Musings of a West Texas Business Venture
Running a High-Tech Company ... In a Low-Tech World.
Subscribe / Syndicate
Previous Posts
- New Client ... Servosity
- The Difference Between Marketing, Advertising, PR ...
- Web Page Sizes
- TimeForge.com version 1.4 Press Release
- TimeForge.com Update
- Speed of a Website
- IEEE Paper on Embedded Mesh Networks
- Accounting Software
- Latest Updates
- Employee Scheduling Software At Its Best
Archives
Thursday, June 29, 2006
Pow Wow
Yesterday, I got a document from my own "gang of four", as it was very apparent that they were not talking or communicating well. We had allocated two of our programmers to work on the "front end", and two of the programmers to work on the "back end". In theory, this sounds like a good plan.After about a week of this, it became apparent that neither group was really excited about talking with the other group. The "introverts" were in the back-end group, and the "extroverts" were in the front-end. Who was working on the AJAX portion? Who was working on the XML formats? These questions were not being answered by the groups, and each group decided to go their own way with these .... non-trivial decisions.
So, I stuck them in a room, with a catered lunch, a whiteboard, and a notepad.
Six hours later, presto, document with defined requirements.
PDF Libraries
One of our clients has asked for even more PDF generating and storing capabilities, with a new set of systems and procedures requiring (likely) more than the Open Source community has provided. JPedal may be an option for us.The new project will require pulling information out of a database, and placing this information onto multiple pages within a PDF (say, 3 pages). The first page will also need a graphic placed on the side, which is pulled from another database, and is likely a different size / orientation every time. It will need to be resized and oriented and super-imposed onto the first page, underneath the data already on the page (like a water mark). Other pages will be a bit simpler, requiring placing data onto PDF forms, and merging them together (using something like the PDF Stamper from iText).
Should be interesting.
Sunday, June 25, 2006
Website Changes Launched
We updated some of our website over this past week, including some new information about Online Marketing. More will be coming, as we revise the website to target more small and medium business owners, both for computer consulting and online marketing.Right now, we're looking for clients in the Houston, Lubbock, and Dallas markets.
Check out the new website here.
Lubbock Auction Items
Several weeks ago, I offered up an item to auction for the Lubbock Chamber of Commerce. However, the auction at the annual Lubbock Business Expo didn't go over like the planners had anticipated, and they stopped the auction.The Lubbock Chamber of Commerce has decided to auction off the items using Parker Auctions, who appears to be a private auctioneer. My major concern with Parker Auctions doing the auction is the fact that there is a 10% fee to the high winner. What? A 10% fee? That's crazy. eBay has no fees for buyers, neither does Yahoo Auctions or pretty much any other Internet auction site that I know of. The credit card fees are 3% at most, I would imagine. So what are the other fees for Parker Auctions? Hosting / traffic? Maybe $5 for all of the auctions at the high end. I am aware that someone needs to get paid (and Walter Parker apparently is getting paid), or perhaps the profits go to the Chamber of Commerce. I don't know, but it seems awfully high.
Regardless, our item is listed here. The title would make you think that we only do marketing, but that's an incorrect assumption.
If you are in need of our services, regardless of your geographical location, feel free to bid on the auction item.
Saturday, June 24, 2006
Advertising News
Some interesting Internet advertising news has come up recently, which I thought I would point out. I'm not sure how much of this is applicable to our local market (Lubbock), but it certainly is very applicable to our clients which have regional and national coverage.1. It has come to my attention (I haven't been paying attention) that ValueClick and FastClick merged some months back. I remember seeing the headlines, but didn't realize that it would mean that ValueClick's entire model would be changed up. Our rep at ValueClick recently called me to let me know that our old program, which was up for renewal, wasn't likely going to happen. Instead of driving traffic to our sites that we manage, they want to provide the "new" version of co-registration leads, which means that they are leads that signed up for something else (or maybe they really want what we are offering), and they come with qualified addresses, phones, etc...
