It may turn out that the way I geek out may be blogable, but it will require some changes of habit. I'm on my way to those changes.
The past couple months have been quite a revolution in my own thinking, and in many ways, I haven't been so excited about my profession in a very long time. The essential deal is that I'm finally moving towards what I should have been moving towards - virtualization, cloud services, elastic BI and all that crazy infrastructure. Now that I have begun spending a lot more time in Ubuntu and off traditional Windows, it gives me a lot more opportunity to break navigation habits. Plus, I'm exposing myself to things and people that I ordinarily would not, and that's all good.
What I'm likely to do is come across a number of new technologies and packages and whatnot and just catalog it here. The vistas are very large and that may be all the time I have to do it without looking ridiculous. It's refreshing to poke my head in places where I'm a complete newb... So anyway, here goes.
First.
Apatar / Talend / PDI
These are the three open source ETL tools that I'm likely to be playing with. Apatar seems the simplest and most straight forward. Talend, I haven't gotten open yet, although I'm sure to and the PDI has so many bells and whistles it's hard to know where to start. Chances are, however, that I'm going to have to find some very specific reasons to use them, depending on a couple things.
1. I have yet to build a datamart out of ERP that sent data into an EPM system that was too big for MSSQL and DTS to handle. That may change, but as facile as I am with T/SQL (and PL/SQL) I'm really thinking long and hard about the test cases I will require.
2. The chances are good that I will become familiar with Amazon's Virtual Private Cloud, and that essentially means that (maybe) wherever I am, I can spin up a server or three which is instantly ready to handle my ETL requirements. As fast as I can push data through a pipe, I can send it to an AMI with the ETL of my choice and not neet to install either of the tools. So while it's clear that ETL becomes even more important in cloudy ways, what is in your toolkit is even less dependent on customer requirements.
3. Essbase Studio is much improved from the EIS of old, and it actually seems to perform in the ways that SQL heads prefer. I can remember their zero tolerance for whatever kind of pig EIS used to be, so that's a wonderful thing. Less need, I percieve, for ETL packages to take a central role.
Its still a bit weird to take all this data at arms' length through all the interfaces now available - but it the package is reliable and performs well, why not abstract? As we used to say of disk being cheap, now environment is cheap.
--
Hackerspaces
There are six hackerspaces in LA and Orange County. Continuing my efforts to make acquaintances in the Open Source world, this is my next step. So I first found some guys at Nullspace Labs via IRC, and that's quite a wakeup call. I haven't used IRC in a decade, and forgot what kind of heads are hanging out here. There still isn't much going on in the EPM arena, but in the noSQL area there are plenty. It's just a matter of bending tech to see where the markets overlap technically, and technically is what I care about these days. So I'm going to be able to work with folks who hack their way through everything, and I'm attracted to the physical hacking as well. This is part and parcel of my getting to understand the new levels of security that will become increasingly important.
--
Speaking of security, the last time I started thinking about it, I kept wondering why it wasn't taken quite so seriously in the BI world. Back in 2001 I had only two customers who bothered to get to what I believe was called C1 level security by the DOD's red book or blue book, or whatever that was. That was Boeing Huntington Beach, and what they did was simply anonymize the Essbase server name, cube names and user names. So one couldn't casually know what was going on when ZLKN00, sent a request to ONDK23 on server UFNS39, although I imagine a clever person could suss out the port mapping.
--
Anyway, I got XChat running on my Ubuntu plus Colloquy on my Mac and iPhone. Really good resources out there. Completely forgot about IRC, I did. It remains superior to Twitter in form and content. I'll try to keep that in mind again.
--
Tor
Thinking about Tor. My first guess is that I don't need it, of course, but I'm trying to learn everything I don't know, which includes a lot. So I'll try that out.
I also crashed my Iomega again, but am very well Backblazed. I'm also going Western Digital and considering the flaws of Dell hardware. It turns out anyway that I'm going to make my compute hardware light on the fat client side, have all my data duped in the cloud, and use more of AWS on demand. Which means at long last, I'm going to have the server power to keep all of my accumulated Master Data up and spinning.
I'm sure I have several scores of dimensions I've keep in outlines over the years. As soon as I get my Essbase AMI up again, I'll reverse them down into denormalized tables and then peruse the library. That's where the ETL tools will come in handy - to have redundancy in multiple formats. JSON, xml and the xml flavors, plus simple old Access, which I'm finding is a really good ubiquitous format. But we'll see more about that later.
--
There's also something cool called Bitcoin I'm going to check out, sucker that I am. Yeah I contributed to SETI online back in the day. Why not get some bitcoin?
--
Did a bit of flexing the old infrastructure muscles. I did pretty much a full install of EPM 11.1.1.3 on my home server in half a day. Couldn't get EAS working but did get MSSQL 2K8 Express, Oracle 10g, Essbase Studio, Foundation, Shared Services up. I had to dump some old oracle homes which never seem to die as services, and I've got some legacy 9.3.1 stuff lying around that complicated things, but it was nice to at least move stuff from place to place. Interesting how the MSSQL data wizard makes it easy to move stuff into Oracle, but I know I'm over-relying on that tool. But it does work...
--
The Python's coming along slowly, I am on vacation...
Comments