Thursday, September 27, 2007

Weekly Blog #4

This week in class we learned many things. One of them was virtual private networks. I virtual private network is a network that uses encryption to transfer data over a public network like the internet. In order to have a virtual private network you do not have to have a physical network, wireless or cable lines etc, and the costs associated to it. You could just rent network use or use a public one. Another point I learned this week is that databases have several configurations: hierarchical, relational and object oriented. I already new about databases, I used a few at work, mostly Microsoft Access, but I never knew how they worked.

Last week, Dell announced that it is rolling out a new line of servers preconfigured with Microsoft’s SQL Server 2005 database software. Dell claims that “the combination can top trendy data warehouse appliances on price, performance and ease of installation.” They come in three configurations with 1TB, 2TB or 4TB of storage and they are ready for to update to the SQL Server 2008 when it comes available next year. It is amazing to see how Dell can already make servers that are upgradeable to a new software almost a year before the new software is sold. I imaging that Microsoft is providing some kind of information about its new software configuration to allow Dell the make the upgradeable servers. For more information click on the article link: Microsoft and Dell Team Up

I am also including a link to a video related to this area: http://www.youtube.com/watch?v=OcCTJsx3AxI

Saturday, September 22, 2007

Weekly blog #3

This week in class I learned many things. One that o found very interesting was that there are several types of networks, like Local-area networks and Wide-area networks. I also learned that they have various configurations and architectures. I found interesting that networks are not just made of cables but can also be made of microwaves or satellites and they are also made of different pieces of hardware including routers, bridges or gateways. This week I also learned about this amazing piece of software: CyberLink magic Sport 4. I found amazing how software, can make a summary of a football game’s most important moments. I think it is incredible how the software can distinguish of what parts are important or not without any human interaction.

This week mortgage data, including over 5000 social security numbers were distributed trough an online peer-to-peer network. Once that data is downloaded it can be stored on millions of different computers and impossible to delete or track. It is sad how something so great that allows instant communication can be used to hurt people. Of course it is one of the disadvantages or dangers of networks. To read the complete article click the link: http://news.yahoo.com/s/ap/20070921/ap_on_hi_te/file_sharing_leak_2

Friday, September 14, 2007

Weekly blog #2

This week in class I learned about virtualization. I had no idea, it was easy and convenient to create virtual servers that can be moved and updated easily. It is very easy to have several virtual servers for different applications or purposes but only have one physical server or if you rent server space you don't even need to own one. Another point I learned this week was that there are hundreds of computer languages on several levels.

One article that got my attention this week was "Microsoft Preps for Daylight-Saving Time Headaches" on PC Magazine. It seems hard to believe that such a small thing like a change on the daylight savings time can cause so much trouble. Even after issuing a patch, outlook and calendars experienced problems like appointments being adjusted to the wrong time. What shocked me the most is the fact that even making such a small change to an application, like a time savings change, can be extremely complicated. On top of that, even after the changes on the software were made and the patch was applied, it still caused so much trouble.
For a video illustrating some of the problems click on the link: Daylight Saving Change Issues

Individual blog assignment #1

Sampling Software Screen

About a year ago I was involved with an IT company to try to develop a database and an automated sampling system for our loan reviews. Most financial institutions have custom made information systems. Most of our clients were no exception. Even though we required our clients to give us a complete data sheet in Excel, they all were slightly different. In addition, due the nature of the contracts, we require different levels of data. For example, Abc Bank has loans both in the US and in China, and thus they require that we review 30 % of Chinese loans and 60% of US loans. On the other hand, Xyz Bank only has loans on the US, but they require that we review 100% of all new loans over $250,000 and an additional 25% of the remaining portfolio. We would have to develop a separate Access database and sampling query for each bank. From the beginning we had multiple problems. The first problem was that the IT contractor would be located on South Carolina and with the exceptions of a few times were I had to travel there or they came to our offices all communication was over the phone or email. A second problem was that we had a dateline for each review and we had to develop each database before the date of the review and get a sample of the loans that need to be looked at. We tried to solve this issue by requesting the data from the banks 3 weeks in advance. Most banks would cooperate, but some would not. A third and the biggest problem was that the IT contractor could almost never make the sampling software work. Every time I ran it, it had a different problem. I would have to send them back to them to be fixed and it would come back a separate issue it didn’t have before. I think that this issue occurred because they always had different people working with us. In addition, I requested several times to add features to the software. The most important one was the ability to detect affiliated loans since we always have to review all of them together. They mentioned it couldn’t be done. I had to do that manually at every review. The other feature I wanted implemented was that the data updates weren’t so labor intensive. Some data had to be imported manually. It always took twice as much work for me to select a sample using the new software that doing it manually. After 18 months with no progress, the president of the company decided to call it off. As of today we still don’t have an automated sampling system (that works).

Saturday, September 8, 2007

Weekly blog #1

With the code name Barcelona, Advance Micro Devices will launch its new quad-code processor on September 10th to compete with Intel's quad-core processors, including its server version Xeon. AMD will sell its server version under the name Opteron.
The clock speed, 2.3GHz by December, will be slower that Intel's which already reach 3GHz.
This will most probably start another price competition.

As per AMD, with its new 10h architecture, Barcelona will deliver a better price-to-performance ratio than Intel. the new features include:
"A move to 128-bit floating-point execution units in 10h from 64-bit FPUs. The wider design is expected to double the performance of floating-point vector operations;
A wider fetch window -- 32 bytes from 16 bytes. This is expected to allow the processor to handle a complete sequence of three large instructions per cycle;
Instruction-set improvements that include the addition of two advance bit-manipulation instructions, which operate on general-purpose registers;
And virtual machine optimizations in 10h to boost the performance of AMD's virtualization technology, as well as compiler-related optimizations."

when processor improvements approached the physical barriers to improve clock speed and reduce size, new strategies had to be implemented. Intel and AMD started to increase the number of core processor inside each unit instead or increasing the clock speed how they have been doing over the last 20 years. They maintained the clock speed and they improved the bits per cycle, the cache memory and the number of cores. Eventually, this will not be enough and another improvements or a new technology will have to be developed.

For a view of the complete story click on the link: