In March 2019, the first IntelliFlex complex arrived to Belarus. And this is really impressive. IntelliFlex is currently the most powerful Teradata platform. More details of IntelliFlex benefits can be found on the Teradata website, where the platform has a whole separate section. Hybrid storage on SSD, InfiniBand, new BYNET, flexible scalability – many factors influenced the decision to switch to IntelliFlex. We took into account the current workload of the complex and the planned growth of our needs.
So far there are only commissioning and internal tests. Already made the first launches and test connections. In the internal space there is still no harmony, but already now noticeable how quickly take place the changes in the IT.
How fast the development goes is clearly seen when comparing the internal space of two top-end platforms released at different times. In our case, we are talking about the Teradata 6700 and Teradata IntelliFlex. Of course, much depends on the specific configurations, but still. Let this purely visual comparison, let the difference between platforms not less than 5 years, however, the new system with the declared performance, more than ten times, and the storage capacity, increased 4.5 times, fits into a similar-sized cabinet with a noticeable reserve for the occupied space.See how tightly packed the cabinet of Teradata Active Enterprise Data Warehouse 6700.
It is symbolic that Teradata seems to have switched to the bright side of life. This impression arises when you look at the two complexes standing nearby. During the installation, the cover had to be temporarily removed. Temporarily, IntelliFlex has returned to the dark side, but I hope everything will be in place when the new complex takes on the burden of Production environment. It’s funny that when you look at the photos of the cabinets, there is no sense of their actual size.
So, the first Teradata IntelliFlex complex in Belarus!
This is a significant, grand event. And we are at the center of these events. The fact has happened, and the page of our client’s history began to turn over. We are waiting for the start of this new page with an expectancy of something good thing. Everyone instinctively understands what should be faster, but how? How much? In one of the articles from Teradata there was such one slide ( на original source ◊).
As soon as commissioning works are completed, the most interesting will begin. We will be able to realistically assess how the duration of the processes will change in the reason of the replacement of only one complex within one automated system. We will try to prepare a whole cycle of articles with statistics on real changes in individual processes in connection with the transition to IntelliFlex.
How will the comparison be made?
After all the preparings will be completed, the migration to the new complex will begin. We will be able to get preliminary results of the comparison, but such a test will not be entirely fair. One of the complexes, among other things, will serve real business users. Performance comparisons after migration will be much more revealing. We will compare the duration of the regulatory processes after the full transition will be compared. What types of processes will we consider separately?
How much the duration of Stage jobs for all ETL processes will reduce, if the only thing that will be changed is the Teradata platform? Here, the data sources and the amount of downloaded data, network speed, ETL server and tool should remain unchanged. The peculiarity of the process: practically only bulk data loading operations.
Buffer and Work
How much will change the duration of the processes of formation of current data snapshots? It remains the same ETL server and tool, volume of data. Feature of the process: data conversion in non-temporal tables, mainly DML queries (insert, update, delete)
CoDM (Customized Teradata FSDM)
How much the duration of the execution of temporal data storage processes will change. It remains the same ETL server and tool, data volumes. Feature of the process: data conversion of temporal tables, mainly DML queries (insert, update, delete).
Compare the speed of the data extraction. The peculiarity of the process: mainly data sampling requests and analytic requests for read.
In general, it remains to be patient (perhaps with popcorn) and wait for the results. Perhaps will be posible to lay out and intermediate results. The transition to the next article in the cycle will be added here, after writing it, of course. Approximately not earlier than April.
P.S. I post photos with a landmark, it’s easier to estimate the size. My height is 1 meter 82 centimeters.