My take on this issue. I am a mainframe / DBMS advocate so of course I think there is some sort of general flaw in a move of critical applications to smaller less understood systems.
Think there are 3 parts to this problem 1. The triger failure its self which was hardware. 2. Work flow and organization. It appears that the data for the small system was harvested from main frame systems. I would expect that the new system contatained some sort of sync function that ran from time to time to update data in both the legacey system and the new one. The system did not contain a working DR process for total data loss in the new system, meaning data recovery probally had to reharvest the data from main frame sources. No problem for the first implementation as you just start the new system a week after harvesting begins. Big problem for DR as you are expected to be up during the harvest period. 3. DR configuration ... Big DBMS systems are unique in that they are designed to be recovered to current rather than too copy. This is done by restoreing the most recent backup and applying logs. I assume that new DBMS systems even on microsoft platforms have the capability to recover to current. What happens when one puts the large database, the backup copies and the logs on the same massive storage device and the device is lost? Recover to current does not work well. I have always held that data base people have a unique understanding of pratical data recovery to current. It is not all technology. As an example ever since VSAM recoverable catalogs (and we are several generations beyond that) catalogs coud be recovered to current by restoring backups and applying records from the log (in this case SMF). The process is rarly used and historically presents problems in DRs. As an example some logs needed for a DB recover to current tend to be weirdly cataloged or not cataloged in VSAM depending on the timing of the catalog back ups. This problem is slightly changed in 'beeding edge" mirror systems. The non DBMS world skipped the copy + log recovery era ... they simply don't understand it at the pratical level. "They" often incudes the non DBMS systems, application design, and DR people. Regards Av Friedman PS This comment happens to be posted as a reply to a note correctly poining out out sighting is not the same citing. Why does this outage prevent the offcers from writting tickets for expired licenses when the driver is stopped for other reasons. After all the experation date is on the drivers license? The answer is the govermnent may be partly responsible for the expired license as the removed the availabily of options to renew. If is can "sight" a diffrent case. One of the reasons why airlines usually do not issue paper tickets any more is there is a serious problem when a passenger presents at the ticket counter or gate with a valid ticket but has been canceled by phone or electronicly. It is not easy to deny such a passenger boarding (they use to get bumped to first class all the time) as thy hold evidence of a vaild contract between them and the airline. In the drivers license case the expire date is not all that important if there is no renewal oppertunity. In the airline case what is in the computer is not all that important if the flyer holds a valid contract. When ever a system is built with two or more seperatly managed 'copies' of the same data problems result. Some DBMS systems like IMS were designed to reduce data redundancy. Some DBMS systems like DB2 cause data redundancy. On Thu, 2 Sep 2010 00:00:46 -0400, Robert A. Rosenberg <[email protected]> wrote: >At 20:25 -0400 on 09/01/2010, J R wrote about Re: Virginia DOT outage: > >> > ... police are not sighting people who are stopped and found to >>have recently expired licenses. >> >>If the police are not sighting them, how are they stopping them -- >>and how do they know their licences are recently expired? > >While they are SIGHTING them, the problem is that they are not CITING >them. When they are stopped and hand over an expired license there is >no need to check the computer - the document itself has an expired >date on it. All checking with the computer would do is verify if the >license had been renewed (and thus was in force). Also the computer >check would be needed to spot a license whose expiration date had not >yet been reached but which had been suspended. > >---------------------------------------------------------------------- >For IBM-MAIN subscribe / signoff / archive access instructions, >send email to [email protected] with the message: GET IBM-MAIN INFO >Search the archives at http://bama.ua.edu/archives/ibm-main.html ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
