Friday, March 24, 2006

Playing with vista

For some reason Vista is so particular about the hardware it goes on. Its almost best suited to be run in a Virtual Server. It doesn't even run right in the newest Vmware Server builds. Oh well, its cool when its working :)

Tuesday, March 14, 2006

CMDBs and what it needs to contain

1. Should a CMDB contain specific application configurations?
2. Should it contain what is installed and what it requires (both other hardware/software/etc)?

I am under the impression that it should be answer number 2. Our Chief Architect thinks number 1. I can see how number 1 is important, but I think that number 2 is also easier to manage and helps you understand the basic systems, without getting into specifics. I also think that the deployment and configuration of software should not be controlled by the CMDB (although it can have a part in feeding its data into a build process).

Some of our developers want to build a APPDB which has these application specific items tracked in it, but it doesn't help to answer number 2 which is more important in my opinion.

All very interesting stuff. Either way we are very far off here, but we need to get our asses in gear.
Java Profilers

I have to run a profiler on some old apps as we move them over to modern app servers. This means finding a profiler which supports the old stuff and the new stuff.

Jprobe's older revisions support it, but they dropped it in the newest revision. I may check it out, but its not a good idea I don't think.

OptimizeIT is an old profiler (not updated since 2004), but it support the older stuff and the newer. We also have special pricing on Borland stuff. After they are trying to sell off their tools its probably not a good product to buy.

I am talking to CA (Wyly) this week too. I don't know how that will turn out.

More on this later.

Saturday, March 11, 2006

Symphoniq

This product is still causing us pain. I may be pulling the plug, not sure. Depends how the replication of problems that we saw goes. I am continuing to evaluate I3 more, but that too is having all kinds of issues on the install that we did. Nothing is ever easy.

Back to my mantra of if it doesn't interfere (non-intrusive) then its much easier to implement without causing product problems.
Welogic vs Websphere

I haven't used either of these products for around 3 years. I have to make some decisions as far as where we move our products to. Right now a lot of these legacy (java) apps run on Netscape application server 2.0 (which is from 1998). I need to move some of these onto a modern app server. I have been going with WebLogic, as we have a site license. I am working on the first app move over to it. From what I see WebLogic hasn't gotten much better since version 7.0 (we are moving to 9.1). I wonder if anyone has comments about websphere? I know IBM has been gaining market share, but I'm not sure if the product is actually that much better.





Tuesday, March 7, 2006

Symphoniq is working

I finally got symphoniq up and running here (with the .NET monitoring). The product looks very solid, but I'm a bit concerned about the instrumentation they do on the webserver (AKA it can break our products). It does the low level instrumentation, as well as the end user (round trip time). Its a pretty cool product, and the use of MS Reporting services is a great addition.

I am evaluating this against Symantec i3, which we have a site license for. i3 is pretty cool, but its complex and the agent are hard to deploy.

We shall see what the other technical folks and business side people think of the product and if its something we cna move forward with.

I will keep you all posted about what I think of it.

Sunday, March 5, 2006

yum or apt-get?

I have been a avid FreeBSD supporter, and before that NetBSD. I started using NetBSD in 1995, and moved to using mostly FreeBSD in 1997. At work as I moved to larger and larger companies, I used way more Linux (Aside from the 80% Windows). Getting more and more busy I started to appreciate the simplicity of Linux and upgrading versus the painstaking upgrading in FreeBSD. Yes FreeBSD is better, but do I have time to compile all of the software and kernels, and do upgrades which are multistep?

cvsup
build world
build kernel
portupgrade

versus a single command like yum or apt-get?

I really like yum in comparison to the other Linux upgrade systems that RedHat has used before. The problem is that you can't use Yum to move from something like Fedora Core 4 to Fedora Test 5. They say that they are adding this functionality into the upgrade system, but it will be beta for a while, and not really trustworthy yet.

This is why apt rocks, and there is nothing that does as good as job. Its tried and true and just plain works well. The alien commands to rebuild rpms and such works like a charm (not 100% accurate). That's why on my personal servers I run Debian, and love apt. That's where I am now on the lay on the free *NIX land.