I was never impressed with co-registration style leads, and once watched, helplessly, while a client received 14,000 of them, at $12 a piece, and only 5 of them actually paid any money -- about $1,000 each. Which means that they spent $168,000 and received $5,000 back. That's a pretty negative ROI.
2. One of our clients is wanting to work with AdValiant, having heard something good about them, somewhere. I'm a bit skeptical of the company, since there is little, or nothing, to be found of them anywhere online. They have no media kit, and the website is very, very bare. However, one of the positives is that the test is costing almost nothing.
3. Google appears to be offering CPA style advertising. I haven't seen any mention of it from our rep, but I did see a post here about it. I wish Google well with their endeavors conquering the world. I hope it works much better than their video advertising program, which was (and is) a complete failure in my opinion.
4. Microsoft recently blasted me with some information about their new product, AdLabs. I am going to be playing with it later in the week, and it looks like an interesting cross of WordTracker with some behind-the-scenes information from Microsoft Search.
5. Yahoo Search is also still in the game, and I have a WebEx presentation with our rep there sometime next week to see some of the power of their new platform. I hope it is all that they promise it to be.
Friday, June 23, 2006
I am back in Lubbock, after a grueling trip to New York. I helped a client move from one location (Roslyn Heights) to their new location (Garden City), over the course of a few days. Wednesday and Thursday were the physical move, and then Friday through Monday was mostly clean up.
The move was not exactly uneventful from a technical side of things, consisting of shutting down four servers (a firewall, a database / web server, a file and email server, and a fax server) and moving them over to the new location. Simple? Perhaps.
We did have several "gotchas" during the weekend. The biggest issue was that the fan and heat-sink fell off the CPU of the fax server, grounding itself to the bottom of the case, but still plugged into the mother board. When we powered it back up (not knowing that the issue had occurred inside of it), it shorted out and made a quick, high-pitched whine, and quit. Luckily, the hard drive remained intact, so it was just a matter of finding a new computer to host the hard drive and the Digiboard (for the modems).
Other technical issues that we ran into included swapping RAM around (we moved 1GB from the database server to the email server, but the email server will only accept ECC RAM), and a corrupt file on the database server which disabled some reports. Once the corrupt file was replaced, it worked fine. Likely, the culprit for the corrupt file was a previous unclean shutdown of the hardware at the old location (the machine hung while being shut down), and the old version of ext3 that it is running doesn't like to be shut down uncleanly.
We also took the time to add another processor to the email server, and removed a SCSI RAID-1 array that was causing some issues. One of the drives is bad, but I'm not sure which one (when I tested them independently, they worked fine).
Most of the issues we encountered were already known, or were telecom issues. It took six (6) hours more than expected to bring the data T's online, and when the phone circuit came up, we discovered that the phone numbers had not all been ported, and would be more than 3 weeks before some of them are. The client went without phone service for the entire weekend.
From the data side, with the exceptions of the few issues above, the transition was smooth. Reverse DNS and SSL certificates were known to work, and the router re-configuration and DNS updating (we host this clients' DNS systems) worked very well.
The move was not exactly uneventful from a technical side of things, consisting of shutting down four servers (a firewall, a database / web server, a file and email server, and a fax server) and moving them over to the new location. Simple? Perhaps.
We did have several "gotchas" during the weekend. The biggest issue was that the fan and heat-sink fell off the CPU of the fax server, grounding itself to the bottom of the case, but still plugged into the mother board. When we powered it back up (not knowing that the issue had occurred inside of it), it shorted out and made a quick, high-pitched whine, and quit. Luckily, the hard drive remained intact, so it was just a matter of finding a new computer to host the hard drive and the Digiboard (for the modems).
Other technical issues that we ran into included swapping RAM around (we moved 1GB from the database server to the email server, but the email server will only accept ECC RAM), and a corrupt file on the database server which disabled some reports. Once the corrupt file was replaced, it worked fine. Likely, the culprit for the corrupt file was a previous unclean shutdown of the hardware at the old location (the machine hung while being shut down), and the old version of ext3 that it is running doesn't like to be shut down uncleanly.
