Archive for Uncategorized
July 4, 2009 at 6:21 am
· Filed under Uncategorized ·Tagged free beer, MonetDb, open source bi
The last few weeks I’ve been working with MonetDb on a SCADA project. Just like the last version (that I used a few years back) – with the SQL interface, it is still stable, simple and extremely fast at certain types of querys (it really can be an order of magnitude faster than Ora, DB2, MySQL and other traditional RDBMS for retrieving singe records or aggregate values from huge but simple datasets).
Then there is the XML interface with the same blinding speed and a completely different optimum application…
So why does it remain relatively unknown?
Is it the altruistic hardcore open source commitment (didn’t stop Apache)? Lack of a pretty package and slick admin tool (doesn’t seem to be a roadblock for SQLite)? Or perhaps a lack of acceptance of column-oriented databases (but Infobright and Vertica are far less mature, far more complex and are shit-hot right now)?
Go figure … (or comment on what I’m missing)
Permalink
January 28, 2009 at 4:59 am
· Filed under Uncategorized
…so one morning this dog starts to sing and, throughout the day, demonstrates his repertoire of folksy classics. The dog’s owner grows more and more excited and that eveing at a neighborhood bar, he boasts to anyone and everyone who will listen that his amazing pet can sing! Everyone digested the news with skepticism and doubt.
Determined to prove his claim, the next evening he drags the dog to the bar, quiets the crowd and encourages the dog to sing. The dog does sing and the two are promptly thrown out and told to never come back. The dog was way off key and sounded like shit.
Saw a bit of this when Ora 10g RAC came out … some co-workers were so excited that they could parallelized queries across the interconnects to multiple LINUX boxes that they ignored the fact it was slow and did not work well.
The latest technology this seems to is the new generation of integration tools – I’ll call them EAI tools though many will take issue with this lable – while I’ve seen tools like WebMethods work to support very specifc solutions, the notion that it is a general purpose enterprise swiss army knife for integration is bunk. Please comment, especially if you have first hand knowlege of even one deployment that met expectations and was on or under budget!
Permalink
November 22, 2008 at 12:06 am
· Filed under Data Integration, misc ·Tagged betrayed, Jasper, Open Source Integration, Pentaho, SnapLogic, Talend
The Jitterbit site now makes it difficult to find a community build other than a latest build – try looking for a GA or stable release without paying first! Jitterbit claims (disclaims?):
Jitterbit Community edition is a development snapshot, offering the latest build of Jitterbit as we move towards our next major release. The Jitterbit Community Edition is primarily tested and supported by the open source community and because of its “early release” nature it is NOT recommended for production use. The Community edition of Jitterbit comes with limited installation support from Jitterbit and does not include certification, QA (quality assurance) or support and services for production deployment.
Technicalities aside, this is contrary to the spirit of Open Source Software – it may technically comply with the OSD but …
You don’t see Redhat implying that Fedora is not stable enough for prime-time!
I want to be clear that I respect the company’s right to sell their very high quality software and hope it is successful but InfoWorld should take back thier BOSSIE award (if it was this year) and their use of an org domain is a sham … Open Source it isn’t.
Twice, I have had clients purchase five-figure subscriptions to open source software based on the results and confidence they acquired through consulting engagements with the community versions – Can’t do this with Jitterbit!
They are just using the label open source for marketing hype and impact leaving Pentaho (Kettle), Jasper (Talend) and SnapLogic in my os integration tool bag.
Permalink
July 20, 2008 at 4:33 am
· Filed under Data Integration, Uncategorized ·Tagged Data Integration, enterprise, ETL, Kettle, Pentaho, Tactical Data Integration
I was just looking over reviews of Integration tools by Gartner ans Forrester … pretty scary.
Two faults with the Gartner report:
The exclusion of open source contenders Pentaho and Jasper – Gartner does qualify that they only analyze tools from companies with minimum revenue or minimum number of production customers – While Pentaho may not yet have 300 subscribers to the commercial version (I have worked for two companies that do or .67% of the threshold), I’d bet just about anything that there were way over 300 customers supported by the community in 2007 when the Gartner report came out.
The other flaw with the Pentaho report is in their naming of the report – they review monolithic ENTERPRISE data integration tools with ENTERPRISE criteria but title their report Magic Quandrant for Data Integration Tools, 2007.
The Forrester Wave™ Evaluation of the Enterprise ETL Market is more appropriately named but kinda scary … I don’t know much about Forrester, but recognizing OWB as an option is a joke and Pervasive is given some credit – while pretty powerful, Pervasive is just a mash of some small scale tools – last time I used it, it requires multiple different scripting languages depending on the part of the suite you were in!
Permalink
July 9, 2008 at 1:29 am
· Filed under misc ·Tagged aoe, ata over ethernet, coraid, storage
I do not normally get too excited or geeked-out over equipment but today I helped a client install Coraid LINUX and setup a SAN for her data – she does statistical work and needs to store and access billions of records in pretty simple tables for analysis.
WOW! Coraid is cool. My client will likely buy one of their appliances but we went to prove it out with the open-source package (designed for there appliances and so only for x86_64). If I can install and configure this, you can rest assured it is not overly complex.
Not as fast as a SAN but still pretty fast. Imagine adding unlimited usable TBs (nfs) for just over 2x the street price of a reasonable SATA drive …
http://support.coraid.com/support/cln/faq.html
Permalink
July 7, 2008 at 3:17 am
· Filed under Uncategorized
A few characteristics are key to succeeding with tactical data integrations -
- They are concise – the purpose and business value can be summarized in a few lines
- They are quick – at Business Informatics, we avoid efforts that are expected to take over four weeks, from commission to delivery of a tested solution
- They provide immeadeate value – returns expected from future efficieincy and cost avoidance can be very real, but our tactics and methodology drive towards tangible returns shortly after deployment
- They are not tool specific – if you hear there is only one tool that can be used for a task, chances are, someone wants to exercise a specific technology. We always define the problem and design the solution before considering the tool-kit.
Enterprise IT and monster projects are approrpriate to address certian challenges and will always be necesary – We generally avoid these situations.
Permalink
