I found it in the book store next to my university. Bought it.
Read the introduction, and browsed through the first chapter (Ruby). Failed to do the exercises (a mixture of feeling I already know a bit of Ruby, and lack of time and motivation to really get my hands dirty).
Stopped reading the book.
I will start over now, I promise.
Just noticed that somebody had already made a product/service out of the idea to integrate Hadoop and Cassandra, two major hypes names in the Big Data, NoSQL space. – which is exactly what I had been planning as new architecture for my last project.
Datastax now offers Brisk:
- CassandraFS has the same interface as HDFS. So, in particular, you should be able to use most Hadoop add-ons with Brisk.
- CassandraFS has comparable performance to HDFS on sequential scans. That’s without predicate pushdown to Cassandra, which is Coming Soon but won’t be in the first Brisk release.
- Brisk/CassandraFS is much easier to administer than HDFS. In particular, there are no NameNodes, JobTracker single points of failure, or any other form of head node. Brisk/CassandraFS is strictly peer-to-peer.
- Cassandra is far superior to HBase for short-request use cases, specifically with 5-6X the random-access performance.
Checkout their white paper.
Google announced that the new Google Analytics (currently in beta and only rolled out to selected accounts) supports event goals.
So, if you have been using GA to track events within your Flash game, Social Network app or anything else, you can now interpret them directly as goals and don’t need to fake Pageview-calls of “virtual” (non-existing) pages in a hidden div or iframe (which is what I have been using before).
Interestingly, no mentioning of event funnels so far – which is strange, since funnels can be defined for “normal” goals anyway.
I’m currently digging my way through the Kontagent documentation to learn about their appropach towards A/B testing.
It seems that the normal (i.e. “free”) , subtype-based variant needs to be managed by the application itself, to define ratios, select a variant to show/use, etc. and Kontagent only tracks the resulting metrics, whereas their full fledged “Viral Optimizer” (only in premium) lets you define variants, ratios, etc. within the Kontagent web interface itself, generating code to copy/paste into your app.
Their A/B REST API documentation is lacking though, so we will see if my assumptions are true.
In the meantime, I’m also testing several versions of my home page using the Google Website Optimizer.
Let me tell you a bit about my history:
My background is twofold – I started out as a software developer for Alcatel, but switched into some project management, development lead, CTO and managing director positions, both inside and outside of the games industry (namely Rockstar Games, VeriSign, and a small german startup called plazz entertainment).
From my perspective, BI or data mining (if you can call it that) in the “traditional” retail-oriented games industry (think Rockstar shipping Manhunt 2 or Grand Theft Auto) back in 2005 has been pretty basic, and mainly focused on usability and play testing, like generating 2D heat maps for level architectures from people playtesting the game – with that, the game designers and level designers could determine which were the most and least frequently visited areas in a level, and be use that information to change the level structure and/or level elements accordingly to guide the player better.
Of course, Microsoft/Bungie (Halo) has been very active in that field, doing much more than that, and also talking about their approach and findings quite openly at conferences.
Things changed for me when I became CTO of a german startup, doing their own “Club Penguin” clone in Germany called “Atlantis“. Initially, we had a freemium (free basic access with premium subscription as option) business model. this was changed to the now typical microtransactions for virtual item sales.
Since we were building the complete platform from scratch, we had a lot of different areas to tackle, so BI was mainly seen as a “hobby” of mine, rather than the strategic necessity for commercial success I thought it is or should be.
Basically, we did aggregated log events from the database (a MySQL single read/write master, multiple read-only slaves setup) per time slot – like number of chat messages per day, time spent online per character per day. We did this using an open source tool called ART.
On the website, Google Analytics (and Piwik) provided the standard metrics like Visits, Average Time on Site, Bounce Rate. I thought about using Ajax-calls into Google Analytics to track ingame events, but decided against it, not to mix web traffic stats and ingame events (that need to be stored in the database, anyway).
My employment there ended before I had time to research and implement something more elaborate and detailed. I’d be very interested to talk about what others did, what tools or methods you use, and so on.