We also took the time to add another processor to the email server, and removed a SCSI RAID-1 array that was causing some issues. One of the drives is bad, but I'm not sure which one (when I tested them independently, they worked fine).
Most of the issues we encountered were already known, or were telecom issues. It took six (6) hours more than expected to bring the data T's online, and when the phone circuit came up, we discovered that the phone numbers had not all been ported, and would be more than 3 weeks before some of them are. The client went without phone service for the entire weekend.
From the data side, with the exceptions of the few issues above, the transition was smooth. Reverse DNS and SSL certificates were known to work, and the router re-configuration and DNS updating (we host this clients' DNS systems) worked very well.
Wednesday, June 14, 2006
New Cluster
The new cluster that is going into this New York location will be a significant step in the right direction. The current database system is running on a 4-year old dual-Athlon 1900 MP system, which I custom built. It's currently running (2) IDE drives, RAID'd together using linux software RAID, and (3) SCSI drives, two of which are also RAID'd together. It's got 3.5GB of RAM, and has been a good workhorse for the last 3.5 years.About two years ago, we noticed that this particular clients' email systems were slowing down (specifically SPAM checking), and recommended a new system to offload local printing, DHCP, Samba, and email processing. A new system, a dual-capable Opteron with 2GB of RAM and (4) 200GB hard drives was placed at the clients' site, and necessary files and systems were transferred over. What a change two years make – that server is now having problems keeping up!! I'm currently en-route to the location with another processor, and another 1GB of RAM which should help with the majority problem (SPAM processing).
I shipped, from Lubbock, 3 brand-new Monarch Computer servers, each one with (2) 320GB SATA drives (hot swappable), 4GB of RAM, (3) 10/100/1000MB network cards, DVD drives, and (1) dual-core Opteron CPU. These machines can take (2) more SATA drives, (1) more dual-core Opteron, and an additional 12GB of RAM, in case the client sees any of the systems slow down. We don't anticipate that will happen within the next three or four years.
However, and more important than the actual hardware, is the move away from Informix to Enterprise DB, a derivative (with commercial support) of the wonderful PostgreSQL 8.1 software. We've been working for a few months now to port over 250,000 lines of code to run on this new platform, which should give the client significant redundancy within the cluster, and almost unlimited scalability (more nodes can be added on the fly).
More on the software architecture at a later date.
Client Trip
I'll be headed out of town early this next week to go to New York to visit a client. On the trip, I'll be installing two new T1's, a new router, a secondary firewall, and a new cluster of back-end servers. We're anticipating a rollover which will start on Friday around 2 am for this client, and should be done by mid-morning on Monday. If all goes well, DNS should be propogating changes within a few hours of the change over.
We ended up using HeartBeat as a decent solution to provide roll-over capabilities to the firewalls. This means that there will be two firewalls, both generic (white-box) linux systems with fast, redundant hard drives, and a medium amount of RAM. One firewall (the more powerful of the two) will be running as the primary system, both for incoming and outgoing connections. This includes web browsing (using squid), web requests (using apache as a reverse proxy), DNS requests, email filtering (SPAM filtering is actually done on another system), and all VPN requests.
This firewall (fire1) will be using regular “heart beats” to check on the secondary firewall (fire2). In fact, the secondary firewall (fire2), will also be checking on fire1. If the secondary (fire2) discovers that the primary is down, it will take over the responsibilities of the primary (fire1). This roll-over will provide the necessary hot-swap capabilities to the firewalls. And, one of them fails, we can determine when to fix it (if there was only one, and it failed, we'd have to immediately repair it).
The largest problem that we are likely to have while working with this type of a system is the fact that changes on one system are not necessarily made to the second. One way to get around this is to use a third system to store configuration files (like DNS records, squid configuration files, and apache configuration files), but this would create another potential problem. Changes for this particular client are few-and-far between, so the sys-admin making the changes will just need to remember to change both.
