transactions, because apparently banks don't use them! I'm sure that'll come as a surprise to many of my banking friends and colleagues, but it has to be true because they're not used in ATMs (in some countries). Based on this one fact (which I'm sure it correct in some circumstances/locales), it seems that all bank transactions are BASE and not ACID. I've discussed ACID, BASE and CAP before, so I won't go into details. However, it seems that this meme has been picked up by a lot of people:

Therefore, I just wanted to make sure everyone understood that there's a lot more to banking than ATM machines. ACID transactions are a backbone of a lot of what goes on throughout the financial services sector. Of course there are areas where ACID transactions aren't needed and shouldn't be used. I've said as much myself over the years. With all of the work we've been involved with around extended (non-ACID) transactions, that should be obvious.

So I do not have an issue with someone suggesting that ACID is not suitable for a specific use case. Where I do have a problem is when people jump to the conclusion that just because ACID isn't right for one use case it seems to mean it's not right for all use cases. That's like saying that atoms don't exist because you can't see them with the naked eye!

SRDC 2013

SRDC 2013

TransForm School on Research Directions in Distributed Computing

CALL FOR ABSTRACTS

June 10-14, 2013

Heraklion, Crete Island, Greece

http://www.ics.forth.gr/carv/transform/srdc/

************************************************************************

TransForm School on Research Directions in Distributed Computing aims at
the dissemination of advanced scientific knowledge in the general area
of distributed computing with emphasis on multi-core computing,
synchronization protocols, and transactional memory. A major goal of the school
is to explore new directions and approaches on hot topics of current research
in these areas (and more generally in distributed computing) and to promote
international contacts among scientists from academia and the industry.
Research work from all viewpoints, including theory, practice, and experimentation
can be presented at the school.

The school will include a series of talks by renowned researchers.
A list of the invited speakers is provided below (this list is expected
to be expanded in the future):

- Carole Delporte, Universite Paris Diderot - Paris 7 (France)
- Shlomi Dolev, Ben Gurion University (Israel)
- Hugues Fauconnier, Universite Paris Diderot - Paris 7 (France)
- Pascal Felber, University of Neuchatel (Switzerland)
- Rachid Guerraoui, EPFL (Switzerland)
- Maurice Herlihy, Brown University (USA)
- Anne-Marie Kermarrec, INRIA-Rennes Campus
- Petr Kuznetsov, TU Berlin/Deutsche Telekom Laboratories (Germany)
- Mark Little, Red Hat (UK)
- Maged Michael, IBM Thomas J. Watson Research Centre (USA)
- Alessia Milani, University of Bordeaux (France)
- Eliot Moss, University of Massachusetts (USA)
- Michel Raynal, University of Rennes I (France)
- Eric Ruppert, University of York (Canada)
- Nir Shavit, MIT Computer Science and Artificial Intelligence Laboratory (USA)
- Assaf Schuster, Technion (Israel)
- Corentin Travers, University of Bordeaux (France)
- Pawel T. Wojciechowski, Poznań University of Technology (Poland)


Abstract Submissions
*************************
PhD students or young researchers interested in presenting their work
should submit a 1-page abstract motivating the main research challenge
they are addressing and stating the approach being taken. A selection
of proposals will be chosen for presentation. Eah such presentation
will be of 15 minutes.

To submit a talk, please send an email to faturu AT csd DOT uoc DOT gr
by April 26, 2013. The subject of this e-mail should be of the following form:
": proposal for SRDC talk".
Every submission should be in English, in .ps or .pdf format, and
include the title, the names of the presenter and his/her collaborators
in the research work of interest, their affiliations, and a one page abstract
of the work. Students should also provide the name and contact information
of their advisors.

Abstracts will become available to participants electronically. Authors
will be given the option to upload their presentation on the school’s website.

Financial Support
*********************
TransForm will provide financial support to a number of researchers/students.
Those who intend to apply for financial support should send a 1 page application
(in addition to their talk proposal) which will provide a short description of their
travel expenses. Applications should be sent to faturu AT ics DOT forth DOT gr by April 26, 2013.


Important dates
******************
Submission deadline: April 26, 2013
Notification: April 30, 2013
School Date: June 10-14, 2013

Is Cloud the death of open source?

Over the last few years I've been hearing from various quarters that Cloud (specifically PaaS) doesn't need or work well with open source. At least what some of these people mean is that business models that have worked well for non-PaaS open source don't necessarily work for PaaS. I think the jury is still out on that one. However, if you look around at PaaS implementations out there, or even further up and down the stack to include IaaS and SaaS, it's clear that open source is playing a major role. Whether it's OpenShift, OpenStack. MySQL, Linux or a plethora of other components, it's hard to find environments that aren't built on open source in one way or another. (Excluding closed source companies, of course!)

Now why do I mention this? Because I'm just back from JUDCon Brazil and this topic of conversation came up with some of the attendees. In fact they were suggesting that several of the most significant waves in software over the past few years and into the next few years, are fuelled by the innovation within disparate open source communities. When you look at cloud, mobile, ubiquitous computing etc. it's hard to disagree!

A Raspberry Pi and vert.x related cross-posting

I've been making progress on another Pi-related project. Since it also involves transactions, I posted it on the JBossTS blog, but wanted to cross-post here for those who may not track that blog separately.

Travelling is a PITA at times!

I spent pretty much all of the week before last in New York and Boston visiting many of our financial services customers. It's a great opportunity to hear what they like, what they don't like and work together to make things better. I'd do this much more often though if it didn't involve flying! It seems that every time I fly I end up getting ill; usually a cold or (man) flu. Unfortunately this time was no different and when I got home at the weekend I could feel something coming. Sure enough, it was a heavy cold and it laid me up for days (I'm still not recovered completely yet).

Then while I'm recovering I remember that I missed QCon London. I vaguely remember many months ago while planning that my trip conflicted with QCon, but it had slipped from my mind until last week. It's a shame, because I love the QCon events. However, what made this one worse was that I appear to have completely missed the fact that Barbara Liskov was presenting! It's been the best part of 20 years since I last saw Barbara, so it would have been good to hear her and try to catch up. Back in the 1980's I visited her group for a while due to the similarities between what they were doing around Argus, replication (both strong and gossip based) and transactions, and of course what we were doing in Arjuna. She was a great host and I remember that visit very fondly. Oh well, maybe in another 20 years we'll get a chance to meet up again!

Updating vert.x examples on the Pi

As I mentioned earlier, I've been repeating earlier experiments around running vert.x on a 256 Meg model B Raspberry Pi. I was working through all of the Java examples and didn't have time to go through them all again. But here's an update on the rest.

For a start, I should Fanout Server works, but again you may need to install a telnet client on your Pi, or run remotely. However, the HTTPS example failed (on Safar, Chrome and Firefox) and I remember this happened in December when I first did this work. I reported the problem to Tim and hopefully we'll get to the bottom of this eventually:

The Proxy example works too, but you need to remember that the Pi isn't an i7 multicore laptop, so just be patient:

PubSub is another example that often requires a little patience before trying to telnet, but the results are worth it:

The Upload example is relatively simple:

Route Match works too, but remember that the text in the example is wrong and the actual directory is route_match and not routematch!

Make sure you have your CLASSPATH set correctly before running the Resource Load example:

And I think that's it. As I mentioned before, if you have any problems reproducing any of this then let me know and I'm happy to try to help.