xlmpp is a multi author blog about Xtreme Large Scale computing.

We are fortunate to work in a team of highly skilled experts with individuals that have worked on MPP (massive parallel processing) systems for more than 20 years. Some of us have practically grown up with parallel systems and as such we have accumulated a wealth of knowledge about extreme scalability. Every day presents new challenges - that's what keeps us fresh and motivated.

As you will read in our blog, we enjoy what we are doing. We want to be open to new ideas but also mindful about old concepts or common pitfalls that sometimes get ignored at first. We try to balance theory and practical application and are fast to try out new things.

Total Cost of Ownership is paramount. We have seen many 'cheap' solutions that failed to deliver the TCO once all dimensions of cost and return have been put together. Certain technology can excel in some areas but create so much overhead or cost in others that the total picture does not make sense.

We have learned to be mindful of alternatives, as long as they can fit into an overall larger - arching - framework.

So here we are:

Oliver Ratzesberger

I have been with eBay since 2004 and lead Architecture and Operations for ultra large scale analytical and data processing infrastructures. With more than 15 years background in massive parallel systems I have worked on large scale parallel technology in Telecommunication, Financial and eBusiness. In the mid 90s I was involved with one of the first large scale telecommunication projects collecting and analyzing call detail records with more than 10^10 records in a single database table.

Before I got into massive parallel work I spent my time since the early high school days in software development. I taught myself C long before I was able to take classes and by the time I got to College I had pretty much mastered C++ and was working part time on development projects. From there I went into full-time software development and IT with a strong focus on Quality Management and Maturity Modeling.

I thrive on large challenges and complex algorithms or throughput problems. The systems we work on today are thousands of times larger than what we used to deal with a few years back. It has been a fun ride so far and there is so much more to come.

Darren Bruntz

Michael McIntire

Lucky is the best way to describe my career.

More than 30 years ago, as a young teenager, I wrote my first application to track contractor mail rates. Simple, stupid, and effective. Some years later the GIS industry found me, where I started administering systems and writing applications such as Digital Line Graph edge matching, ground water models, and toxic waste proximity analysis. By the late 80’s I had moved on to work with minicomputer Operating Systems performance, and then discovered MPP Databases, where I have remained and filled many roles from engineering to architecture.

My work in Parallel Data Systems started in 1988, but I was hooked by a 1991 project with 250GB of data in two big tables. There were few more difficult problems at the time than making that big data set produce meaningful information in a self service model. Do you remember your first 100GB table? Now, joining multi-TB tables 10,000 times a day is routine - how times have changed.

Five years ago I started working on eBay’s Massively Parallel Data Analytics and remain so today. eBay’s Analytic systems can best be described as a Data Democracy in which the business holds congress. I continue to advise and service that Data Democracy, which essentially becomes the balanced implementation of Cost, Performance, and Usability. If it’s possible, we do it. If it’s not possible, we figure out how to make it possible.

Lucky Indeed.