Friday, June 09, 2006
Only a Few Minutes
After explaining my reservations about the OpenLaszlo platform to our summer interns, and the feasibility of having our front-end render in HTML (rather than Flash) within a matter of weeks, not months, we made an immediate switch yesterday to another View component of our MVC framework. [John Sundman seems to have found my post, and offers up some insight here.]We've looked at Rails in the past, and I think it's a nice framework (it certainly has a lot of people excited over the possibilities), but I'm not sure how quickly we can decouple the Rails View from the Model and Controller (MVC pattern). As such, we'd likely end up writing a lot of the code by hand, kind of defeating the whole point of Rails. Instead, we moved toward PHP.
PHP certainly has a lot going for it as a framework and language, and it's been around for quite a while. We mostly focus on Java in-house, but the necessity for a quick, easy-to-learn framework which can be prototyped fast has left Java in the dust. Before the end of the day was up, we had PHP installed and running on the necessary machines, and then had downloaded and installed the other necessary components.
Our interns were able to research a few of the PHP frameworks (which can be easily decoupled from the back-end), and settled on Smarty as a template framework, and Zephyr as an AJAX framework. Both projects appear somewhat stable, and we'll also be throwing in Prototype.JS and hand-coding where necessary.
Our only kink is going to be the (perhaps major issue) combination of the View with the Model and Controller. Our back-end functionality uses the Acegi Security layer to integrate with Spring. In theory, this is a back-end only security library ... but we're well aware that Acegi can also set cookies, etc... to validate users. There could be some interesting integration issues between the two layers.
Thursday, June 08, 2006
Manamana
This video on Google caught my eye today -- I most recently heard the song in a Dr. Pepper commercial featuring a waiter and a couple dancing and singing in a restaurant. I couldn't place the original until I saw this today.OpenLaszlo is not going to work
Our two summer interns have been toiling in their own individual ways, learning what appears to be the next-big-thing, OpenLaszlo. It's very remarkable technology, capable of taking some written scripts, and then kicking out Flash files which look amazing. Even better, and the feature that enticed us to use them, was the new render engine that they purport to have, which kicks out DHTML, using the same scripts on the back-end. Very, very neat.However, we found out yesterday, after doing some digging, that the actual support will not be around until sometime in late 2006. Since my summer interns will likely only be around for the summer (though this is yet to be determined), we will need a way to either get the OpenLaszlo team to hurry up (not likely), or look at a different front-end technology.
My main requirement for a front-end technology is to have something that can be picked up very, very quickly. Light speed. I don't want my two developers sitting on their hands for three weeks while they learn the intricacies of Spring and Tomcat and Hibernate, etc.... Just so they have the stack that they need to develop a front-end. That's not very much fun, and is a poor use of our dollars, probably taking at least 6 weeks to pick up fully.
We've been discussing some other possibilities, including writing the View component in an entirely different language. Java is being used on the back-end, because it gives us lots of integration capabilities, some of which are not yet found in the Ruby or pop's of the world. It also gives us a wealth of back-up programmers in case our current employees decide to mutiny.
However, the complexity of the stack on the back-end has cause for some difficulties going forward, and certainly for the summer. Development with java, of a web-based front-end, can take hours and hours, assuming you picked the right library. And which one to choose? GWT is getting a lot of press. But there's also dojo, or AjaxAnywhere. Or we could roll our own using Prototype.js or any of the other libraries.
Our current train of thought is to use something much simpler, something like PHP or Ruby on the front-end, to be able to bind to the XML services (this would force our system to have web services). Now, the main question is, which one?
Sunday, June 04, 2006
New Intern
On the 1st we had a new intern start with us, and he'll be working with another employee on integrating Open Laszlo with our upcoming software product. So far, he's been working out very well.A Few Weeks of Efficient Frontier
We've been using Efficient Frontier for a few weeks now, three to be precise. So, I thought I'd write about some of our experiences. The initial change over occurred on May 25th, and they started bidding for us. Our first few days we saw a significantly lower-than-average results from them. However, after calling them about the much lower conversion ratios, we determined that the daily spend was much lower than expected. By almost a grand. After updating this, it appears that our conversion ratios are back to where they really should be.Overall, I'd say I'm about 50/50 on their advertising platform at this point. Perhaps the upcoming weeks will show the power of the platform.