Let me know what you guys think?
This one is a little off topic, but I wanted to highlight some of the Firefox extensions which I am in love with and I would be lose without them. They are in order from most important to least important:

http://www.longfocus.com/firefox/gmanager/ - Lets me manage my gmail accounts, and keep tabs on how much email and spam each account has.

http://fasterfox.mozdev.org/ - Another must have, allows you to tweak FF to enable the browser to run faster, things like allowing more connections per webserver and such.

http://adblock.mozdev.org/ - Must have, can block ads and other items you don't want to see, allows for regexp matches.

http://www.customizegoogle.com/ - This one you need to have it will do things like force gmail over SSL, customize the display of ads, and change the way pages are displayed and ranked in google.

http://216.55.161.203/theonekea/tabprefs/ - You love tabs? We all love tabs. This one lets you manage the way your tabs behave. I love the fact that I can switch tabs by holding my mouse over them. It probably saves me 500 clicks per day :)

http://www.mrtech.com/extensions/local_install/ - With the number of extensions you see I run, and always playing with another one or two, you know we need a solid extension manager. Well this tool takes the cake, its the best extension manager out there.

http://ietab.mozdev.org/ - Lets you run certain tabs with the IE engine, and use regular expressions to auto load certain sites in ie. For example I load *.msn.com Domains with IE from within FF.

http://users.rcn.com/shoofy/forecastfox_enhanced/ - Gotta keep in touch with the weather and sat maps right in your FF status bar.

http://smoothwheel.mozdev.org/ - Makes the scroll wheel on a mouse act more "smooth", and allows you to tune the way the wheel works in FF.

http://jennifermadden.com/ - This one pretties up the view source, color coding the html and such. Makes it easier to read and analyze the source.

http://mozmonkey.com/ - (Switch Proxy) This one allows you to manage and switch proxy servers quickly. I am often doing things that I don't want to be tracked to me, so I use this and my list of proxy servers to better hide myself as needed.

http://searchsidebar.mozdev.org/ - This one does search result aggregation, its pretty good, but I prefer using Copernic Agent for the most part.

https://addons.mozilla.org/extensions/moreinfo.php?id=1912 - This extensions allows you to search for a person on many sites at once. Its pretty handy, but I don't use it a whole lot.

I'm currently playing with this extension :
http://showcase.uworks.net/ - It shows the open tabs in a mini view, I'm not sure how useful it is yet.

Thursday, March 2, 2006

Why is WebLogic so complex? These app servers just have no simple way to manage and configure them. I haven't used WebLogic since version 7. Now I'm dealing with a port from an ancient Netscape Application Server (version 2.0 or something) to WebLogic 9.1.

We have a site license for the BEA products, so that's the strategy that we are moving towards. Makes sense to have a common app server.

You'd figure these J2EE app server companies would simplify configuration and then enable the "advanced" configuration for the small number who need to do that. Compared to Jboss and Resin these servers are just poorly designed and overly complex.

Microsoft seems to have it down right on IIS, when is the Java world going to simplify things, so that the common user doesn't have to mess with 100 dials to make a simple web-app work?

Wednesday, March 1, 2006

How come everyone has scalability problems, but no one really talks about them. People have no idea how to build a fully scalable infrastructure. Google, and Akamai seem to do this very well in general, but for the rest of us we aren't so gifted.

EX:
As things, especially in market and financial data scale up, we have a hard time keeping up with it. Look at the breakout of options data, and new regulations which require us to track more data, and shove it all down the pipes we have. We need some smarter burstable technology to handle the mass updates. Someone at work was talking about pulsing data. Cool ideas, minimize the updates per stock to 3 per second versus updating all of the changes to a ticker.
I live in Boston, but I have to go to New York a lot for work. The joys of day trips. So I had meetings this morning from 8:30am until now... around 2pm. I have been talking to a lot of people about some of the work I am proposing we do around here:

1. Fix entitlement systems - replacing them with off the shelf software (SSO and IM).
2. Figure out application dependencies - by using tools like nlayers
3. Build a SOA management infrastructure - by using tools like Actional and Amberpoint

Figure out how we are going to build a CMDB with the data we get from these tools. What vendors are good for CMDBs.

Any advice on CMDBs?

Can any of the SOA tools build the dependency maps? From what I understand and have been told by vendors they will build some of these maps. At least logical layer maps, that's a good first step.

I need to figure out how to feed this data to our business folks and give them usage, cost, and consumption numbers. If we can do that then we will really be making progress on this stuff.

This goes into monitoring and moving from active monitoring to passive and real-time monitoring. Understanding behaviors and anomalies or stastical deviation as an alarm. Move from building manual monitors into actually monitoring the behavior of a lot of metrics and then alerting based on changes of those. These are forward thinking ideas that we need to bring us into a proactive and predictive mode from a reactive mode.

How about product usage analytics capability and usage tracking. That's another project I am working on. Many proposals on the table here.
I am a technical guy, but my transfer into the arena of understanding how technology grows is a new area I am looking into. I have a strong background in everything from development, networking, security, hardware, software, and just about anything you can think of.

My quest to stay on top of whatever is new and hot is never ending. This could be a latest phone, or the newest database platforms.

I will explore some of the issues and products in this blog which will give you a better understanding of the issues that my employer faces.

I work for one of the largest financial data providers in the market. We provide data and software to pretty much every major financial institution out there. We also have products which do trade execution, tracking, and messaging.