Wednesday, June 24, 2009

Scaling Hibernate

Very interesting talk about Hibernate Shards and Hibernate Search:

Emmanuel Bernard (JBoss) is an author of Hibernate Search (among others) and Max Ross (Google) is one of Google engineers behind the Hibernate Shards.

It is interesting to see what kind of challenges developers have to deal with when it comes to horizontal scaling. One of the take aways from this talk for me is that horizontal partitioning logic should be a part of application code. One should not rely on database built in partitioning. Hibernate Shards takes this idea one step further by adding one more abstract layer called virtual shards.

Illustrations and Paintings on the Web

I haven't been posting any illustration or painting related stuff lately. Unfortunately, I haven't finished any new piece myself so I have nothing to show now. On the other hand, there are tons of great artists on the web so here comes a list of few links which I found interesting:

Wednesday, June 17, 2009

Open Source and Cloud Computing

Some people expect the Cloud Computing to hit the top of Gartner's Hype Cycle [1,2] by the autumn this year.
As a fan of this "new emerging" architectural approach I am happy to see that open source community joins this hype as well. I think there are actually two important activities going on right now:
  • Apache community started shaping their ideas about how to build a software stack for the cloud. Click here for the main Lab wiki page, of particular interest can be a list of presentations too.
  • The other important moment is a strong recognition of cloud by the SpringSource (the company behind the Spring framework). Rod Johnson clearly stated in his keynote (see 01:51 of the footage) at SpringOne this year that:
    The cloud is really going to change the world of IT and therefore is something we absolutely need to consider in terms of our view of the world.
    I am sure Spring developers will keep this on their minds which can reslut in smooth integration of Spring and cloud deployemts and architecture requirements. The proof that SpringSource guys prefer action over words is recent acquisition of Hyperic.
From now going on, the fact that clouds are getting darker does not necessarily have to mean that something bad is coming. The opposite can be the truth - the darker the clouds are the brighter the future for them can be.

Monday, June 15, 2009

How Google handles large-scale graph computation

A Pregel seems to be the answer. It seems that this parallel computation model is worth keeping an eye on. I hope Google will reveal more details about their solution soon.

Wednesday, June 10, 2009

Monday, June 1, 2009

Quick Glance at Bing

Microsoft released its Bing search service. I gave it a quick glance this morning as I wanted to find out if it can be useful to me (as a Java developer). So I tried two simple tests:

Test #1

Query: Java File API

Two top most results are for JavaDoc, that is what I am lookng for.

Two top most results are irrelevant to me. It seems to prefer localized sources (though I did not check local search option) and provided some links to local Czech java related portal ( but the articles were not specific to Java File API.

Google provided good links but first of them was for 1.3 and the second for 1.4 API version. Because I was looking for more uptodate API of Java File object I decided to give both search engines one more hint.

Improved query: Java File API 1.5

The link I was looking for was second on the result list.

The link I was looking for was first on the result list.

Test #2

Query: Groovy clousure

It realized that clousure is a typo. It provided corrected "Did you mean" service link and the two top most results were relevant, then it provided results for original search terms.

It did not realize that clousure is a typo and provided some results having the same typo in the text. As of writing the first result is some page in japanese which is little help to me (no transaltion link provided).

My conclusion:
As of now I am staying with Google. But I welcome Bing as it will make Google engineers think how to improve their search.

As for the Google search service it was probably using my personla profile and search history to improve search results (I was logged into my Google account). However, I think that all the search queries were quite simple and straightforward so still the results should be more or less comparable.