Friday, June 02, 2006
Double Nickel Steakhouse
Last night, Steph mandated that I take her out to a nice steakhouse. We decided to try Lisa Wests Double Nickel Steakhouse (which, as far as I can tell, does not have any kind of website). I was expecting a good dinner, having heard some rave reviews of the place, similar to what I might get at a Ruth Chris' Steakhouse. Following is our experience.The place is located across from the South Plains Mall. I'm not sure if that is bad, or good, but it's the old Spanky's location (which Lisa owns, but this location didn't work out so well). When we walked in, the ambiance was quite "up scale", with dimmed lighting and tiled flooring in the entry way. Our hostess (and everyone else) greeted us as Mr. / Mrs. and plenty of "sirs" and "ma'ams" were heard. The waitress didn't sit down at our table (as they would at Outback or Logan Roadhouse).
Our first impression upon sitting down was "This could have been done better". A place with a $45 Porterhouse should have white table-cloth linens. Not patterned plastic tablecloths. Ours was especially nice, with several cuts / slices in it whereby it looked like someone set a hot pot on it and burned it.
Our waitress showed up and took our orders for adult beverages, and filled our glasses with water. I didn't notice at first, but the wrapped white cloth that she dropped off was bread. It was harder than I expected (the bread), and not as tasty as many of the other steak houses in town.
We ordered a shrimp cocktail ($9 or so) which came with 5 medium-sized prawns. Additional shrimp were $2 / each (our waitress didn't think we should order more). When that arrived at our table we had a few bites -- I wasn't expecting much, but it did taste like United Supermarket frozen shrimp, in my opinion. No flavor what-so-ever.
For dinner we each ordered the Filet Mignon, with the "Double Nickel Salad" and a side of Asparagus. I had a 12oz, and my wife had an 8oz. Our salads were great -- an infusion of spring mix salad with caramel apple tastes and blue cheese. The most surprising aspect of it was when you bit into an apple, a candied nut, and a blue cheese chunk at the same time. The taste is not ... traditional, but for those who are a bit varied, it was wonderful.
A few more minutes went by and our dinners were served, during which time we decided that the plates were very ugly. Obviously, that's subjective, but I thought that the plates and bowls would look more at home at a Chinese buffet table. The steaks looked and smelled terrific, but after a few bites into mine, I realized that the back half of the steak was cold. Cold like a refrigerated soda can. I called our waitress over, and she whisked it away on our butter plate (from the bread) to reheat it.
While I was waiting on a re-heat of what must have been due to one of:
- A pre-cooked and then cooled steak
- An errant cook didn't turn the meat, and that side was never directly heated
- They don't use hot plates to keep the food hot while waiting for the side items
I did like two of four sides. The creamed corn was very good, and appeared to have some sort of molasses in it, making it sweet. The whipped potatoes were very good as well. The tomato slice was .... dubious. It wasn't great, nor horrible. I couldn't decide if it was pan seared, or baked for a few minutes, because it wasn't crisp, and it wasn't mushy. Something in between. The last side item was a deep-fried (I assume) cheese stick, which I didn't much care for.
Our tab ended up being around $100, plus another $20 for tip. I did notice that the combination of carpet, lighting, shoes, and college-aged wait staff meant that there were a lot of people tripping over themselves, and almost falling on their faces while serving people. This was more than humorous to me.
Overall, I'd say that it's probably one of the better places to have a steak in Lubbock. There's a plaque on the wall stating that it's the 9th best steak house in Texas, according to someone who ranks steak houses. I wouldn't go that far, but would say it has some of the tastiest food in Lubbock -- and is quite above the average franchise or chain that is so common in this town.